Computer-aided robotics for applications in fracture reduction surgery: Advances, challenges, and opportunities

Summary

The advancement of information technology and AI has boosted global economic and social development. Robot systems (RS) and computer-aided technology (CAT) are used in various domains, including social production and human existence. Traditional fracture reduction surgery relies on the expertise and surgical skills of surgeons to realign fractures in patients. Researchers have developed robotic and assisted systems to automate fracture reduction surgery in recent decades. Computer-aided fracture reduction robot system (CARS) is used to replace the manual reduction performed by conventional physicians. A partial CARS has been used successfully in clinical fracture reduction surgery. This study provides an overview of CARS. First, the RS and CAT used in fracture reduction surgery are overviewed. Furthermore, a comprehensive analysis of CARS is presented, encompassing their design, experimental validation, and clinical applications, while highlighting recent advancements and potential future directions in this domain. The suggested CARS for fracture reduction are compared in different ways. The learning curve and technical ethics of CARS are summarized. The paper addresses unresolved research gaps and technical challenges, providing recommendations to guide future study.

Graphical abstract

Applied sciences; Health sciences; Natural sciences

Introduction

The prevalence of orthopedic disorders has risen due to economic growth and increased transportation. Fracture is common in clinical orthopedics and endangers the health of susceptible individuals. Osteoporosis contributes to the rising occurrence of senile fractures as individuals age. Fractures are less common in younger individuals compared to the elderly, but the outlook is still unfavorable.¹ The fracture rate among individuals under 50 is elevated due to high-intensity labor, road accidents, and sports injuries.² While most fractures in these groups are not deadly, they can result in different levels of disability, such as restricted movement, limb deformities, and reduced motor function post surgery. These effects can significantly impact the growth and development of adolescents and children, as well as the work capacity of adults. Fracture rates globally differ according to age, gender, geographic region, food, lifestyle, and various other factors. The World Health Organization reports around 180 million fractures globally annually, with the hand, wrist, shoulder, femur, and spine being the most commonly affected areas.³ Figure 1 shows the distribution of human fractures. The musculoskeletal system is the most crucial and intricate motor system in the human body.⁴ It has a complex structure and is surrounded by significant vascular and nerve tissues, leading to a high risk during surgical procedures.

Figure 1.

CARS for applications in fracture reduction surgery

Researchers have developed various robot systems for fracture reduction surgery, such as spine surgery robots,^5,6 femur fracture reduction robots,^7,8 long bone fracture reduction robots,^9,10 pelvic fracture reduction robots, and others designed for different body parts.^11,12,13 This is due to advancements in computer, robot, and AI technology. Currently, the use of CARS is a growing area of scientific interest. CARS have been utilized in many orthopedic clinical procedures such as spine surgery,¹⁴ joint replacement,¹⁵ bone tumor removal,¹⁶ and traumatic fracture fixing.^17,18 CARS surpasses traditional surgery. It provides automatic preoperative planning, real-time intraoperative navigation, precise collaborative control of robotic arm and operating table, and postoperative rehabilitation detection feedback.¹⁹ It meets the demands for accurate, minimally invasive, and low radiation fracture reduction surgery.²⁰ Figure 1 shows the application of orthopedic surgical robots. While certain strategies have been used in clinical settings, research on CARS is still in its early stages.

Several literature reviews have examined the use of CARS in orthopedic surgery.^21,22,23 These studies have covered various technologies, focusing on summarizing the principles and historical development of robot technology and navigation technology in fracture reduction surgery.^24,25 This paper thoroughly examines the use of CARS, discussing the robot system, structure, navigation technology, virtual reality technology, and learning curve in detail. This analysis is crucial for advancing research and innovation in this field. Figure 2 depicts the primary technologies utilized in current CARS and their interconnections. Due to continuous advancements in computing, robotics, AI, and related technologies, CARS is poised to progressively replace traditional surgical procedures performed by surgeons, ensuring precise and autonomous completion of the surgical process;²⁶ CARS and computing technologies to perform preoperative preparation, intraoperative navigation, and fracture reduction.²⁷ Depending on the specific location of the patient’s bone fracture, CARS can perform customized surgeries.²⁸

Figure 2.

The main technologies used in CARS and their interrelation

The following sections of this review are organized as follows. The text provides a comprehensive summary of the current status of CARS. It categorizes and describes the system, robot structure, and specialized robot technology used for various fracture sites. The text introduces the evolution of several CARS, focusing on preoperative planning, navigation technology, and virtual reality (VR) technology. Section 4 analyzes the learning curve of CARS to determine the learning and usage patterns of various surgical systems by surgeons. It also compares and analyzes the learning curve of a similar type of robot system. The fifth segment examines the ethical concerns arising from the use of CARS. Unresolved scientific concerns and technical challenges are addressed, along with recommendations to support future study.

Basic composition of CARS

To provide a comprehensive review of the latest advances in CARS, a rigorous literature search and screening process was conducted. To ensure the breadth and accuracy of the literature search, carefully selected keywords and phrases were used to cover the core concepts in the field. Key search terms included, but were not limited to: “computer-assisted fracture reduction”, “fracture reduction robot”, “intelligent fracture reduction system”, “robot-assisted orthopedic surgery”, “automated fracture treatment”, and related technical terms such as “image guidance”, “navigation system”, “3D reconstruction”, “virtual reality”, and “learning curve”. Additionally, time-related qualifiers were added to capture and clarify the development history and significant research results of CARS, considering the rapid technological advancements.

Several well-known scientific databases and search engines were searched, including PubMed, Web of Science, Scopus, IEEE Xplore, SpringerLink, and Google Scholar, to cover the widest range of academic resources. These platforms, covering multiple disciplines such as medicine, engineering, and computer science, helped to collect relevant literature across disciplines. The inclusion criteria focused on selecting high-quality literature directly related to the application of computer-assisted and robotic techniques in fracture reduction, with in-depth discussion of the technology’s working principles, system design aspects, applications in clinical practice, evaluation of effects, and prediction of future trends. Only original research papers, comprehensive reviews, systematic reviews, or articles from major academic conferences were considered to ensure that the information reviewed was modern and cutting-edge in technological development. The exclusion criteria were designed to screen out literature that was not closely related to the topic or had limited academic contributions, including content that did not directly discuss the use of computer-assisted and robotic techniques in fracture reduction; literature in the form of case reports, pure reviews, letters, and abstracts that did not contain key information unless they made a significant contribution to the field; research methods that were not rigorous enough, such as vague study design, insufficient sample size, or insufficient statistical methods; and possible comprehension errors due to language barriers. Only English publications were included to ensure the accuracy of the review and the consistency of international perspectives. An initial search was conducted based on preset search terms and databases, followed by further selection based on title and abstract. A total of 346 articles were selected.

CARS is gaining attention as a research focus in medicine and engineering. It is tailored to various clinical application scenarios. Traditional fracture reduction relies on the surgeon’s medical expertise and clinical experience. Computer-assisted robotic fracture reduction surgery, on the other hand, utilizes preoperative planning,²⁹ intraoperative registration,³⁰ navigation technology,³¹ and robotic technology³² to perform fracture reduction surgery precisely. This study compares various systems, structures, and application scenarios of CARS. It summarizes their structural characteristics and functional aspects comprehensively. CARS refers to the combined use of computer-assisted technology and robot technology in various stages of fracture reduction surgery.

Figure 3 displays the complete workflow of fracture reduction surgery using CARS. Preoperative planning for computer-assisted fracture reduction surgery mostly involves reconstructing the fracture model and performing virtual reduction. Intraoperative registration approaches primarily consist of picture-free registration,³³ two-dimensional (2D)/three-dimensional (3D) image registration,³⁰ 3D/3D image registration,³⁴ and image fusion registration.³⁵ Navigation technology primarily consists of optical navigation,³⁶ electromagnetic navigation,³⁷ and opto-magnetic cooperative navigation.³⁸ RS is primarily categorized based on its structural and connection forms. It performs fracture reduction based on various application circumstances. The study examines the learning curve of CARS and tackles the issue of responsibility attribution.

Figure 3.

Flow chart of CARS

Robotic fracture reduction system

The CARS can be categorized into three types based on their action mechanism and interaction level during fracture reduction surgery: fully automatic RS, active-assisted RS, and navigation-assisted RS.

Fully automatic RS

During the operation of the fully automated RS, the surgical robot operates autonomously without external assistance; the surgeon intervenes or halts the surgical process only in emergencies. Automated RS typically consist of high-precision robotic arms, 3D imaging devices (e.g., CT or MRI), advanced image processing software, and control algorithms. The robotic arm acts directly on the bone, performing operations according to a preset plan.

As shown in Figure 4, the world’s first fully automated RS, ROBODOC, was developed by Curexo Technologies, Inc., in Follimont, CA, and is currently in clinical use.³⁹ It was originally designed to perform total hip replacements and was first used in clinical orthopedic surgery in 1992. As the first RS for clinical orthopedic surgery, ROBODOC was approved for sale by the European Union in 1994 and by the FDA in 2008. The ROBODOC system consists of an ORTHODOC workstation and a 5-axis robotic mechanism. The system uses preoperative CT scans to guide the surgical process, creating a detailed 3D model of the bone through an ORTHODOC workstation and subsequently developing a customized surgical strategy. Finally, the operation is completed by a five-axis robotic structure. ROBODOC ensures surgical accuracy by using a reference point of about 5 mm implanted in bone tissue. The reference points help the system precisely monitor bone movement and track it, ensuring that surgery is not interfered with by debris or fluids. As a fully autonomous system, after ROBODOC performs a surgical operation, the surgeon can only abort the operation through the emergency stop function, highlighting the system’s autonomous operation capability. In 2014, TSolution-One replaced ROBODOC and was approved by the FDA in 2019.⁴⁰ TSolution-One is the only fully automated RS currently available for clinical surgery. In addition to ROBODOC and TSolution-One, CASPAR was introduced in 1997 as an early fully automated 6-DOF RS for orthopedic surgery. CASPAR, an interactive computer system, was used to perform preoperative planning for CT image guidance and tasks such as bone milling. Compared with ROBODOC, CASPAR had a longer operation time, increased blood loss, and longer complications and postoperative recovery time.⁴¹ Due to these shortcomings, the CASPAR fully automated RS is now discontinued.

Figure 4.

Orthopedic surgical robot with fully automatic RS

(A) ROBODOC. Reproduced with permission from ref.³⁹. Copyright 2013 The Korean Orthopedic Association.

(B) TSolution-One. Reproduced with permission from ref.⁴⁰. Copyright 2017 The Authors. Published by EDP Sciences. Licensed under the CC BY 4.0 license.

(C) CASPAR. Reproduced with permission from ref.⁴¹ Copyright 2021 The Authors. Published by Elsevier B.V. Licensed under the CC BY-NC-ND license.

Since the inception of fully automated RS to the current generation of multiple products, these systems can automate the entire process from diagnosis to treatment, significantly improving surgical accuracy and reducing human error. Their primary feature is completely autonomous operation, which reduces dependence on the manual skills of doctors and is suitable for the treatment of complex fractures. However, it also requires highly complex control systems and advanced security mechanisms. Currently, research on automated RS is focusing on improving the intelligence of the systems, including more efficient fracture recognition algorithms, adaptive reduction strategies, and enhanced human-computer interfaces. Safety and ethical considerations are also a focus of current research, and ensuring efficient treatment while maintaining patient safety is a key direction for future development.

Active-assisted RS

The active-assist RS consists of a robotic arm, sensors, a computer control unit, and an operating interface for the surgeon. It does not operate completely autonomously but acts as an “assistant” to the surgeon, providing power support and precise guidance.

As shown in Figure 5, the mainstream active-assisted CARS include the MAKO system,⁴² the NAVIO system,⁴³ and the CORI system.⁴³ In the MAKO system, surgeons create surgical plans based on preoperative images. Once registration is complete, the system displays the surgical plan on the screen using data from multiple sensors. With the system’s assistance, the surgeon manipulates the robotic arm to perform tasks such as bone removal or screw placement. If the surgeon causes the robotic arm to deviate significantly from the intended trajectory, the system stops operating. In a semi-automatic system, the surgeon’s actions play a decisive role in the system’s operation. Similar to the MAKO system, the NAVIO and CORI systems use a model without image processing to achieve accurate registration of anatomical sites during surgery. Active assistive robots enhance the surgeon’s ability to operate by reducing hand tremor, providing precise force control, and delivering real-time surgical feedback, leading to a reduced risk during surgical procedures. These systems retain the surgeon’s subjective judgment and control and are applicable to various fracture types.

Figure 5.

Active-assisted orthopedic surgical robot

(A) MAKO.

(B) NAVIO. Reproduced with permission from ref.⁴³ Copyright 2023 The Authors. Published by MDPI. Licensed under the CC BY 4.0 license.

Currently, research on active assistive RS mainly focuses on improving flexibility, response speed, and cooperation with surgeons. By integrating machine learning algorithms, these systems can better understand and predict the surgeon’s operational intent, providing more personalized assistance. Compared with fully automated systems, semi-automated systems can better ensure the safety of the surgical process. A disadvantage is that the surgeon, affected by the system’s automatic control, cannot adjust the robotic arm in real time.

Navigation-assisted RS

The navigation-assisted RS mainly includes imaging equipment, image processing software, a navigation system, and a mechanical guidance device. In contrast to fully automated and actively assisted RS, it does not directly participate in the physical reduction of the fracture but provides the surgeon with precise surgical path and positioning information.

As shown in Figure 6, typical navigation-assisted RS currently include the ROSA system,⁴⁴ the OMNI-Botics system,⁴⁷ and others. The function of the navigation-assisted RS is to use software and sensing technology to construct a 3D model of the fracture site, providing a better reference for surgeons to complete the operation. The RS can be combined with CT images to complete the calibration and registration of specific points in the surgical environment by the robot. The OMNI-Botics system plans the path for the size and position of the surgical implant and guides the mechanical device’s path using navigation technology.⁴⁸ In the navigation-assisted RS, the surgeon creates a surgical plan based on the patient’s fracture and inputs it into the system. The surgeon moves the device to the designated position according to the displacement and angle calculated by the system, and then performs the surgical operation. The navigation-assisted RS is fully controlled and operated by the surgeon, and its surgical accuracy is significantly improved compared to traditional surgery. Table 1 presents the orthopedic surgical navigation systems (RS) that have been validated at this stage.

Figure 6.

Navigation-aided orthopedic surgical robot

(A) ROSA. Reproduced with permission from ref.⁴⁴. Copyright 2021 The Authors. Licensed under the CC BY 4.0 license.

(B) TiRobot. Reproduced with permission from ref.⁴⁵ Copyright 2023 The Authors. Published by Elsevier. Licensed under the CC BY-NC-ND license.

(C) OMNI-Botics.

(D) Mazor X Stealt. Reproduced with permission from ref.⁴⁶ Copyright 2020 Elsevier.

Table 1.

Orthopedic surgical robot systems that have been validated at this stage

System Type	System Name	Application Category	R&d institutions and countries
Fully automatic robot system	TSolution-One49	Active robot-controlled milling/cutting	THINK Surgical, California, USA
	Acrobot50	Trial phase	Acrome Robotics, California, USA
	CASPAR41	Active robot-controlled milling/cutting	NDI, Ontario, Canada
Active-assisted robot system	MAKO51	Haptic feedback, tandem surgeon–robot controlled milling	Stryker, Michigan, USA
	NAVIO52	Haptic feedback, tandem surgeon–robot controlled milling	Smith & Nephew, London, UK
	VELYS53	Haptic feedback, tandem surgeon–robot controlled cutting	DePuy Synthes, Johnson and Johnson Group, Indiana, USA
	Versius54	Robotic cutting/manipulation with direct surgeon control	CMR Surgical, Cambridge, UK
	Da Vinci54	Robotic cutting/manipulation with direct surgeon control	Intuitive Surgical, California, USA
	CORI43	Haptic feedback, tandem surgeon–robot controlled milling	Smith & Nephew, London, UK
	ZEUS55	Robotic cutting/manipulation with direct surgeon control	Computer Motion Inc., California, USA
Navigation-assisted robot system	TiRobot56	Drilling or wire trajectory guidance	TINAVI Medical, Beijing, China
	ROSA57	Cutting jig positioning	Zimmer-Biomet, Indiana, USA
	Omnibiotics58	Cutting jig positioning	OmniGuide Surgical Corin Group, Cirencester, UK
	Spine Assist59	Drilling or wire trajectory guidance	Mazor Robotics, Medtronic Group, Minnesota USA
	ExcelsiusGPS49	Drilling or wire trajectory guidance	Globus Medical, Pensylvania, USA
	Mazor X Stealth46	Drilling or wire trajectory guidance	Mazor Robotics, Medtronic Group, Minnesota USA

Currently, research on navigation-assisted RS focuses on improving the accuracy and robustness of navigation systems by integrating multimodal image data, developing advanced tracking technologies, and optimizing user interface design to enhance surgical efficiency and safety. Additionally, the application of VR and augmented reality technologies is being explored to further enhance the intuitiveness and interactivity of surgical procedures.

The basic structure of RS

Robots in surgery have evolved from early industrial robots to specialized medical surgical robots. Surgical robots have demonstrated distinct benefits in the medical industry after years of ongoing research and advancement. The growing need for orthopedic surgical robots drives the varied advancement of their systems and the ongoing investigation and innovation of their structures. Orthopedic surgical robots can be categorized as external fixed structures, serial structures, parallel structures, and hybrid structures.

External fixation orthopedic surgery robots follow the traditional way of fixing broken bones. They accomplish fracture reduction by compressing one or both ends of the broken bone, altering the structure, and continuously adjusting the position. Figure 7 shows some applications of external fixation robots. Koo et al. and Seide et al.^60,61,62 pioneered unilateral fixation for fracture reduction, where one end of the fracture is immobilized and the other end is attached to a mechanical mechanism that allows some movement. The surgeon can realign the fractured bones by continuously altering the position of the surgical apparatus. Although the robots used to achieve unilateral fixed reduction are inexpensive and simple in structure, their serial structure results in low surgical accuracy and cannot meet practical clinical needs. Majidifakhr et al.⁶³ applied the Stewart platform in fracture reduction surgery by modifying the mechanical structure linking the two sides of the damaged bone. Tang et al.⁶⁴ incorporated CAT into the RS to enhance the precision of the reset accuracy. The external fixation robot at both ends of a broken bone has advantages such as small size, light weight, and easy operation in the operating room, and can automatically achieve high-precision bone restoration. However, due to its design, the robot is only suitable for long bone fractures and is not suitable for other complex fracture types. This type of robot is limited by motor power, resulting in poor load capacity. Therefore, the development of a simple, high-load, and easy-to-control robot has become the main focus for researchers.

Figure 7.

External fixed fracture surgical robot

(A) 6-DOF external fixator. Reproduced with permission from ref.⁶¹ Copyright 2004 Robotic Publications Ltd.

(B) Dynafix unilateral external fixator. Reproduced with permission from ref.⁶² Copyright 2005 Elsevier.

(C) 6-DOF displacement system. Reproduced with permission from ref.⁶⁵ Copyright 2014 John Wiley & Sons, Ltd.

(D) New 3D CARS. Reproduced with permission from ref.⁶⁴ Copyright 2011 John Wiley & Sons, Ltd.

The widespread utilization of industrial robots has enhanced global industrialization and spurred researchers to explore their application in other scientific domains. Füchtmeier et al.⁶⁶ developed a serial surgical robot for fracture reduction inspired by industrial robots. The surgeon adjusts the clamping device’s position at one end of the robotic arm during the fracture reduction process. This technology achieves precise fracture positioning through high-precision sensors and advanced control, provides stable force control to protect tissue, reduces the surgeon’s labor intensity, and improves surgical outcomes and adaptability with the aid of visualization and intelligent assistance. However, it also faces challenges such as high cost, technical complexity, system integration difficulty, emergency response limitations, low acceptance, and challenges in maintenance and upgrades. As shown in Figure 8C, Jamwal et al.⁶⁷ developed an intrinsically compliant parallel robot for femoral fracture reduction. The system utilizes six pneumatic muscle actuators to achieve a six-degree-of-freedom (6-DOF) motion and precise trajectory tracking. This design significantly enhances alignment accuracy and minimizes radiation exposure compared to traditional methods. Oszwald et al.⁶⁸ enhanced the industrial robot and validated the enhancements by testing it on mice and human bones which is shown in Figure 8D. Kim et al.⁶⁹ created a 6-DOF force-feedback control scheme designed to reduce excessive force applied to the fractured bone during robot-assisted fracture reduction and to provide the surgeon with tactile force feedback, as shown in Figure 8F. The system effectively addresses the limitations faced by traditional single-sensor systems in protecting vulnerable tissues, integrates human-robot collaborative control to improve surgical flexibility and safety, and reduces X-ray exposure through a 3D navigation system, meeting the needs of modern medical surgery to minimize radiation exposure. Although this scheme provides an advanced force-feedback mechanism, its complexity and cost may be higher than those of earlier single-sensor systems, and adequate clinical validation and user training are still required for widespread adoption and implementation. Wu et al.⁷⁰ integrated a 6-DOF series manipulator with a robot-assisted traction device to decrease the mechanical force needed by the robotic arm to the specified load. The system effectively reduces the dependence of surgery on the surgeon’s experience, minimizes radiation exposure during surgery, decreases the instability of manual operation through precise control, and balances the soft tissue binding force in the complex pelvic structure. However, the system faces challenges such as balancing flexibility with load capacity, a lack of clinical reset parameter guidance, and an inadequate robotic test simulation system, which limits its large-scale clinical application. As is shown in Figure 8G, Pan et al.⁷¹ proposed an augmented A∗ algorithm for robot-assisted pelvic fracture reduction. The approach integrates collision detection and path optimization, significantly improving surgical safety and operational efficiency in complex anatomical environments. Ge et al.⁷² created an innovative RS tailored for pelvic reduction. The system can identify fracture patterns and carry out surgical planning and execution of reduction techniques automatically. The pelvic RS effectively completed the clinical procedure in 110 min, achieving successful anatomical reduction. The system demonstrates a high degree of automation and precision in operation, and although it is still in the research stage it shows great application potential.

Figure 8.

Serial fracture reduction robot

(A) Serial RS. Reproduced with permission from ref.⁷² Copyright 2022 The Authors. Published by MDPI. Licensed under the CC BY license.

(B) Long bone fracture reduction RS. Reproduced with permission from ref.⁷³ Copyright 2006 John Wiley & Sons, Ltd.

(C) Pelvic reduction robot. Reproduced with permission from ref.⁶⁷ Copyright 2022 John Wiley & Sons, Ltd.

(D) Femoral shaft CARS. Reproduced with permission from ref.⁶⁸ Copyright 2008 Orthopedic Research Society.

(E) Robotic restoration of distal femur fragment. Reproduced with permission from ref.⁷⁴ Copyright 2010 Orthopedic Research Society.

(F) Six-degree of freedom positioning robot. Reproduced with permission from ref.⁶⁹ Copyright 2016 Elsevier.

(G) Minimally invasive intramedullary nail implantation robot. Reproduced with permission from ref.⁷¹ Copyright 2021 Elsevier.

(H) Minimally invasive intramedullary nail technology. Reproduced with permission from ref.⁷⁵ Copyright 2021 Elsevier.

Research on tandem robots is maturing, particularly in improving the precision of reduction, reducing radiation exposure for the surgical team, and decreasing the workload of the surgeon. Current research trends focus on system intelligence, automation, and seamless integration with existing surgical processes. Despite the challenges of cost, technical complexity, and acceptance, tandem RS are expected to play a more significant role in fracture treatment in the future as technology advances and costs decrease.

Researchers have conducted experimental studies on the medical use of parallel robots to address the poor carrying capacity issue of serial surgical robots. Ye et al.⁷⁶ introduced an innovative 6-DOF surgical robot with improved positioning precision and stiffness. As is shown in Figure 9A, Wang et al.⁷⁷ developed a parallel manipulator robot (PMR) for femoral fracture reduction, employing a traction table setup with 6-DOF movements. This system demonstrated precise alignment with errors as low as 0.63 mm during fine adjustments. Fu et al.⁷⁸ proposed an indirect visual servo-guided fracture reduction robot based on external markers. Utilizing binocular cameras, this approach minimized radiation exposure and ensured precise fracture alignment with angulation errors of 3.3° ± 1.8° and transverse errors of 2 ± 0.5 mm which is shown in Figure 9B. Abedinnasab et al.⁹ introduced a 6-DOF parallel mechanism for computer-assisted fracture reduction with increased reset precision compared to the Stewart platform which is shown in Figure 9C. The system not only reduces interference between the motion chains and expands the robot’s working space but also greatly facilitates the placement and surgical operation of the affected limb through its semi-planar structure, providing surgeons with a better visual field and closer wound proximity. Compared to the well-known Gough-Stewart platform, this mechanism has demonstrated advantages in precise control and positioning, especially for precise repositioning against strong muscle loads. Experimental tests have shown that the required repositioning steps can be applied with high precision. Although high-precision reduction has been achieved through simulated clinical surgery, safety in practical clinical applications, especially for soft tissue protection, still needs further validation. An et al.⁷⁹ optimized a fracture reduction robot (FRR) controller using an improved Sparrow Search Algorithm (SSA) with Levy flight, achieving a 98.74% reduction in motion trajectory error compared to traditional methods. This innovation significantly improves the precision and stability of FRR systems which is shown in Figure 9D. Du et al.⁸⁰ introduced a fluoroscopic-guided robot-assisted orthopedic surgery system designed to improve the accuracy of long bone fracture reduction and the success rate of intramedullary nail locking, while reducing surgical time and the surgeon’s radiation exposure to the C-arm X-ray machine. The system combines the parallel robot simulation of traditional manual reduction and series robot guidance of intramedullary nail fixation and integrates the remote operating system, surgeon console, and control software to support surgical planning and teaching functions. The advantage of this method is that it can improve reduction precision and intramedullary nail locking rate, reduce the surgeon’s workload, and has been verified by clinical sample tests. However, it only targets the reduction and intramedullary nail locking of long bone closed fractures, especially femur and tibia fractures. Its function is relatively specific and does not widely cover all types of fracture treatment.

Figure 9.

Parallel fracture reduction robot

(A) A parallel manipulator robot on a traction table. Reproduced with permission from ref.⁷⁷ Copyright 2013 Wiley & Sons, Ltd.

(B) Path planning of parallel RS. Reproduced with permission from ref.⁷⁸ Copyright 2020 Wiley & Sons, Ltd.

(C) Fully open three-legged parallel robot. Reproduced with permission from ref.⁹ Copyright 2017 ASME.

(D) 6-DOF parallel robot for femur fracture reduction. Reproduced with permission from ref.⁷⁹ Copyright 2023 Elsevier.

Traditional RS have disadvantages such as low precision and one-way reduction, making it challenging to maintain the accuracy of bone alignment in fracture reduction surgery. In contrast, the parallel structure robot offers greater freedom in the fracture reduction process and can achieve accurate fracture alignment and rotation adjustment. However, due to the limited operating space of this structural form, the parallel robot structure is only suitable for lower limb fracture reduction, which limits its practical clinical application.

The series structure robot has a wide range of motion and high flexibility, which is suitable for a wide range of work movement, while the parallel structure robot has light weight, strong stability and high reset precision compared with the series structure robot. In order to fully combine the advantages of the above structures, researchers proposed an RS based on the hybrid structure of series and parallel. Mixed structure surgical robot has become a popular research field because of its advantages in clinical surgery.

Tao et al.⁸¹ compared the accuracy of dental implant placement using a dynamic navigation system and a robotic system. Results showed the robotic system outperformed the navigation system with lower entry and exit deviations, as well as reduced angular error, as shown in Figure 10A. Zhao et al.³⁴ developed an intelligent robot-assisted minimally invasive reduction system for unstable pelvic fractures. The system achieved an average reduction error of 3.41 mm and a 95.5% “excellent and good” rate using the Matta criteria in clinical applications which is shown in Figure 10B. Kuang et al.⁸² developed the HybriDot, a passive/active hybrid robot for orthopedic trauma surgery which is shown in Figure 10C. This system integrates circular prismatic and back-drivable joints, achieving high positional accuracy and angular precision. Clinical trials demonstrated its efficacy in supporting tasks like distal locking of intramedullary nails. The system is still in the preliminary research stage, and the long-term efficacy and safety have not been fully verified. Further evaluation and optimization through bone models and clinical trials will be required in the future. As is shown in Figure 10D, Sandoval et al.⁸³ developed a collaborative framework for minimally invasive surgery using a 7-DOF robot. The system ensures compliance with surgical tasks under shared workspaces through null-space control strategies. Wang et al.⁸⁴ proposed a tracking control method for cable-pulley-driven surgical robots, combining non-linear hysteresis compensation and force estimation which is shown in Figure 10E. This method achieved 99% positional accuracy and introduced a self-sensing force capability for dynamic load adjustments. Nguyen Phu et al.⁸⁵ proposed an innovative hybrid mechanical structure designed to improve the accuracy and efficiency of computer-assisted bone reduction surgery which is shown in Figure 10F. The system combines the 3-RPS tripod parallel mechanism with the double triangular plane parallel mechanism. Kinematic analysis and simulation show that a complete fracture reduction trajectory can be generated. Compared to the traditional Stewart platform, it significantly improves the workspace, especially in the treatment of long bone fractures such as the femur. The system provides an advanced solution for bone reduction surgery, but further research is needed to optimize its clinical practicality and ease of use. Song et al.⁸⁶ designed the HyBAR, a hybrid bone-attached robot for joint arthroplasty, combining serial and parallel kinematics which is shown in Figure 10G. The system improves precision and structural rigidity while enabling high-speed bone shaping during minimally invasive surgery.

Figure 10.

Mixed structure RS

(A) Movable hybrid robot system. Reproduced with permission from ref.⁸¹ Copyright 2022 Elsevier.

(B) Intelligent robot-assisted fracture reduction system. Reproduced with permission from ref.³⁴ Copyright 2022 Elsevier.

(C) HybriDot system. Reproduced with permission from ref.⁸² Copyright 2012 Wiley & Sons, Ltd.

(D) Minimally invasive surgical RS. Reproduced with permission from ref.⁸³ Copyright 2018 Elsevier.

(E) Visual control-based surgical robot. Reproduced with permission from ref.⁸⁴ Copyright 2023 Elsevier.

(F) Novel hybrid structural surgical robot. Reproduced with permission from ref.⁸⁵ Copyright 2019 The Authors. Licensed under the CC BY 4.0 license.

(G) Hybrid structural robot. Reproduced with permission from ref.⁸⁶ Copyright 2009 John Wiley & Sons, Ltd.

Currently, by fully combining the structural advantages of multiple types of robots, the hybrid structure robot can achieve high-precision fracture fragment reduction, effectively avoiding the infection and radiation hazards associated with traditional open surgery, and greatly enhancing surgical safety. This structural form not only improves the efficiency of fracture reduction surgery but also adapts to the surgical needs of complex fracture environments, ensuring the safety and success rate of surgery while reducing dependence on the surgeon. Currently, the hybrid structural robot has become the mainstream structural form of fracture reduction surgical robots and has received increasing attention.

Kinematic modeling of CARS

Kinematic modeling is the foundation of CARS design and control. The kinematic modeling of CARS primarily includes forward kinematics and inverse kinematics.⁸⁷ As shown in Figure 11, forward kinematics is the mapping between the joint space of a CARS and the working space of the robot. In other words, the position and orientation of the end-effector are determined given the joint angles. Inverse kinematics is the study of the mapping from the robot workspace to the joint space. It involves determining the joint angles of the CARS from the end-effector’s pose, which is called the inverse kinematics solution. In this section, the kinematic models of several typical robot structures will be discussed in depth to provide effective theoretical support for the design of CARS controller.⁸⁸

Figure 11.

The role of manipulator kinematics in CARS

The kinematic modeling methods of CARS mainly include geometric modeling method and D-H modeling method.⁸⁹ The geometric modeling method is suitable for robots with relatively simple structures, especially planar robots. For example, due to the relatively fixed geometric configuration of the external fixed structure robot, the positional relationship between the joint variables and the end orientation of the CARS can be directly deduced using linear and planar geometric relations. The D-H modeling rule is more versatile and more suitable for robots with complex structures, including serial, parallel, and hybrid structures. The D-H method describes the pose transformation of the robot by defining the homogeneous transformation matrix between joints, which is convenient for dealing with systems of multiple degrees of freedom. For the series-connected structure, each joint corresponds to a D-H parameter set, and the kinematic model of the entire system is obtained by multiplying the D-H matrices. The parallel structure needs to consider the simultaneous transformation of multiple branch chains and solve the inverse kinematics problem using an optimization method.⁹⁰ The hybrid structure robot has characteristics of both series and parallel structures, and its modeling needs to combine both methods.⁹⁰

Kinematic model of a serial CARS

The Figure 12 shows the schematic diagram of the 1-DOF robotic arm. For the 1-DOF robotic arm, its kinematic model is established by geometric method as follows:

$$x=lcos\theta$$ (Equation 1)

$$y=lsin\theta$$ (Equation 2)

Where l stands for joint length; θ is the joint variable.

Figure 12.

Kinematic model of CARS

The Figure 12 shows a schematic diagram of a 2-DOF robotic arm. For a 2-DOF robotic arm, its geometric motion model is expressed as:

$$x=l_1\mathit{cos}{\theta }_1+l_2\cos \left({\theta }_1+{\theta }_2\right)+···+l_i\cos \left({\theta }_1+{\theta }_2+···+{\theta }_i\right)$$ (Equation 3)

$$y=l_1\mathit{sin}{\theta }_1+l_2\sin \left({\theta }_1+{\theta }_2\right)+···+l_i\cos \left({\theta }_1+{\theta }_2+···+{\theta }_i\right)$$ (Equation 4)

Where l₁ represents the length of joint 1; l₂ represents the length of joint 2; l_i represents the length of joint i; θ₁ is the variable of joint 1; θ₂ is the joint 2 variable; θ_i is the joint i variable.

The forward kinematic model of a serial robot is established by determining the end-effector’s pose relative to the base coordinate frame based on the joint angles.⁹¹ To create the forward kinematic model, it is essential to first establish the robot’s coordinate frames according to the widely used D-H convention. The D-H conventions are as follows.

• (1)The z axis is established along the rotation axis of each joint, with $z_i$ aligned with the direction of rotation of joint $i+1$.
• (2)The x axis is defined along the common perpendicular between the z-axes of joint i−1 and joint i, with the positive direction extending from $z_{i-1}$ to $z_i$.
• (3)Once the x axis and z axis are established, the y axis can be determined using the right-hand rule.
• (4)The translations between the origins of each coordinate frame consist of movements along the x axis and z axis; let di represent the translation along the z axis, while the link length a_i denotes the translation along the x axis.
• (5)The twist angle α_i is defined as the angle of rotation about the x_i−1 axis.
• (6)The angle θ_i is defined as the angle of rotation about the $z_{i-1}$ axis.

In a rotational serial robot, d_i, a_i, and α_i are constant values, while only θ_i is a variable. Taking a 6-DOF serial robot as an example, the coordinate frames are first established. From Figure 12, it is observed that the transformation between the coordinate frames of the robot’s axes can be achieved by first rotating around the z axis by an angle θ_i, followed by translations along the z axis and x axis by d_i and a_i, and finally rotating around the x axis by an angle α_i. The corresponding transformation matrix is expressed as follows:

$${}_i{}^{i-1}A=\text{Rot}\left(Z_{i-1},{\theta }_i\right)\text{Trans}\left(0,0,d_i\right)\text{Trans}\left(a_i,0,0\right)\text{Rot}\left(X_i,{\alpha }_i\right)$$ (Equation 5)

In the above expression, Rot(Z_i−1,θ_i) and Rot(X_i,α_i) represent the rotational homogeneous transformations, while Trans(0,0,d_i) and Trans(a_i,0,0) denote the translational homogeneous transformations.

When only a translational transformation occurs in the coordinate frames within the space, the translations along the x, y, and z axes are defined as a, b, and c, respectively. The corresponding translational transformation matrix is expressed as follows:

$$\text{Trans}\left(a,b,c\right)=\left[\begin{array}{cccc}1& 0& 0& a\\ 0& 1& 0& b\\ 0& 0& 1& c\\ 0& 0& 0& 1\end{array}\right]$$ (Equation 6)

When performing rotations around the three coordinate axes, the corresponding transformation matrices are well-established. The rotation matrices for each axis are defined as follows:

$$\text{Rot}\left(x,\theta \right)=\left[\begin{array}{cccc}1& 0& 0& 0\\ 0& \cos \theta & -\sin \theta & 0\\ 0& \sin \theta & \cos \theta & 0\\ 0& 0& 0& 1\end{array}\right]$$ (Equation 7)

$$\text{Rot}\left(y,\theta \right)=\left[\begin{array}{cccc}\cos \theta & 0& \sin \theta & 0\\ 0& 1& 0& 0\\ -\sin \theta & 0& \cos \theta & 0\\ 0& 0& 0& 1\end{array}\right]$$ (Equation 8)

$$\text{Rot}\left(z,\theta \right)=\left[\begin{array}{cccc}\cos \theta & -\sin \theta & 0& 0\\ \sin \theta & \cos \theta & 0& 0\\ 0& 0& 1& 0\\ 0& 0& 0& 1\end{array}\right]$$ (Equation 9)

In the equation, Rot denotes the rotation transformation, which represents the rotation transformation matrix around a certain vector f originating from the origin.

$$\text{Rot}\left(f,\theta \right)=\left[\begin{array}{cccc}{f_x}^{2}vers\theta +c\theta & f_y{f}_{x}vers\theta -f_{z}s\theta & f_x{f}_{x}vers\theta +f_{y}s\theta & 0\\ f_x{f}_{y}vers\theta +f_{z}s\theta & {f_y}^{2}vers\theta +c\theta & f_z{f}_{y}vers\theta -f_{x}s\theta & 0\\ f_x{f}_{z}vers\theta -f_{z}s\theta & f_y{f}_{z}vers\theta +f_{x}s\theta & {f_z}^{2}vers\theta +c\theta & 0\\ 0& 0& 0& 1\end{array}\right]$$ (Equation 10)

In the equation, sθ and cθ represent sinθ and cosθ, respectively, while versθ denotes the versine function.

By substituting the formulas for the translation and rotation homogeneous transformation matrices into Equation 5, the general transformation matrix is derived:

$${}_i{}^{i-1}A=\left[\begin{array}{c}\mathit{cos}{\theta }_i\\ \mathit{sin}{\theta }_i\\ 0\\ 0\end{array}\begin{array}{c}-\mathit{sin}{\theta }_i\mathit{cos}{\alpha }_i\\ \mathit{cos}{\theta }_i\mathit{cos}{\alpha }_i\\ \mathit{sin}{\alpha }_i\\ 0\end{array}\begin{array}{c}\mathit{sin}{\theta }_i\mathit{sin}{\alpha }_i\\ -\mathit{cos}{\theta }_i\mathit{sin}{\alpha }_i\\ \mathit{cos}{\alpha }_i\\ 0\end{array}\begin{array}{c}{a}_i\mathit{cos}{\theta }_i\\ a_i\mathit{sin}{\theta }_i\\ d_i\\ 1\end{array}\right]$$ (Equation 11)

By substituting the parameters for each joint, the corresponding transformation matrices can be obtained. Ultimately, the kinematic model of the serial manipulator is expressed as the product of these transformation matrices:

$${}_0{}^{n}T\left(q_1,q_2,q_3\cdots q_n\right)={}_0{}^{1}T\left(q_1\right)·{}_1{}^{2}T\left(q_2\right)·{}_2{}^{3}T\left(q_3\right)\cdots {}_{n-1}{}^{n}T\left(q_n\right)$$ (Equation 12)

$${}_0{}^{1}T\left(q_1\right)·{}_1{}^{2}T\left(q_2\right)·{}_2{}^{3}T\left(q_3\right)\cdots {}_{n-1}{}^{n}T\left(q_n\right)=\left[\begin{array}{c}{n}_x\\ n_y\\ n_z\\ 0\end{array}\begin{array}{c}{o}_x\\ o_y\\ o_z\\ 0\end{array}\begin{array}{c}{a}_x\\ a_y\\ a_z\\ 0\end{array}\begin{array}{c}{p}_x\\ p_y\\ p_z\\ 1\end{array}\right]$$ (Equation 13)

In scalar form, the equation is expressed as follows:

$$n_{3\times 1}=n\left(q\right)=n\left(q_1,q_2,q_3\cdots q_n\right)$$ (Equation 14)

$$o_{3\times 1}=o\left(q\right)=o\left(q_1,q_2,q_3\cdots q_n\right)$$

$$a_{3\times 1}=a\left(q\right)=a\left(q_1,q_2,q_3\cdots q_n\right)$$

$$p_{3\times 1}=p\left(q\right)=p\left(q_1,q_2,q_3\cdots q_n\right)$$

The preceding expression can be simplified as follows:

$$P_{12\times 1}=f\left(q\right),q\in R^n$$ (Equation 15)

To determine the matrix:

$${}_0{}^{n}T={}_0{}^{1}T\left(q_1\right)·{}_1{}^{2}T\left(q_2\right)·{}_2{}^{3}T\left(q_3\right)\cdots {}_{n-2}{}^{n-1}T\left(q_{n-1}\right)·{}_{n-1}{}^{n}T\left(q_n\right)$$ (Equation 16)

Since the matrix on the right side is known and the joint variables on the left side are unknown, the right matrix equation is first multiplied by ${}_0{}^1{T}^{-1}$. Equations involving the joint variables are then identified and established, and these variables are solved iteratively until all are determined.

$${}_0{}^1{T}^{-1}·{}_0{}^{n}T={}_1{}^{2}T\left(q_2\right)·{}_2{}^{3}T\left(q_3\right)\cdots {}_{n-2}{}^{n-1}T\left(q_{n-1}\right)·{}_{n-1}{}^{n}T\left(q_n\right)$$ (Equation 17)

$${}_1{}^2{T}^{-1}·{}_0{}^{n}T={}_2{}^{3}T\left(q_3\right)\cdots {}_{n-2}{}^{n-1}T\left(q_{n-1}\right)·{}_{n-1}{}^{n}T\left(q_n\right)$$ (Equation 18)

Although the D-H method is widely applicable to most robotic structures, it may not be directly suitable for specific configurations, such as parallel robots. In such cases, special transformations or modified D-H parameters are required for accurate modeling. For instance, in parallel robots, the presence of multiple kinematic chains can complicate the representation of all degrees of freedom when directly applying the D-H method; thus, it is essential to incorporate closed-loop constraints in the modeling process.

The inverse kinematics of robots can be complex, and unlike forward kinematics, which typically yields unique solutions, inverse kinematics may have multiple solutions. To more accurately reconstruct the joint angles of the robot, this paper employs analytical methods to derive the inverse kinematics equations.

To determine θ₁, θ₂, and θ₃, the relationship equations that connect these angles to p_x, p_y, p_z, and the approach vector a_x, a_y, a_z are established. Meanwhile, θ₄, θ₅, and θ₆ can be solved by computing the matrix inverse. To solve for θ₁, Equation 13 is referred to, which states:

$$P_x=d_a{a}_x-c_1\left(d_4{s}_{23}-a_3{c}_{23}-a_2{c}_2-a_1\right)$$ (Equation 19)

$$P_y=d_6{a}_y-s_1\left(d_4{s}_{23}-a_3{c}_{23}-a_2{c}_2-a_1\right)$$

From this, the following is obtained:

$${\theta }_1=\arctan \frac{-p_y+d_6{a}_y}{-p_x+d_6{a}_x}+k\pi ,\left(k=0,\pm 1\right)$$ (Equation 20)

Let p_x1, p_y1, and p_z1 represent the end-effector pose of Joint 3. From this, the following is obtained:

$$P_x^{\prime }=p_{x1}-{a_{1}c}_1=c_1\left(-d_4{s}_{23}+a_3{c}_{23}+a_2{c}_2\right)$$ (Equation 21)

$$P_y^{\prime }=p_{y1}-{a_{1}s}_1=s_1\left(-d_4{s}_{23}+a_3{c}_{23}+a_2{c}_2\right)$$

$$P_z^{\prime }=p_{z1}-d_1=-d_4{s}_{23}-a_3{s}_{23}-a_2{s}_2$$

Convert the formula：

$$M={P_x^{\prime }}^2+{P_y^{\prime }}^2+{P_z^{\prime }}^2=a_2^2+a_3^2+d_4^2+2a_2\left(a_3{c}_3-d_4{s}_3\right)$$ (Equation 22)

$$N=a_3{c}_3-d_4{s}_3=\frac{M-a_2^2-a_3^2-d_4^2}{2a_2}$$ (Equation 23)

$$P_z^{\prime }=p_{z1}-d_1=-d_4{s}_{23}-a_3{s}_{23}-a_2{s}_2$$ (Equation 24)

From this, the following is obtained:

$$\left(N+a_3\right){\tan \left(\frac{{\theta }_3}{2}\right)}^2+2d_4\tan \frac{{\theta }_3}{2}+N-a_3=0$$ (Equation 25)

$$\tan \frac{{\theta }_3}{2}=\frac{-d_4{s}_3\pm \sqrt{d_4^2+a_3^2-N^2}}{N+a_3}$$ (Equation 26)

$${\theta }_3=2\arctan \frac{-d_4{s}_3\pm \sqrt{d_4^2+a_3^2-N^2}}{N+a_3}+k\pi ,\left(k=0,\pm 1\right)$$ (Equation 27)

It can be referred to Equation 21 to solve for θ₂, which indicates the following:

$$\sqrt{{P_x^{\prime }}^2+{P_y^{\prime }}^2}=\pm \left(d_4{s}_{23}-a_3{c}_{23}-a_2{c}_2\right)$$ (Equation 28)

$${P_z^{\prime }}^2=-d_4{c}_{23}-a_3{s}_{23}-a_2{s}_2$$ (Equation 29)

$${\theta }_2=\arctan \frac{\sqrt{{P_x^{\prime }}^2+{P_y^{\prime }}^2}\left(-d_4{c}_3-a_3{s}_3\right)\pm P_z^{\prime }\left(-d_4{s}_3+a_3{c}_3+a_2\right)}{\sqrt{{P_x^{\prime }}^2+{P_y^{\prime }}^2}\left(-d_4{s}_3+a_3{c}_3+a_2\right)\pm P_z^{\prime }\left(-d_4{c}_3-a_3{s}_3\right)}+k\pi ,\left(k=0,\pm 1\right)$$ (Equation 30)

Based on the three angles obtained, it can be concluded that:

$${}_4{}^{3}A{}_5{}^{4}A{}_6{}^{5}A={\left({}_1{}^{0}A{}_2{}^{1}A{}_3{}^{2}A\right)}^{-1}T$$ (Equation 31)

$${P_z^{\prime }}^2=-d_4{c}_{23}-a_3{s}_{23}-a_2{s}_2$$ (Equation 32)

From the equality of both sides of the equation, it can be derived that the first two rows of the third column on both sides are equal, leading to:

$$c_4{s}_5={-a}_x{c}_1{c}_{23}{-a_{y}s}_1{c}_{23}+a_z{s}_{23}$$ (Equation 33)

$$s_4{s}_5={-a}_x{s}_1+a_y{c}_1$$ (Equation 34)

The calculations yield:

$${\theta }_4=\arctan \frac{{-a}_x{s}_1+a_y{c}_1}{{-a}_x{c}_1{c}_{23}{-a_{y}s}_1{c}_{23}+a_z{s}_{23}}+k\pi ,\left(k=0,\pm 1\right)$$ (Equation 35)

By substituting the known angles into the above steps, the values of θ₅ and θ₆ can be obtained:

$${\theta }_5=\arctan \frac{a_z{s}_{23}{c}_4-a_x\left(c_1{c}_{23}{c}_4+s_1{s}_4\right)-a_y\left(s_1{c}_{23}{c}_4-c_1{s}_4\right)}{{-a}_x{c}_1{s_{23}-a_{y}s}_1{s}_{23}-a_z{sc}_{23}}+k\pi ,\left(k=0,\pm 1\right)$$ (Equation 36)

$${\theta }_6=\arctan \frac{n_z{s}_{23}{s}_4-n_x\left(c_1{c}_{23}{s}_4-s_1{c}_4\right)-n_y\left(s_1{c}_{23}{s}_4+c_1{c}_4\right)}{o_z{s}_{23}{s}_4-o_x\left(c_1{c}_{23}{s}_4-s_1{c}_4\right)-o_y\left(s_1{c}_{23}{s}_4+c_1{c}_4\right)}+k\pi ,\left(k=0,\pm 1\right)$$ (Equation 37)

Kinematic model of parallel CARS

During the process of fracture reduction, the distance and angular relationships between the ends of the fractured bone are often complex. Currently, parallel robots used for fracture reduction surgeries often employ six degrees of freedom to accommodate complex surgical scenarios and address the diverse reduction needs of patients. This paper uses the 6-DOF spherical parallel robot (6-SPS) as a case study to analyze the kinematic forward solution of such a parallel robot.

As shown in Figure 13, the 6-SPS parallel robot is composed of an upper platform, a lower platform, and six telescopic rods. The telescopic rods are composed of an upper section and a lower section. The change in the length of the rods enables adjustments to the pose of the upper platform. The center of the upper platform is designated as O_P, and the centers of the hinge points are designated as P_i (where i = 1, 2, 3, 4, 5, 6). These points are distributed on a circle centered at O_P and with radius r_P, forming a symmetrical hexagon. The points P₁, P₃, P₅, and P₂, P₄, P₆ respectively form two equilateral triangles.

Figure 13.

Schematic diagram of the 6-SPS configuration parallel robot structure

The center of the lower platform is designated as O_B, and the centers of the hinge points on the lower platform are designated as B_i (where i = 1, 2, 3, 4, 5, 6), which collectively form a symmetrical hexagon with a circumradius of r_B. The upper and lower hinge points are arranged such that the corresponding regular hexagons are oriented 180° relative to each other.

Based on the structural characteristics of the 6-SPS parallel robot, its basic structure can be fully characterized by six geometric parameters as follows: the circumradius of the upper platform, denoted as r_P; the circumradius of the lower platform, denoted as r_B; the short edge spacing of the upper platform hinge points, denoted as d_P; the short edge spacing of the lower platform hinge points, denoted as d_B; the minimum length of each rod, denoted as l_min; and the maximum length of each rod, denoted as l_max.

For the sake of analysis, a right-handed coordinate system {P} is established with point O_P as the origin, where the platform normal direction is defined as the z axis, and the line connecting point O_P to the midpoint of points P₁ and P₆ is defined as the x axis. A right-handed coordinate system {B} is similarly established at the lower platform. Due to the symmetry in the arrangement of the upper and lower hinge points, the position vectors of the hinge points on both the upper and lower platforms in coordinate systems {P} and {B} can be expressed as follows:

$$p_i={\left[r_P\mathit{cos}{\theta }_{P_i}{r}_P\mathit{sin}{\theta }_{P_i}0\right]}^T$$ (Equation 38)

$$b_i={\left[r_B\cos {\theta }_{B_i}{\mathrm{r}}_B\sin {\theta }_{B_i}0\right]}^{\mathrm{T}}$$ (Equation 39)

In the equations:

$${\theta }_P=\left[{\eta }_P\frac{2\pi }{3}-{\eta }_P\frac{2\pi }{3}+{\eta }_P\frac{4\pi }{3}-{\eta }_P\frac{4\pi }{3}+{\eta }_P-{\eta }_P\right]$$ (Equation 40)

$${\theta }_B=\left[{\eta }_B\frac{2\pi }{3}-{\eta }_B\frac{2\pi }{3}+{\eta }_B\frac{4\pi }{3}-{\eta }_B\frac{4\pi }{3}+{\eta }_B-{\eta }_B\right]$$ (Equation 41)

$${\eta }_P=\frac{\pi }{3}-\arcsin \left(\frac{d_P}{2r_P}\right)$$ (Equation 42)

$${\eta }_B=\arcsin \left(\frac{d_B}{2r_B}\right)$$ (Equation 43)

In the equations, θ_P represents the angle formed by the line connecting the upper platform origin O_P to the upper hinge point P_i and the positive direction of the X_P axis. θ_B represents the angle formed by the line connecting the lower platform origin O_B to the lower hinge point B_i and the positive direction of the X_B axis. η_P denotes half of the central angle corresponding to the circumcircle of the upper hinge points, associated with the long edges of the upper hinge points. η_B denotes half of the central angle corresponding to the circumcircle of the lower hinge points, associated with the short edges of the lower hinge points. The distribution of the hinge points is illustrated in Figure 14.

Figure 14.

Hinge point distribution

The kinematic analysis of the 6-SPS parallel robot includes both kinematic inverse solutions and kinematic forward solutions. The kinematic inverse solution is relatively straightforward to solve, while the kinematic forward solution is more complex. The kinematic inverse solution serves as the foundation for solving the kinematic forward solutions and the dynamic inverse solutions. In this section, the kinematic inverse solution for the 6-SPS parallel robot is derived using the closed-loop vector method and differentiation. This includes position inverse solutions, velocity inverse solutions, and acceleration inverse solutions, as well as the derivation of the velocity Jacobian matrix.

Based on the spatial position vector relationships, the vector expression for each actuator can be written as follows:

$$b_i+l_i{s}_i=\mathrm{t}+p_i$$ (Equation 44)

From Equation 44, the following is obtained:

$$s_i=\frac{t+p_i-b_i}{l_i}$$ (Equation 45)

$$l_i=\Vert t+p_i-b_i\Vert$$ (Equation 46)

Let l_i be the length of the i-th actuator. According to the properties of the rotation matrix, the relationship between the angular velocity vector ω_P of the upper platform and the time rate of change of the Euler angles can be expressed as follows:

$${\omega }_P=E\left[\dot{\begin{array}{c}{\varphi }_X\\ \dot{{\varphi }_Y}\\ \dot{{\varphi }_Z}\end{array}}\right]$$ (Equation 47)

$$E=\left[\begin{array}{ccc}c{\varphi }_Z{c\varphi }_Y& -s{\varphi }_Z& 0\\ s{\varphi }_{Z}c{\varphi }_Y& c{\varphi }_Z& 0\\ -s{\varphi }_Y& 0& 1\end{array}\right]$$ (Equation 48)

By differentiating Equation 23 with respect to time, the velocity vector of P_i in the fixed coordinate system can be obtained:

$$v_{Pi}=v_P+{\omega }_P\times p_i$$ (Equation 49)

In this equation, v_Pi represents the velocity vector of the i-th hinge point on the upper platform in the lower platform’s reference frame, and v_P represents the velocity vector of the origin of the moving coordinate system. By multiplying both sides of Equation 49 by s_i, the stretching velocity of the i-th actuator can be obtained:

$$\dot{l_i}=v_{Pi}\cdot s_i=v_P\cdot s_i+\left(p_i{\times s}_i\right)\cdot {\omega }_P$$ (Equation 50)

The stretching velocities of the six actuators can be combined into matrix form as follows:

$$\dot{l}=J_P\times \dot{x_P}$$ (Equation 51)

In the equation, $\dot{x_P}={\left[\begin{array}{cc}{v}_P^T& v_P^T\end{array}\right]}^T$ represents the 6D velocity vector of the upper platform, while $\dot{l}={\left[\begin{array}{cc}\begin{array}{cc}\dot{l_1}& \dot{l_2}\end{array}& \end{array}\begin{array}{cc}\begin{array}{cc}\dot{l_3}& \dot{l_4}\end{array}& \begin{array}{cc}\dot{l_5}& \dot{l_6}\end{array}\end{array}\right]}^T$ represents the vector of stretching velocities of the six actuators. The Jacobian matrix J_P describes the relationship between the 6D velocity vector of the upper platform $\dot{x_P}$ and the stretching velocities $\dot{l}$ of the six actuators.

$$J_P=\left[\begin{array}{c}{s}_1^T\\ ⋮\\ s_6^T\end{array}\begin{array}{c}{\left(p_1+s_1\right)}^T\\ ⋮\\ {\left(p_6+s_6\right)}^T\end{array}\right]$$ (Equation 52)

By differentiating Equation 50 with respect to time, the stretching accelerations of the six actuators can be obtained as follows:

$$\ddot{l}=\frac{d{J}_P}{dt}\dot{x_P}+J_P\ddot{x_P}$$ (Equation 53)

In the equation, $\ddot{x_P}={\left[\begin{array}{cc}{\dot{v}}_P^T& {\dot{v}}_P^T\end{array}\right]}^T$ represents the acceleration vector of the upper platform.

For a 6-SPS parallel robot, the 6D pose of the upper platform, denoted as $q={\left[xyz{\varphi }_X{\varphi }_Y{\varphi }_Z\right]}^{\mathrm{T}}$, consists of both translational and rotational components. The translational component represents the position vector $t={\left[xyz\right]}^{\mathrm{T}}$, which denotes the location of the origin of the upper coordinate system in the fixed coordinate system. The rotational component describes the angles of rotation of the moving coordinate system relative to the fixed coordinate system, represented by ZYX Euler angles, $\Omega ={\left[{\varphi }_X{\varphi }_Y{\varphi }_Z\right]}^{\mathrm{T}}$. Consequently, the rotation matrix of the moving coordinate system {P} with respect to the fixed coordinate system {B} can be expressed as:

$${}_B{}^{P}R=R_Z\left({\varphi }_Z\right)R_Y\left({\varphi }_Y\right)R_X\left({\varphi }_X\right)$$ (Equation 54)

For a 6-SPS parallel robot, the forward kinematics solution involves determining the pose based on the lengths of the six driving rods. In practical applications, a common approach involves installing displacement sensors in each branch to obtain the actual pose of the upper platform through the forward kinematics solution. The Newton-Raphson iterative method is a widely used technique for solving the forward kinematics problem, which provides high precision in the calculated pose with negligible error. This paper also employs the Newton-Raphson iterative method to derive the forward kinematics solution for the 6-DOF parallel robot.

The 6D pose of the upper platform is represented as $q={\left[xyz{\varphi }_X{\varphi }_Y{\varphi }_Z\right]}^{\mathrm{T}}$. From Equation 44, the following is obtained:

$${\dot{x}}_P=\left[\begin{array}{cc}I& 0\\ 0& E\end{array}\right]\dot{q}$$ (Equation 55)

Let $\mathrm{J}=J_P\left[\begin{array}{cc}I& 0\\ 0& E\end{array}\right]$. From Equation 51, the following is obtained:

$$\dot{l}=J\dot{q}$$ (Equation 56)

In this equation, J represents the Jacobian matrix relating the time derivative of the 6D pose to the extension velocities of the six driving rods. By multiplying both sides of the equation by J⁻¹, the following is obtained:

$$\dot{q}=J^{-1}i$$ (Equation 57)

Assuming the initial pose is q₀, the estimated value after one iteration of numerical computation from the above equation is given by:

$$q_1=q_0+J^{-1}\left(l_m-l_0\right)$$ (Equation 58)

In this equation, l₀ = [l₁₀ l₂₀ l₃₀ l₄₀ l₅₀ l₆₀]^T represents the lengths of the six rods at the initial pose during the iteration, while l = [l₁ l₂ l₃ l₄ l₅ l₆]^T denotes the actual measured lengths of the six rods. The iterative formula for the forward kinematics algorithm is expressed as:

$$q_{n+1}=q_n+J^{-1}\left(l-l_n\right)$$ (Equation 59)

Based on this iterative formula, the forward kinematics solution for the 6-SPS parallel robot can be derived. The iterative steps for the forward kinematics solution are as follows.

• (1)An initial pose value, denoted as q₀, should be selected.
• (2)The value of q₀ should be substituted into the relevant equations to obtain the lengths of the six driving rods at pose q₀, denoted as l₀.
• (3)The exit condition should be checked: it is necessary to determine if ||l−l₀||≤ε, where ε represents a predefined accuracy threshold. If the condition is satisfied, the iteration should be terminated; q₀ will represent the forward kinematics solution corresponding to the currently measured rod lengths. If the exit condition is not met, the next step should be executed.
• (4)The value of l₀ should be substituted into the iterative formula to compute q₁. q₁ should be used as the new initial pose value, and iteration should continue until the exit condition is satisfied. The final computed pose value will represent the forward kinematics solution corresponding to the currently measured rod lengths.

It is important to note that, when employing the Newton-Raphson iterative method to solve the forward kinematics of parallel robots, the selection of the initial pose is crucial. Selecting an appropriate initial pose can reduce the number of iterations and accelerate the solution process; conversely, an inappropriate initial value may increase computation time and potentially lead to incorrect solutions. When establishing the initial iterative pose, it is common practice to select the configuration in which each driving rod is positioned at its midpoint. If the input rod length values significantly differ from the midpoint lengths of the driving rods, multiple sets of rod lengths may be utilized as intermediate inputs between the target and midpoint lengths. The kinematic solution obtained from these intermediate inputs can subsequently be employed as the initial pose for the next iteration, thereby mitigating the impact of the initial pose on the algorithm. Furthermore, if the target point coincides with a singular configuration of the parallel robot, the forward kinematics algorithm may fail to yield a valid solution.

Application of CARS

Orthopedic surgical robots are used in clinics for spinal surgery, joint replacement surgery, limb fracture reduction surgery, and other fracture trauma procedures.

Spinal surgery robot

During conventional spinal fracture reduction surgery, a screw is inserted into the pedicle to finalize the reduction of the spinal fracture site. Incorrect screw positioning can result in temporary or permanent nerve injury, cerebrospinal fluid leaking, and hemorrhage. Using fluoroscopy during screw installation surgery can be harmful to the surgeon’s health. Researchers are developing and enhancing RS for spinal fracture reduction surgery to minimize risks for patients and surgeons. Spinal surgery robots can be categorized into minimally invasive spinal surgery robots and open surgical robots based on the type of surgery they perform.

Guided spinal surgery robots are commonly used in spine surgical procedures. They assist in stabilizing the vertebral body using robotic guidance and are primarily used in navigation-based procedures like interbody fusion and spinal orthosis.

Mazor created the first spinal surgery robot, Spine Assist, in 2001.⁹² As is shown in Figure 15A, it uses a 6-DOF Stewart parallel mechanism and provides precise positioning and guidance for the spine. Renaissance, the second-generation system, offers improved performance.⁹³ SpineAssist/Renaissance has shown a 96% success rate in pedicle screw implantation surgery, surpassing the traditional approach.⁹⁴ Both systems are FDA-approved for clinical usage. Several organizations and researchers have conducted significant research on spinal surgery robots. Badaan et al.⁹⁴ designed a robotic needle driver for image-guided procedures, demonstrating that continuous needle rotation could reduce targeting errors by up to 70% in controlled experiments which is shown in Figure 15B. Balicki et al.⁹⁵ developed an image-guided robot tailored for hybrid ORs to enhance pedicle screw placement accuracy in minimally invasive spine fusion surgery which is shown in Figure 15C. The system integrates augmented reality navigation and achieves a targeting error of 1.27 ± 0.57 mm in cadaveric studies. He et al.⁹⁶ evaluated the bi-planar navigation system for femoral neck fracture surgeries, demonstrating a reduction in fluoroscopy time and surgical errors. The system outperformed freehand methods in screw placement accuracy and operational efficiency which is shown in Figure 15D. As is shown in Figure 15E, Chen et al.⁹⁷ proposed a 3D robot-assisted navigation system for spine surgery, showing high localization accuracy (2.54 ± 0.15 mm) and reliable surgical path planning in experimental settings.

Figure 15.

Guided spinal surgery robot

(A) SpineAssist robot. Reproduced with permission from ref.⁹². Copyright 2006 John Wiley & Sons, Ltd.

(B) AcuBot robot. Reproduced with permission from ref.⁹⁴. Copyright 2011 John Wiley & Sons, Ltd.

(C) CT-Bot prototype. Reproduced with permission from ref.⁹⁵. Copyright 2022 John Wiley & Sons, Ltd.

(D) Robotic system using a bi-planar fluoroscopy. Reproduced with permission from ref.⁹⁶. Copyright 2019 The Authors. Published by John Wiley & Sons,Ltd. Licensed under the CC BY 4.0 license.

(E) A body-mounted surgical assistance robot system. Reproduced with permission from ref.⁹⁷. Copyright 2019 John Wiley & Sons, Ltd.

(F) Star marker connected to the surgical arm. Reproduced with permission from ref.⁴⁶. Copyright 2020 Elsevier.

In 2016, Mazor launched the Mazor X robotic surgical system, which combines a 6-DOF serial manipulator and navigation system for preoperative CT scans and real-time guidance during surgery. Following its acquisition by Medtronics, Mazor’s Stealth navigation software was integrated with a robotic arm to create the Mazor X Stealth robot, enabling real-time positioning through photoelectric navigation. As is shown in Figure 15F, the integration of “Robot $+$ Navigation” has become the standard approach for guided spinal surgery robots.⁹⁸ Other systems that utilize this technology include ROSA Spine, Tianji robot, and ExcelsiusGPS. ROSA Spine has undergone extensive clinical trials, while ExcelsiusGPS is currently used in clinical settings and is known for its reduced learning curve and user-friendly system.⁹⁹

Active spinal surgery robots can perform needle placement, drilling, nail placement, laminae grinding, and other tasks. They have addressed the surgical challenges associated with older methods, including hand drilling, nail placement, and radiation exposure. The first active orthopedic robot for joint replacement was released in the United States in 1992, while research on active spinal surgical robots started in the early 2000 s. Table 2 outlines the current details of the active spinal surgical robots produced by different scientific research institutes and firms worldwide, including the proposed time, system composition, and implementation functions. Despite significant progress achieved by researchers and medical device firms in developing spinal robot systems, the advancement of active spine surgery robots is now limited to the prototype level mainly due to medical ethical concerns. Currently, only a limited number of commercial drilling and nail placement robotic surgical systems are being used in real clinical settings.

Table 2.

Application of active spinal surgery robots

Author	System composition	Function realization	Characteristic
Chung et al.100	Surgical robot, planning system, optical tracking system	Instrument guided, automatic drilling	Drilling and nail placement accuracy is improved
Ortmaier et al.101	Surgical robot system, planning system, optical tracking system, light robot arm	Multiple types of spinal surgery	Can perform many types of surgery
Zhang et al.102	Six degrees of freedom robotic arm, six dimensional force/moment sensor	Autonomous drilling	Effectively reduce radiation exposure
Jin et al.103	Series 6° of freedom robotic arm, force sensor	Recognition of solid layer status by dragging and drilling of mechanical arm	Assist the doctor to complete pedicle nail fixation
Rezazadeh et al.104	Six degrees of freedom robotic arm	Autonomous drilling	Precise remote operation
Smith et al.105	Six degrees of freedom robotic arm, end effector, quick change mechanism	Automatic needle placement, drilling, tapping, nail placement	Automatic rapid instrument replacement
Opfermann et al.106	Five degrees of freedom mechanical arm, piezoelectric material	Surgical instrument placement	Flexibility and accuracy are improved
Li et al.107	Six degrees of freedom robotic arm, binocular vision system, intelligent bone drill, master car	Automatic Kirschner needle placement	Minimally invasive, flexible and highly automated
Chen et al.108	Master-slave series-parallel hybrid manipulator	Control the hand actuator for drilling and nail placement	Lighten the burden on doctors during surgery

Joint replacement surgery robot

Joint replacement surgery mainly refers to hip or knee replacement, which, from a clinical perspective, does not belong to the traditional category of fracture reduction. In some special cases, such as complex fractures with joint surface injury, joint replacement may be an option; however, this involves a more comprehensive reconstruction of the joint structure rather than simple fracture reduction.^109,110 Considering the correlation between joint replacement robots and CARS, as well as the universality of the technology, both joint replacement robots and CARS are included within the scope of orthopedic surgery robots, and a detailed elaboration is provided. Joint replacement robots are the primary focus of research in orthopedic surgical robot systems. They offer advanced surgical planning and precise surgical procedures. Researchers are incorporating robot-controlled surgical tools into joint replacement procedures. The RS uses CT-based 3D imaging to plan the cutting path and perform tasks like bone cutting, using computer-controlled cutting devices.

In 1988, Honl et al.¹¹¹ proposed an RS with two planar robotic arms for joint physical rehabilitation therapy which is shown in Figure 16A. In 1992, Marelli et al.¹¹² created a robot-assisted system for knee prosthesis installation, allowing surgeons to perform accurate joint removal. Musits et al.¹¹³ developed ROBODOC Surgical Assistant, a robot for joint replacement surgery that can create chambers for prosthetic implants with high accuracy. Bauer et al.¹¹⁴ confirmed the precision of robot-assisted implant insertion in hip replacement surgery.

Figure 16.

Joint replacement surgical robot

(A) The ROBODOC surgical robot.

(B) Mini bone-attached robotic system. Reproduced with permission from ref.¹¹⁵. Copyright 2005 Robotic Publications, Ltd.

(C) Overview of the developed bone cutting robot. Reproduced with permission from ref.¹¹⁶. Copyright 2018 Elsevier.

(D) Novel Image-free handheld robot. Reproduced with permission from ref.¹¹⁷. Copyright 2022 Elsevier.

(E) Image-guided CARS joint fracture surgery. Reproduced with permission from ref.¹¹⁸. Copyright 2017 The Authors. Licensed under the CC BY 4.0 license.

(F) Skywalker robot. Reproduced with permission from ref.¹¹⁹. Copyright 2021 The Authors. Published by Elsevier. Licensed under the CC BY-NC-ND license.

Subsequently, various RS for joint orthopedic surgery have been developed. Mitsuishi et al.¹²⁰ created a 9-axis bone-cutting machine for enhanced precision and reduced patient hospitalization. Wolf et al.¹¹⁵ proposed an active microrobotic system for creating bone chambers in joint replacement surgery which is shown in Figure 16B. Liao et al.¹¹⁶ developed a mechanistic model for predicting cutting force and temperature during bone milling, incorporating bone anisotropy and osteon orientation which is shown in Figure 16C. The model demonstrated improved accuracy compared to the Johnson-Cook model and was validated with experimental data to assist robotic surgery and optimize cutting parameters. Kim et al.¹²¹ created a robotic surgical device for bone cutting in minimally invasive joint replacement procedures. Boiadjiev et al.¹²² developed a 2-DOF bone cutting tool to address precision and surface smoothness issues. Yen et al.¹²³ created a surgical robot for joint replacement surgery with CT-free navigation software and coordinated control. Doan et al.¹¹⁷ investigated the accuracy of the VELYS robotic-assisted system for total knee arthroplasty (TKA) in a cadaveric study which is shown in Figure 16D. Results showed superior alignment accuracy in the femoral and tibial coronal and sagittal planes compared to conventional instrumentation, with fewer outliers in resection alignment. Giulio et al.¹¹⁸ developed an image-guided CT/fluoroscopy image registration framework for robot-assisted joint surgery, enabling precise fracture reduction and minimizing soft tissue damage which is shown in Figure 16E. Shalash et al.¹²⁴ created a novel technology for autonomous bone cutting in joint replacement surgery, enhancing the smoothness of the bone surface.

Mitra et al.¹²⁵ introduced the NAVIO system, a robot-assisted device designed to enhance the precision of implant placement and soft tissue balance in joint replacement surgery, specifically for partial and complete knee replacements. The system design not only enhances the joint replacement procedure but also minimizes extra radiation exposure from CT scans and utilizes a handheld robotic tool for precise cutting. Research involving 128 patients who underwent joint replacement surgery using NAVIO revealed a 99.2% survival rate for knee implants during the procedure. In 2020, Agarwal et al.¹²⁶ conducted an extensive comparative analysis between conventional and robot-assisted joint replacement surgeries, concluding that robot-assisted surgery is superior to traditional joint replacement surgery. Image comparison results demonstrated higher accuracy of robot-assisted surgery compared to traditional joint replacement surgery. Rahman et al.¹²⁷ evaluated the clinical precision of alignment and alignment correction in robot-assisted joint replacement surgery, finding that positioning errors are comparable to implantation errors in static reference robot-assisted technology. As is shown in Figure 16F, Xia et al.¹¹⁹ examined clinical data from joint replacement surgeries utilizing a Skywalker robot from June 2020 to January 2021. Upon the robotic arm reaching the specified position, the surgeon performs the bone cut using the cutting fixture and evaluates both the actual versus expected cut thickness and the lower leg’s alignment pre- and post-operation. The analysis suggests that this approach provides high osteotomy accuracy, potentially aiding surgeons in achieving precise bone excision and lower limb alignment in clinical settings.

Although numerous systems have been developed and prototyped, only a few have been successfully implemented in clinical settings. Currently, widely used joint surgery robots include the Mako, Navio, Rosa, and VELYS^TM systems.

Limb fracture reduction CARS

Fractures of long bones in extremities are more common in industrial manufacturing, construction, and transportation incidents. The fracture rate of long bones and limbs is increasing due to advancements in production, construction, and transportation. The reduction process must overcome significant traction forces due to coverage by diverse muscle groups, nerves, and soft tissues. Current machines for fracture reduction surgery do not meet necessary standards. Consequently, numerous researchers have conducted various studies on limb fracture reduction CARS, achieving significant progress and advancements.

Browbank et al.¹²⁸ integrated computer and mechanical control technologies with surgical fracture treatment methods to develop a robot vision system and specialized reset manipulator using fluoroscopic images. They demonstrated the potential of medical robot technology, outlined the fundamental criteria for robot reset, and addressed safety and sterility requirements. Seide et al.⁶¹ developed a six-axis automatic fixator employing the Stewart platform and TSF technology, incorporating an electric drive and a load-sensing device on six driving rods. The doctor inputs the desired action into the software, which automatically directs the external frame to precisely reduce deformities and fractures and measure bone healing stress. Joskowicz et al.¹²⁹ examined issues with existing fracture reduction techniques, developed an RS to assist with intramedullary nail locking in long bone fractures, and discussed the objectives and capabilities of a computer-assisted surgical system. This technology primarily aids in locating the keyhole of the intramedullary nail, potentially reducing the need for intraoperative fluoroscopies, minimizing radiation exposure for the surgeon, and shortening procedure duration. Oszwald et al.¹³⁰ developed a reset control system that incorporates touch and distance feedback, using a joystick for control input, two orthogonal cameras for enhanced imaging, and an industrial robot RX90 as the reset actuator. Reduction of synthetic femur fractures was performed in 15 cases under various settings, and a comparison of the reduction outcomes indicated that radiation exposure time can be minimized and reduction accuracy improved with robot assistance. The study was limited to experiments on artificial bones and did not involve further tests. The University of Tokyo and Osaka University in Japan collaborated to develop the surgical robot FRAC-Robo to assist in reducing femoral fractures.¹³¹ In 2008, this collaboration also resulted in the development of the FRAC-Robo system, which facilitated anatomical reduction through robot-assisted traction and rotation. Moreschini et al.¹³² developed a novel robot for reducing lower limb fractures, featuring automatic control of knee joint flexion and extension, independent traction of thighs and legs, foot rotation, and additional functions to assist orthopedic surgeons in fracture reduction through motion control. Abedinnasab et al.¹³³ developed an innovative robotic device with six degrees of freedom to treat long bone fractures. Experimental testing has demonstrated that this technology holds significant promise for clinical application by minimizing the need for repetitive procedures and reducing radiation exposure for surgeons and patients. Current approaches to designing fracture reduction trajectories are hindered by inefficient collision detection, lengthy reduction paths, and increased muscle strain, which challenges the provision of precise and safe therapy. Li et al.¹³⁴ introduced an automated planning approach integrating collision detection, force analysis, and shortest path planning to address these issues which is shown in Figure 17A. Experimental results demonstrate that the method exhibits short duration, high reduction efficiency, and superior trajectory safety, providing a novel approach for precise and safe fracture resetting treatment. Bang et al.¹³⁵ developed a robotic bone fracture reduction system based on a hexapod structure, achieving precise alignment control while reducing radiation exposure to 0.11 mSv compared to 0.52 mSv in conventional methods. As is shown in Figure 17B, Preclinical trials using bovine bone models showed improved accuracy with fewer alignment errors and reduced surgeon fatigue. In 2020, Jingtao et al.¹³⁶ introduced a safety technique for human-machine collaborative surgery in robot-assisted long bone fracture reduction, focusing on reducing route envelope error and enhancing the artificial force field. Simulation results show that the safety method significantly enhances the safety and accuracy of the robot-assisted resetting operation. Lee et al.¹³⁷ introduced and fine-tuned a new intramedullary limb extension robot to address mechanical and electromagnetic issues from the miniaturization of medical devices. The RS has been shown to effectively decrease treatment duration and minimize postoperative complications, as indicated by modeling and experimental data. Zhu et al.⁸ introduced a teleoperated robotic system for femoral shaft fracture reduction. The system minimized alignment errors in artificial lower-limb models while reducing radiation exposure which is shown in Figure 17C. Experiments demonstrated its high repeatability and operability, with mean axial displacement errors of 1.8 mm and rotational errors of 1.2°.

Figure 17.

Limb fracture resetting robot

(A) The parallel orthopedic robot. Reproduced with permission from ref.¹³⁴. Copyright 2022 The Authors. Published by MDPI. Licensed under the CC BY license.

(B) The limb fracture reduction robot. Reproduced with permission from ref.¹³⁵. Copyright 2024 IPEM.

(C) Integration of robot and optical tracking system. Reproduced with permission from ref.⁸. Copyright 2017 Elsevier Ltd.

(D) Heterogeneous master-slave robot. Reproduced with permission from ref.¹³⁸. Copyright 2024 John Wiley & Sons Ltd.

The field of limb fracture reduction CARS is advancing rapidly, with ongoing research and applications driving innovation in orthopedic surgery. It is anticipated that these robots will become routine tools for orthopedic trauma therapy in the future.

Universal fracture reduction CARS

The single-type CARS is specifically designed for a particular type of fracture, such as limb fracture reduction. Such robots may be highly specialized in specific areas but are limited in their versatility and application across different fracture types. Unlike robots designed for a single type of fracture reduction, universal CARS are intended to meet the treatment needs of a wide range of fracture types, from simple limb fractures to complex intra-articular fractures. Their design emphasizes versatility and adaptability, allowing for use in different fracture reduction surgical scenarios by replacing components or adjusting procedures. To address the application limitations of single-type CARS, many researchers have conducted various studies on universal CARS.

In 1989, a study on computer-assisted hip replacement surgery was conducted at the University of California Davis Hospital.¹³⁹ The paper highlights the system’s application in plastic surgery, head and neck surgery, and cancer surgery, marking it as the first concept of a universal surgical robot. Malvisi et al.¹⁴⁰ introduced an innovative RS that integrates electronics, controls, and a new mechanical design for precise positioning and safe interaction in surgical settings. Evaluated in vitro, the technology has shown encouraging outcomes in robot-assisted joint replacement surgery, suggesting a potential new approach in orthopedic surgery. Glozman et al.¹⁴¹ developed a new registration strategy to explore the geometric relationships between reference frames, and reported study results on robot-assisted registration methods in orthopedic surgery. This registration method was evaluated with a 6-DOF parallel robot designed for medicinal purposes. The robot’s six-axis force sensor rapidly and precisely determines the surface’s normal direction at the sampling location. This technology demonstrated a seamless integration between robotic and collaborative control in surgical procedures. Fu et al.¹⁴² introduced an innovative robot-assisted technique for achieving fracture site docking. The technique aids surgeons in identifying fractured long bones and securing intramedullary nails. Screw insertion necessitates an operating table, a precision fluoroscopy device, a fracture reduction robot, a guiding robot, a control system, and a surgeon control platform.

Kong et al.¹⁴³ developed HIT-RAOTS, a robotic-assisted orthopedic remote surgery system that provides doctors with fracture information by capturing imaging data that shows the fracture’s position and orientation, aiding in fracture reduction surgery and screw placement. The system integrates a 6-DOF force sensor robot with a parallel manipulator powered by six AC servo motors, ensuring precise placement in remote surgeries. Hu et al.¹³⁹ classified orthopedic robots into categories such as computer control, device ontology, sensors, vision systems, and operating systems based on clinical features. They proposed a clinical safety strategy for the orthopedic surgical robot based on the system components' functions and characteristics. The authors summarize the clinical applications of surgical robots, explore research hotspots, and discuss potential applications, providing a foundational design technique. Wu et al.¹⁴⁴ integrated an existing robot-assisted surgery system to explore the creation of virtual surgical environment models within a VR setting. Computer graphics and medical imaging combine to accurately model the complex robot system, patient, and bone during surgery. An accurate virtual surgical setting is crucial for preparation, simulation, and conducting remote surgeries.

Xia et al.¹⁴⁵ developed an image-guided RS for neurosurgery to prevent unintentional damage to neurovascular systems during surgical drilling. As is shown in Figure 18A, this system integrates a navigation system, a 6-DOF robotic arm, and 3D visualization software, enabling precise collaboration among surgeons. Clinical trials show that this RS protects neurovascular structures by stabilizing the drill bit and employing a virtual fixture, thus improving drilling safety. Sun et al.¹⁴⁶ introduced a novel robot-assisted system for fracture reduction, enabling remote surgery. The system includes a traction device, a reset unit, and a remote operating system. Experimental results confirm the system’s reset accuracy, demonstrating an average axial error of 1.86 mm, transverse error of 1.48 mm, vertical error of 1.9 mm, and rotational inaccuracy of 2.1°. These results validate the efficacy of the suggested robot-assisted fracture reduction system and highlight its potential for further research and development. Dagnino et al.¹⁴⁷ developed a 3D imaging system specifically for virtual joint reduction. This device processes and segments fracture images from CT scans to create 3D models of bone fragments, displayed on a screen. After virtually reducing the fracture, the surgeon performs the actual reduction using a robotic manipulator. The system underwent a fracture reset test, revealing a reset accuracy of 1.04 ± 0.69 mm and 0.89 ± 0.71°. Dagnino et al.¹⁴⁸ introduced a novel robot-assisted fracture surgery system featuring an innovative robotic structure and live 3D imaging. This method primarily targets distal femoral fractures but can be adapted for other surgical areas. The system’s functionality was demonstrated through 10 distal femoral fracture reductions, with trial outcomes validating its precision, efficiency, and safety. Peng et al.¹⁴⁹ examined features of minimally invasive surgery, designed a novel robot-assisted system for precise tool placement and alignment, and evaluated the workspace, singularity, and kinematics of the hybrid robotic arm. This system guides the structural design of minimally invasive surgical robots.

Figure 18.

Universal fracture reduction robot

(A) Robotic assistance for skull base surgery. Reproduced with permission from ref.¹⁴⁵. Copyright 2022 John Wiley & Sons, Ltd.

(B) Robot-assisted bone fracture reduction system. Reproduced with permission from ref.⁶⁹. Copyright 2016 Elsevier.

(C) Percutaneous fragment manipulation device. Reproduced with permission from ref.¹⁵⁰. Copyright 2019 The Authors. Licensed under the CC BY license.

(D) Bone connection robotic hand. Reproduced with permission from ref.¹⁵¹. Copyright 2023 Elsevier Ltd.

Kim et al.⁶⁹ proposed a force-feedback system to reduce the excessive force applied by the CARS, providing surgeons with sensory feedback on the contact force between bones which is shown in Figure 18B. This force feedback system is expected to enhance the reliability of the RS and reduce secondary soft tissue damage during surgery. Georgilas et al.¹⁵⁰ introduced an innovative transdermal device for manually or robotically manipulating bone pieces which is shown in Figure 18C. An optical tracking system quantifies the device’s deformation under load. The results show that the system’s performance matches the simulation outcomes. The percutaneous fragment manipulation device effectively achieves fracture reduction and facilitates external fracture fixation. Liu et al.¹⁵¹ proposed an innovative system integrating robotics and 3D printing for tibial shaft fracture reduction which is shown in Figure 18D. The method demonstrated high reduction accuracy with mean alignment errors of 2.78° and length errors of 1.95 mm in phantom bone models, offering significant improvements in precision and radiation safety.

The universal CARS is characterized by its wide adaptability, high flexibility, precise operation capabilities, integration of advanced technology, and powerful software support. It meets the reduction needs of different fracture types through modular design and multifunctional accessories, utilizes high-precision imaging and navigation technology to achieve accurate positioning, integrates intelligent control algorithms to adapt to complex changes during surgery, and emphasizes surgical safety and operator interactivity. Additionally, it promotes the development of telemedicine, has the potential for teleoperation, and provides comprehensive education and training support to improve the skills of medical personnel and surgical success rates. As technology continues to advance and clinical evidence accumulates, it is expected that such robots will be more widely promoted and used worldwide.

Computer-assisted fracture reduction technique

Preoperative image-guided reconstruction of broken bone model

In complex fracture cases, surgeons cannot gather sufficient information about fracture fragments from 2D imaging.¹⁵² Surgeons perform anatomical reductions through open surgery to accurately align and join the damaged bone. After reduction, the fracture site is stabilized with plates and screws.¹⁵³ Open surgery, which requires cutting through soft tissue to access bone, increases the risk of infection. The CARS uses small incisions in minimally invasive surgery to join fracture pieces, contrasting with traditional methods.¹⁵⁴ Combined robot technology and 3D imaging guidance enhance the alignment accuracy of fracture fragments in robot-assisted surgeries. This method uses images from modalities like CT and fluoroscopy to guide minimally invasive fracture reduction surgeries. Using image-guided and robot-assisted technologies enables surgeons to perform precise fracture reduction surgeries with minimal soft tissue damage.¹⁵⁵ Image guidance technology analyzes the patient’s CT scan to create a 3D model of the fracture site. Surgeons upload the 3D model to a virtual reduction environment, plan the procedure, adjust the robotic arm by simulating the actual reduction, and create motion instructions.¹⁵⁶ The RS executes these commands to perform the fracture reduction.

Preoperative registration technique

The quality of registration directly impacts surgical precision. RS prioritize registration over computer-aided navigation due to the integration of robotic equipment in surgeries. Prior to orthopedic surgery, surgeons must register patient bones, 3D models, and robotic equipment. Currently, three prevalent registration methods are used: reference point,¹⁵⁷ surface,¹⁵⁸ and image registration.¹⁵⁹

The basic point registration method, an advanced form of the paired point matching approach, requires minor surgery before a preoperative CT scan to insert reference points into the patient’s bone.¹⁶⁰ These reference points guide the treatment during surgery; however, additional procedures may cause pain or complications at the insertion site. Surface registration employs the iterative closest point algorithm and least squares method to align the patient’s 3D bone model with the actual bone surface.¹⁶¹ The shape registration method starts with paired points for baseline registration and enhances surface registration using specific points. Its advantage lies in providing real-time updates on the surgical process, surpassing conventional methods. Image registration uses fluoroscopic images taken during orthopedic surgery.¹⁶² Although confirmed in laboratory settings, this method is not widely used in clinical surgical robots yet.

Intraoperative navigation techniques

Surgical navigation technology is widely used in clinical surgery. Researchers continuously focus on enhancing precision, accuracy, and predictability in fracture reduction surgery. In robot-assisted fracture reduction surgery, the system uses pre-surgery images for computer simulations to devise the optimal virtual 3D operation plan. The system employs a navigation pointer to monitor and track the operation’s progress in real time. Currently, validated computer-aided navigation technologies include infrared optical, electromagnetic, and ultrasonic navigation, categorized by their tracking methods.

The infrared optical navigation system detects patients and surgical tools through the infrared light it emits and detects. As the oldest type of navigation technology, it is widely utilized in orthopedic surgery for its superior precision.¹⁶³ Kahler et al.¹⁶⁴ proposed an optical tracking surgical device to mitigate viral infection risks during pelvic internal fixation surgeries. This device uses computed tomography to capture the patient’s anatomical structure and aid in screw placement. Digioia et al.¹⁶⁵ developed a navigation system enabling surgeons to measure their procedures precisely in real time, integrating preoperative planning, motion modeling of surgical instruments, real-time measurement, and experimental functional verification. Digioia et al.¹⁶⁶ introduced a mini-incision technique using this system to reduce surgical wound sizes. Weber et al.¹⁶⁷ combined the navigation system with a bone clamp, robotic arm, and operating table to create a device for the precise alignment of fractured bones which is shown in Figure 19A. Wan et al.¹⁶⁸ introduced a novel motionless calibration method for augmented reality surgery navigation systems, which simplified the calibration process by using a mixed reality image and an optical tracker, achieving high accuracy without user-dependent procedures.

Figure 19.

Optical navigation technology

(A) Optical navigation system for osteotomy surgery. Reproduced with permission from ref.¹⁶⁷. Copyright 2004 CARS and Elsevier.

(B) Optical navigation system reference system for total knee replacement surgery. Reproduced with permission from ref.¹⁶⁹. Copyright 2014 Elsevier.

(C) Femur resection navigation system. Reproduced with permission from ref.¹⁷⁰. Copyright 2019 Elsevier.

(D) Total knee replacement navigation system. Reproduced with permission from ref.¹⁶⁸. Copyright 2022 The Authors. Licensed under the CC-BY-NC-ND license.

(E) Augmented reality surgical navigation system. Reproduced with permission from ref.¹⁷¹. Copyright 2020 Asia Pacific Knee, Arthroscopy and Sports Medicine Society.

(F) Active navigation optical tracking system. Reproduced with permission from ref.¹⁷². Copyright 2023 Elsevier.

Schlatterer et al.¹⁶⁹ modeled the various phases of the acquisition process by constructing multiple optical navigation system models which is shown in Figure 19B. The study involved 30,000 simulations utilizing Monte Carlo statistical methods to assess variations in anatomical reference frames. The investigation suggests that dynamic changes in the infrared optical navigation system have no significant impact on the alignment angle of fractured bones.

Wang et al.¹⁷³ propose a stereo tracking approach that utilizes machine learning techniques. This technique can precisely identify a specific item within a complex setting and is widely used in computer-assisted navigation systems. The researchers employed this technique to measure the distance between two specific markers. The experimental findings showed that the maximum average error in the positioning accuracy of this method was 0.32 mm. Matassi et al.¹⁷⁰ evaluated an accelerometer-based navigation system for total knee replacement, demonstrating that the technology can achieve precise surgical alignment and determine the ideal implant position which is shown in Figure 19C. Takagi et al.¹⁷¹ used the navigation method to refine computer-aided bone space design and improve bone space technology which is shown in Figure 19E. Hopfgartner et al.¹⁷⁴ introduced an innovative structured light imaging system to confirm and adjust the position and trajectory of the Kirchner needle in orthopedic surgery. Wan et al.¹⁷⁵ introduced an innovative static virtual-to-real calibration approach, which uses a camera coordinate system to map the relationship between the virtual environment and physical space which is shown in Figure 19D. The experiment confirms the practicality of the suggested calibration procedure, with an average registration accuracy of approximately 5.80 mm. Han et al.¹⁷² introduced an online optimization approach to mitigate information loss in optical navigation systems which is shown in Figure 19F. The findings suggest that the optical navigation system can achieve accurate measurements using calibrated equipment in all environments. Despite the optical navigation system’s high precision and stability, it is crucial to avoid any obstruction of the infrared camera and tracking target during operation. The occlusion issue presents a significant limitation, hindering the practical implementation of optical navigation in clinical surgery.

Electromagnetic systems are employed in computer-aided navigation systems to address line-of-sight obstruction issues in optical navigation systems. Electromagnetic navigation technology uses electromagnetic induction to track the endpoints of surgical instruments within the surgical field, facilitating the determination of the spatial coordinates of magnetic markers.¹⁷⁶ Lionberger et al.¹⁷⁷ conducted 46 total knee replacement procedures to assess the accuracy of an optical navigation system versus an electromagnetic navigation system. The test findings suggest that the electromagnetic navigation system approaches the efficacy of the optical navigation system and may substitute it in orthopedic surgery. Tigani et al.¹⁷⁸ examined the postoperative results of 32 total knee replacements performed with an electromagnetic navigation system. 30 out of 32 cases achieved optimal alignment, according to the comparison results. The data indicated that the electromagnetic navigation system ensures the safety of orthopedic surgery. Bouchard et al.¹⁷⁹ conducted an experimental evaluation of an innovative surgical navigation system, which includes an electronic tracking device used for bone marking during surgery. The test findings indicate that the navigation system has an overall average error of 1.55 ± 0.72 mm, meeting the accuracy standards for clinical surgery. Lambert et al.¹⁸⁰ evaluated the practicality of using an electromagnetic tracking device for treating internal aneurysms which is shown in Figure 20A. The test findings suggest that the electromagnetic navigation system has an average inaccuracy of 1.30 mm, which effectively reduces the impact of X-rays and the need for contrast agent injection. Stathopoulos et al.¹⁸¹ introduced a new radiation-free targeting method using electromagnetic navigation technology to accurately position intramedullary nails at a distance which is shown in Figure 20B. Berger et al.¹⁸² conducted a simulation of orthognathic surgery on a plastic skull to evaluate the precision and reliability of a novel electromagnetic navigation system which is shown in Figure 20C. Research data indicate that the electromagnetic navigation system demonstrates superior precision and accuracy compared to traditional surgery. Gao et al.76 developed a new electromagnetic navigation system and introduced a registration technique for external markers on intramedullary nails, along with an identification algorithm for guiding mechanisms. Both a modeling experiment and a clinical test confirm the performance and stability of the system which is shown in Figure 20D. The electromagnetic navigation technique overcomes light blockage issues and enables navigation of various surgical devices, such as puncture needles—a capability lacking in optical navigation systems. However, a drawback is that metal instruments experience significant interference from the electromagnetic field and are susceptible to disruption by nearby magnetic fields. Ultrasonic navigation systems for surgical operations are becoming an increasingly popular research topic, alongside optical and electromagnetic navigation systems.

Figure 20.

Electromagnetic navigation technology

(A) Principle of electromagnetic tracking system for aneurysm repair. Reproduced with permission from ref.¹⁷⁶. Copyright 2012 European Society for Vascular Surgery.

(B) Electromagnetic computer-aided guidance systems. Reproduced with permission from ref.¹⁸¹. Copyright 2012 Elsevier.

(C) Electromagnetic navigation system for orthognathic surgery. Reproduced with permission from ref.¹⁸². Copyright 2015 European Association for Cranio-Maxillo-Facial Surgery.

(D) New dynamic electromagnetic tracking navigation system. Reproduced with permission from ref.⁷⁵. Copyright 2021 Elsevier.

Ultrasonic navigation technology can perform information collection, modeling, and placement for patients based on the principle of ultrasonic distance measurement. Brendel et al.¹⁸³ introduced an approach to align CT and ultrasound datasets focusing on bone structure. The program acquires the dataset through the ultrasonic navigation system and handles the registration of CT data, ultrasound data, and pre-processed data throughout the operation. This approach was validated through tests on lumbar spines and human tissues, demonstrating its effective use. Matsuda et al.¹⁸⁴ used ultrasound technology to assess the precision of total knee replacements. Chopra et al.¹⁸⁵ combined ultrasonic imaging technology with optical navigation technology to propose a navigation system for tumor removal. The ultrasound probe uses images acquired via optical navigation for direct guidance, and ultrasonic navigation is employed within the tumor as an indirect navigation tool. Clinical measurements showed that both approaches significantly improved resection accuracy compared to conventional surgery. Beek et al.¹⁸⁶ introduced a new method for percutaneous stabilization of scaphoid fractures using an optical navigation device and ultrasound imaging to strategically place screws during surgery. An experiment in the laboratory compared the efficacy of ultrasound-guided surgery to traditional surgical methods. The results indicated that the new method met the accuracy standards for percutaneous scaphoid fixation and was superior to the previous method.

Wassilew et al.¹⁸⁷ conducted a comparative study on the clinical outcomes of ultrasound navigation systems versus image-free navigation systems in total knee replacement surgery which is shown in Figure 21A. Ultrasonic navigation systems achieve greater surgical precision and minimize outliers compared to image-free navigation systems. Rosenberg et al.¹⁸⁸ outlined an imaging separation strategy utilizing ultrasound scanning and presented an enhanced method for detecting drilling through intraosseous ultrasound measurements. Zhao et al.¹⁸⁹ employed ultrasonic guidance technology in percutaneous vertebroplasty, confirming its ability to significantly decrease radiation exposure and puncture duration which is shown in Figure 21C. Gueziri et al.¹⁹⁰ introduced a novel method for the systematic assessment of ultrasound registration, classifying the surgical procedure steps based on a new classification system. The classification process involves preprocessing, initialization, registration, and visualization. Hacihaliloglu et al.¹⁹¹ developed an innovative computational method to enhance the utilization of real-time 3D ultrasound for spinal imaging, improving modeling, segmentation, and registration tools. Tang et al.¹⁹² employed convolutional neural networks to perform 3D reconstruction of the spine surface in untracked ultrasound data, marking the first instance of this technique used for this purpose. Miyatake et al.¹⁹³ introduced an innovative navigation system to enhance surgical precision and safety which is shown in Figure 21B. Chen et al.¹⁹⁴ suggested using an annotation-guided encoder-decoder network to identify bone structure in radiation-free ultrasound data which is shown in Figure 21D. Ultrasonic navigation technology has advanced high-precision and high-safety in orthopedic surgery by combining optical and electromagnetic navigation systems. An ultrasonic surgical navigation system offers the benefit of avoiding trauma and radiation to the patient. However, it may face challenges such as picture noise, low quality, and a heavy sound shadow caused by bones, which can hinder correct positioning.

Figure 21.

Ultrasound navigation technology

(A) Ultrasound navigation system for total hip replacement. Reproduced with permission from ref.¹⁸⁷. Copyright 2012 Elsevier.

(B) Pre-operative ultrasound imaging. Reproduced with permission from ref.¹⁹³. Copyright 2023 The Authors. Licensed under the CC-BY-NC-ND license.

(C) Ultrasound navigation system image fusion. Reproduced with permission from ref.¹⁸⁹. Copyright 2018 Elsevier.

(D) Ultrasound-assisted orthopedic surgery navigation technology. Reproduced with permission from ref.¹⁹⁴. Copyright 2022 Elsevier.

(E) Convolutional neural network-based ultrasound navigation technology. Reproduced with permission from ref.¹⁹². Copyright 2021 Elsevier.

Computer-assisted navigation technology in orthopedic surgery currently offers high precision and accuracy. However, further research is needed to achieve greater precision and diversity in orthopedic procedures. Numerous unresolved issues with navigation technology in orthopedic surgery require immediate attention. The optical navigation system is hindered by line-of-sight obstructions; the electromagnetic navigation system requires a high-quality surgical environment; and the ultrasonic navigation system struggles with precise surgical positioning. Integrating multiple navigation technologies could provide a viable solution to these issues, enhancing precision in orthopedic surgery. However, coordinating various navigation systems may introduce issues such as increased costs, occupancy of surgical areas, and challenges in real-time translation. Cost-effective, compact, and efficient navigation systems are likely to drive future advancements in orthopedic surgery.

Application of VR technology in CARS

VR technology, an emerging field, has evolved alongside advancements in modern science and technology. In orthopedic surgery, its use can introduce fresh approaches and techniques for intricate procedures. This technology can minimize surgical injuries and provide essential guidance to physicians during procedures.^195,196 During traditional surgery, surgeons use the patient’s imaging data to plan procedures. Physicians use VR to construct a model of bone deterioration and strategize procedures based on this model. VR enables physicians to determine optimal surgical strategies, including surgical site, screw placement, and internal fixation dimensions. This technology can simulate most orthopedic processes, potentially enhancing the success rate of these procedures.¹⁹⁷ The significant benefits of VR in orthopedic surgery consistently attract researchers.

In 1996, Robb et al.¹⁹⁸ developed a virtual reality-aided surgical system. This technology provides real-time surgical computer support, enabling physicians to control and modify scan datasets instantly. The system can generate and transmit virtual images on the surgeon’s command without disrupting routine operations. Tsai et al.¹⁹⁹ developed an interactive VR simulator for orthopedic surgery. This simulator allows physicians to perform virtual operations, replicating intricate surgical procedures. Comparisons between simulators and actual simulations have shown that simulators are beneficial learning tools for orthopedic surgery planning. Expanding on prior research. Tsai et al.²⁰⁰ developed a haptic system for an orthopedic surgery simulator to replicate bone drilling procedures. Its haptic feedback is highly realistic, effectively improving hand-eye coordination during training. Rieger et al.²⁰¹ assessed the impact of CT virtual preoperative planning on fracture reduction surgery. Evaluation results indicate that CT VR technology is now a crucial supplementary tool for preoperative and intraoperative fracture reduction.

As research on VR surgery technology progresses, the perspectives on its use in medicine become increasingly crucial. As is shown in Figure 22A, Verhey et al.²⁰² reviewed the role of virtual, augmented, and mixed reality in orthopedic surgery, highlighting applications in surgical training, preoperative planning, and intraoperative navigation. These technologies demonstrated improved surgical precision, reduced radiation exposure, and enhanced trainee education outcomes. Cevallos et al.²⁰³ examined the efficacy of VR training for pediatric orthopedic procedures, specifically pinning slipped capital femoral epiphysis which is shown in Figure 22B. VR training reduced pin placement errors and fluoroscopy usage while enhancing spatial awareness and procedural accuracy compared to standard methods. Vankipuram et al.²⁰⁴ designed a virtual orthopedic drilling simulator, providing a lifelike surgical training environment which is shown in Figure 22C. Shi et al.²⁰⁵ evaluated a visuohaptic surgical training simulator for lumbar pedicle screw placement which is shown in Figure 22D. Residents trained with the simulator exhibited significantly lower screw penetration rates and greater accuracy than those taught using traditional methods. Mabrey et al.²⁰⁶ conducted a literature review of VR techniques for orthopedic surgery and compared virtual surgery simulators for orthopedic surgery with virtual techniques for surgical teaching. The summary revealed a scarcity of research on VR technology’s application in orthopedic surgery, despite its effective use in other sectors. Integrating VR with orthopedic training can efficiently reduce surgical time, minimize errors, and enhance doctors' training experience. Physicians are increasingly inclined to train using virtual machines. Gai et al.²⁰⁷ developed a VR system for training in fracture reduction. The system captures skeletal attitude information using an inertial measurement unit and creates a 3D human skeleton model. Skeletal position and function can be displayed in real-time on the Visual Studio platform. In contrast to costly optical instruments, this technology provides cost-effective data and a real-time interactive platform for fracture reduction simulation. Blumstein et al.²⁰⁸ divided medical students with no prior clinical surgery experience into two groups and provided them with standard guidelines and VR training. The results showed that medical students trained in VR demonstrated higher proficiency in simulated surgery compared to those with standard instruction, proving that VR technology is more effective for clinical training. Racy et al.²⁰⁹ proposed a VR femur nail simulator that combines tactile and image enhancement. The simulator consists of a 3D virtual environment, tactile devices, a 3D-printed drill shank, and a VR headset. The virtual environment is created by a video game development engine and enables haptic feedback and image enhancement through plug-ins. To verify the performance of the proposed simulator, participants used it to complete the near-end guide wire entry and remote locking. The elapsed time and tool travel distance were used to evaluate the effectiveness of the simulator. The results show that the simulator demonstrated good authenticity, content, and structural effectiveness. Huber et al.²¹⁰ developed a force-feedback VR simulator for Kirschner needle internal fixation in orthopedic surgery. To evaluate the simulator training, the authors divided 20 participants into two groups: one trained by an experienced surgeon and the other using a VR simulator. The results showed that participants who used the VR simulator performed better on tests of surgical skills than those trained by the surgeon. Vankipuram et al.²¹¹ introduced a virtual orthopedic drilling simulator that was designed to provide visiohaptic interaction with virtual bones. The method demonstrates a rapid learning rate; subjects are expected to achieve expert proficiency after multiple experiments. Cai et al.²¹² proposed a minimally invasive surgery simulator based on VR digital twins and verified the training effect of the simulator on surgical skills through experiments. The authors quantitatively assessed the effectiveness of simulator training by collecting the surgical proficiency data of specialists and novice physicians. The results of the analysis show that the simulator plays a positive role in training doctors to operate robots.

Figure 22.

VR technology for orthopedic surgery

(A) The integrated cameras of the VR headset enable “hand tracking”. Reproduced with permission from ref.²⁰². Copyright 2019 John Wiley & Sons, Ltd.

(B) VR surgery trial. Reproduced with permission from ref.²⁰⁶. Copyright 2022 The Authors. Licensed under the CC BY4.0 license.

(C) Virtual orthopedic drilling simulator. Reproduced with permission from ref.²⁰³. Copyright 2022 Elsevier.

(D) Virtual surgery training system based on force feedback. Reproduced with permission from ref.²⁰⁵. Copyright 2017 Elsevier.

Over the past nearly thirty years, an increasing array of VR technologies has been proposed and utilized in orthopedic surgery. VR technology in orthopedic surgery can be categorized into fracture reduction, bone drilling, prosthetic implant, and surgical training technologies, based on their applications. Although many VR systems remain experimental, researchers have praised their advancement in enhancing orthopedic surgery and surgeon capabilities.

Learning curve of CARS

The progress and improvement of various robotic and computer-assisted orthopedic methods have accelerated the learning process in modern orthopedic surgery research.

The learning curve concept, proposed in robot-assisted orthopedic surgery, focuses on defining and facilitating the progress of orthopedic surgical procedures. Currently, the learning curve not only includes the time required to complete a task but also encompasses various medical and scientific explanations of contextual elements. Figure 23 shows the relationship between physicians and the learning curve. Learning curve indicators for robot-assisted orthopedic surgery primarily focus on the selection and utilization of surgical tools, procedural efficiency, surgical accuracy and quality, and the enhancement of postoperative patient recovery and quality of life. Contemporary studies on the learning curve of robotic and computer-assisted orthopedic surgery primarily focus on total knee replacement and spinal surgeries.

Figure 23.

Relationship between surgeons and the learning curve

Hu et al.²¹³ examined 174 spinal procedures divided into five groups, each consisting of approximately 30 patients, to study the association between screw placement success and the learning curve. The surgeries were conducted with robotic assistance, and outcomes from the five groups were compared to assess their impact on the learning curve. With increasing surgical expertise, the success rate of screw insertion improved, and the frequency of switching between robotic and manual screw placement decreased. Over time, the rate of screw dislocations became more consistent. In 2017, Redmond et al.²¹⁴ analyzed the learning curve of robot-assisted hip replacement surgery, focusing on surgical precision, duration, and intraoperative factors. The study validated that adopting new technologies involves a learning curve, and surgical risk diminishes with increasing surgeon experience. In 2019, Siddiqui et al.²¹⁵ examined 120 instances of robot-assisted orthopedic surgery to analyze the learning process for proficient surgeons and neurosurgeons utilizing robot-assisted technology for pedicle screw placement. The study categorized assessment criteria into patient variables, intraoperative variables, and postoperative variables. Patient variables primarily included the patient’s weight, age, and gender. Intraoperative factors included screw placement accuracy, number of screws used, patient blood loss, and operation duration. Postoperative variables included length of hospital stay, discharge status, hospital readmission, and wound complications. The study demonstrated that robot-assisted spinal surgery systems, when paired with real-time navigation technology, can accurately perform screw placement. There was no discernible difference in surgical outcomes between seasoned surgeons and interns. Robot-assisted technology can effectively reduce the learning curve for surgeons. In 2020, Fayed et al.²¹⁶ examined clinical data from 100 robot-assisted pedicle screw implants to investigate the precision and learning curve of robotic screw placement. According to the data, path planning time, screw insertion time, and surgery duration decreased dramatically as the number of patients increased. As the number of procedures increased, the learning curve plateaued. In 2022, Su et al.²¹⁷ investigated the learning curve of robot-assisted spinal surgery, focusing on the precision of pedicle screw placement and operative duration. The study analyzed instances with unique time series to explore potential inherent relationships between teamwork, individual surgeons, and the learning curve. The results suggest that the learning curve for inexperienced surgeons working in teams parallels that of experienced surgeons. Effective communication between surgeons and supportive hospital culture are crucial for promoting teamwork. Stegelmann et al.²¹⁸ studied how the learning curve affects the likelihood of complications in robot-assisted total knee replacement procedures performed by novice orthopedic surgeons. After performing 50 procedures, novice orthopedic surgeons significantly reduced their average surgical time without encountering difficulties during the initial learning phase.

Mastering the learning curve for robot-assisted orthopedic surgery is crucial for orthopedic physicians. Recent research indicates that a combination of diverse assistive technologies, surgical techniques, and the surgeon’s expertise impacts the outcomes of learning curve studies. However, analogous research shows that the learning curve exhibits a consistent pattern. In the initial learning phase, the learning curve is hindered by factors such as delays to surgery, discharge times, and postoperative complications. However, the learning curve plateaus after a specific number of procedures. Currently, doctors' surgical skills have consistently improved. Surgeons can effectively complete the initial learning phase by thoroughly understanding the surgical indications. Evaluating the number of surgeries with various accurate assessment indicators during the learning phase allows for effective monitoring of surgical progress through the intermediate and advanced stages of the learning curve. Surgical complications and the postoperative quality of life of patients are key factors in determining the accuracy of the learning curve. By scientifically and objectively understanding the principles of the learning curve, clinicians can fully realize the clinical value and relevance of robot-assisted orthopedic surgery.

Technical ethics of CARS

Medicine plays a crucial role in providing ethical oversight in the realm of science and technology.²¹⁹ During operations, orthopedic surgical robots come into contact with patients' bodies, a crucial aspect for patient safety. Therefore, it is essential to ensure the safety, stability, and accuracy of these robots during their development. As the technology advances, researchers have examined the ethical concerns surrounding orthopedic surgical robots. From an ethical standpoint, researchers conducted a thorough analysis of the various types and origins of ethical hazards associated with orthopedic surgical robots.²²⁰ They also explored methods for preventing and managing these ethical risks. Studying the ethical concerns associated with orthopedic surgical robots could enhance their advancement and utilization in clinical settings.²²¹ Figure 24 illustrates the various characteristics and interests of stakeholders in CARS.

Figure 24.

The different characteristics and interests of stakeholders in CARS

Orthopedic surgical robots present ethical issues for robot designers, producers, consumers, medical professionals, medical institutions, and government regulatory authorities.^222,223 Addressing and managing ethical issues related to orthopedic surgical robots is a challenging endeavor that requires long-term, multi-agent solutions to foster strong cooperation among all stakeholders.²²⁴ Stakeholders of orthopedic surgical robots are categorized and organized based on their unique traits and interests.²²⁵ These categorized relationships are illustrated in a diagram. Effective management and mitigation of ethical risks can be achieved by coordinating stakeholders to establish a dynamic and cooperative relationship.²²⁶ Table 3 outlines the different characteristics and interest demands of stakeholders in CARS.

Table 3.

Different characteristics and interest demands of stakeholders of CARS

Name	Character	Function	Orientation
Designer	Provider	The robot’s use intention and manufacturing quality are determined by it.	Ethical ideals are integrated at the forefront.
Producer
Medical worker	Operation and application	Enhance intricacy using medical robots for precise and repetitive medical tasks.	Ethical obligation toward primary stakeholders
Medical institution
Patient	User	The party at risk of ethical implications is relatively vulnerable, however it possesses the right to decline the selection of the medical robot.	The most critical stakeholders
Government	Supervise	Key stakeholders to mitigate ethical hazards	Oversee and synchronize the interests of all parties.

Research challenges, technical difficulties and future work

Extensive research has been conducted on robot-assisted technology for fracture reduction. However, several issues and constraints must be addressed before this technology can be used in clinical surgery. Current research primarily examines the structural design and functional integration of orthopedic surgical robots to assess this technology’s reliability at various fracture sites in terms of precision and safety. Key areas for additional research have been compiled based on the current literature, as presented in Table 4. Orthopedic surgical robots are advancing toward image enhancement, structural downsizing, intelligent human-computer interaction, minimally invasive surgery, and remote operation. They are characterized by modularity, intelligence, lightweight design, and high integration.

• (1)Modularity: Currently, modularity is a key focus in the development of advanced equipment and a defining feature in the trend of orthopedic surgical robot development. It involves designing distinct functional modules that independently carry out the fracture reduction process, which are then integrated through interfaces. The development of modularity can reduce costs and production time for CARS, enhancing their versatility.
• (2)Intelligence: Advanced sensing technology improves the accuracy of CARS in clinical surgery and aids surgeons' decision-making. Orthopedic surgical robots use modern sensor and computer technologies to enhance sensitivity to environmental changes, enabling more precise clinical decision-making.
• (3)Lightweight Design: Conventional studies of orthopedic surgical robots prioritize surgical functionality and often overlook the impact of robot dimensions and mass on the surgical environment. Researchers are increasingly focused on developing lightweight orthopedic surgical robots that provide high-precision operations, occupy minimal space, and offer variable movement capabilities.
• (4)Integration: As high-end equipment processing technology advances, orthopedic surgical robot systems are increasingly using smaller components. Integrated technology simplifies system design and expands the capabilities of surgical robots.

Table 4.

Focuses of orthopedic surgical robots studied and to be studied

Research has been carried out at this stage		Future research trends
Keywords	Function realization	Keywords	Function realization
Structure	The researchers designed the structure for different fracture sites.	Precise, flexible, and compact design	Adjust according to the clinical setting
Precision	The precision of screw implantation or joint replacement during robotic reduction is improved.	Intelligent registration technology	The infographic image is fused with the fracture anatomical structure image to provide the doctor with accurate diagnostic information and spatial information
Learning curve	The surgeon successfully mastered the equipment in a short time.	Man-machine interaction	According to the operation needs can be convenient and fast operating system, improve the efficiency of surgery
Airmanship	Intraoperative guidance of the device can be more accurate.	Clinical sensing technology	Miniaturized, multi-information fusion sensing technology suitable for the surgical environment
Remote surgery	Provide telemedicine services.	Remote surgical control technology	The self-adaptability and collision avoidance of robot control system should be higher.
VR technique	The simulated operation process can provide reference for real clinical operation.	Operation specification	The operation was optimized to obtain the best surgical treatment standard.
Control technique	Control the robotic arm or other actuator to reach the specified position and complete the specified function.	Comprehensive treatment solutions	It integrates with artificial intelligence and other technologies to research new treatment methods and means

Conclusions

This study seeks to present the most recent advancements in robot-assisted orthopedic surgery, including topics such as development history, system composition, operating mechanism, and design of the CARS. Detailed information is provided on the fundamental principles, design, and operational effectiveness of several CARS. Detailed information is provided on the computer-aided navigation technology and the learning curve for the CARS. The text highlights the limitations of current technology and discusses the problems and obstacles in developing CARS in the future, along with outlining future research tasks.

The fracture reduction robot can be categorized based on its structure as external fixed, series-connected, parallel, and series-parallel hybrid structures. External fixation is the initial method suggested for achieving fracture reduction using a robot structure. The design is straightforward, although it lacks precision. The tandem CARS offers increased flexibility, while the parallel CARS fulfills the force and moment criteria necessary for orthopedic clinical surgery. The series-parallel hybrid reset robot combines the advantages of series-connected and parallel reset robots to different extents. Furthermore, detailed information is provided on orthopedic surgical robots designed for various fracture locations. The text describes the application of computer-aided navigation technology and VR technology in fracture reduction surgery, focusing on their operating principles, implementation technology, and functions. It analyzed the learning curve of several surgical robots and described the learning curve of orthopedic surgical robots together with its contributing aspects.

Despite the demonstrated effectiveness of suggested orthopedic surgical robots, certain technological hurdles and issues are often overlooked during actual clinical procedures. Structural Lightweight: This involves simplifying structure, reducing size and weight, minimizing design and manufacturing costs, and lessening the financial burden on patients, while maintaining high precision and safety. High Intelligence: Equipped with advanced sensing and computing technologies, these robots can accurately detect environmental changes, support medical decision-making, or autonomously make surgical decisions. System Modularization: This allows for efficient updates of various module technologies, reducing equipment costs and manufacturing time, and enhancing system adaptability. Functional Integration: Involves using micro-components to make surgical robots lighter, simplifying system design and expanding application ranges. These four issues represent research challenges are not yet fully addressed or emphasized in existing literature. Future research on CARS should involve scholars from relevant fields.

Acknowledgments

This work was supported by the National Natural Science Foundation of China under Grants No. 52075301 and by the Natural Science Foundation innovation and development joint fund of Shandong Province under Grants No. ZR2022LZY020 and by The Key R&D Program of Shandong Province under Grant No. ZR2022LZY020. The asterisk indicates the corresponding author.

Author contributions

Z.Z.: Conceptualization, investigation, visualization, and writing – original draft. Y.C.: Visualization and validation. G.M.: Writing – review and editing. Y.G.: Validation and investigation. Q.Z.: Supervision, resources, writing – review and editing. J.B.: Writing – review and editing.

Declaration of interests

The author declares no competing interests.

Declaration of generative AI and AI-assisted technologies in the writing process

During the preparation of this work the authors used ChatGPT4o in order to improve language and readability. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.