Towards advanced motion control systems for micro-scale magnetic drillers inside viscoelastic media

Abstract

Microrobotic systems have the potential to reach regions of the body with significantly less invasiveness than conventional surgical tools. However, control over clinically relevant workspaces remains challenging due to limitations in magnetic field and robotic trajectory control. Here, we present a planar eight-solenoid magnetic field generator (MFG) that enables programmable velocity and heading of a 250 × 250 × 1250 µm³ microscale driller through continuous field reorientation within a workspace of 20 × 20 × 20 cm³. The microscale driller achieved velocities up to 155 µm/s in 0.6 weight percent agarose gel, which mimics the mechanical properties of brain tissue, and executed turns beyond 90° with angular errors below 3°. These results demonstrate a scalable platform for precise and remote control of magnetically embedded microdevices, and support future development of minimally invasive systems for applications such as targeted therapeutic delivery and microsurgery.

Introduction

Microrobotic technologies have emerged as promising tools for minimally invasive medicine because they offer the ability to reach anatomical locations through minimally invasive and nonlinear control methods¹. Controllable magnetic field-based actuation has gained particular attention and use due to the field’s permeability through tissue, relative safety, and precise real-time control². In recent years, helical magnetic microrobots have demonstrated locomotion in fluidic and biological environments and have shown potential for applications ranging from targeted therapeutic intervention to functional surgery³.

While significant progress has been made in the development of magnetic microrobots, state-of-the-art demonstrations have focused on operations within fluidic or lumen-based environments where locomotion can be achieved through swimming or flow-assisted transport⁴. However, many clinically relevant targets are located within different biological tissue in environments such as the gastrointestinal, nervous, and cardiovascular system, and remain inaccessible through these pathways. Applications such as localized therapeutic delivery, biopsy, targeted tumor intervention, and access to deep anatomical structures require tools capable of traversing these media controllably and accurately. In such environments, propulsion mechanisms must generate sufficient shear forces to overcome tissue stiffness without causing unnecessary damage^5,6. Helical microscale drillers, or “microdrillers,” are strong candidates for these tasks for their ability to convert rotational motion into forward propulsion, which enables untethered navigation through tissue-like or viscoelastic materials. Developing control systems capable of generating and directing this motion therefore represents an important step towards clinically relevant magnetic microrobotic navigation.

Despite this progress, navigation of microdrillers through viscoelastic tissue remains a significant challenge. Systems designed for precise control of a microrobot’s motion frequently operate in highly localized workspaces, while larger-scale systems may not provide the field configurations necessary for accurate steering of micron-scale devices^7,8. In addition to having a large workspace and high robotic degrees of freedom (DOF), such systems for medical procedures should be designed for reconfigurability and compatibility with imaging techniques such as radiography⁷. As a result, achieving controlled motion through tissue-like environments over clinically relevant distances remains an open problem.

This limitation becomes particularly important when considering clinical translation. While current high-control/low-volume systems are suitable for proof-of-concept demonstrations, practical medical applications often require navigation over substantially longer distances while preserving the same degree of sub-millimetric controllability and localization accuracy^3,4. Expanding the workspace without sacrificing control authority therefore remains a critical challenge for effectively translating these systems.

To address this challenge, we developed a planar eight-solenoid magnetic field generator (MFG) (Fig. 1a) designed to produce controlled rotating magnetic fields over a workspace the size of an average human skull (20 × 20 × 20 cm³). This configuration allows for unobstructed access to the workspace, and simplified integration with current clinical workflows (Fig. 1b). Combined with a dynamic trajectory-generation framework capable of real-time reorientation of the applied magnetic field (Fig. 1c), the system enables programmable 5 DOF positional and heading control for magnetic microrobots⁹.

Fig. 1. Planar MFG system used for testing.

As shown, a i)Pixelink M2-CYL camera, ii)solenoid array, and iii) agarose gel sample displayed, microdriller shown in inset, scale bar is1 mm (b) rendering of system integration with patient table in operating room,which allows for simple integration with other medical equipment, (c) rendering of microdriller trajectory (not to scale) through parenchymal tissue,while avoiding functional tissue, and (d) COMSOL Multiphysics simulation magnetic field magnitude graph, in Tesla, located in the 2-dimensional central plane away from solenoid array, and demonstrating minimum fields of 8 mT throughout the full plane

Here, we demonstrate two key capabilities required for future therapeutic navigation, controllable propulsion and programmable steering. Experiments performed in 0.6 weight percent agarose gel, used to replicate the mechanical properties of brain tissue, show that the microdrillers can maintain accurate heading control over multi-path trajectories (Fig. 2a, c) with predictable frequency-dependent velocities (Fig. 2b). Together, these results establish a foundation for future robotic systems capable of navigating complex biological environments.

Fig. 2. Velocity and heading results in 0.6% by weight agarose gel.

Testing demonstrated (a) microdriller final heading versus planned trajectory, shown in white, with a maximum error of 3° and an average error of 1.5°, scale bar is 1 mm (b) microdriller velocities versus rotational frequencies, which show a strong linear relationship to the applied frequency below step-out, and (c) a 90° turn demonstrated by a driller operating at 5 Hz rotational frequency, which utilized a multi-stage turn in 10° increments, scale bar is 1 mm

System design and motion planning

The presented actuation platform consists of eight solenoids arranged in a planar geometry facing a central workspace (Fig. 1a). Each solenoid contains a soft magnetic core (Vacoflux 50) and is independently driven to generate magnetic fields with programmable magnitude and orientation. Field measurements performed with a custom PCB hosting 25 MLX90395 magnetometers, and verified with an Extech Instruments SDL900, confirmed sufficient workspace magnitudes exceeding the minimum requirement of 5 mT for locomotion throughout the functional volume. Figure 1d shows the magnetic field magnitudes in a 20 × 20 cm² center plane from the solenoids.

The microdriller used in this study consists of a two-photon polymerization (2PP) printed helical body bonded to a permanent magnetic base (Fig. 1a). When inside a rotating magnetic field, the helical geometry converts the rotational motion from the magnetic torque into forward thrust (Fig. 2b).

To exploit this behavior, trajectories were represented as a series of user-defined heading paths. For each path, a corresponding rotating field was generated in the plane perpendicular to the desired direction of travel. Current values for each solenoid were computed using constrained minimum optimization and delivered through an electronic control architecture. By continuously updating field orientation between trajectory segments, the system produced chosen paths while maintaining the field magnitudes necessary for propulsion (Fig. 2c).

Rather than focusing solely on field generation, the control framework was designed around navigation as a complete task. By automatically converting desired trajectories into magnetic field commands, the system enables a more intuitive pathway from clinical intent to medical intervention while reducing the complexity of hardware operation. This approach enables the user to define a desired path, for which the control system will automatically generate a sequence of magnetic field orientations for execution. The full platform therefore combines a large-workspace magnetic actuation system with programmable trajectory planning and execution to deliver complete functional control.

Velocity and heading performance

A primary requirement for practical microscale navigation is the ability to control propulsion velocity. To characterize this capability, microdrillers were tested in 0.6 weight percent agarose gel using rotational frequencies from 1 to 5 Hz. Higher frequencies were not investigated for this paper, as excess heating caused inconsistencies in the mechanical properties of the agarose gel samples. Across this range, velocity increased approximately linearly with rotational frequency, reaching a maximum speed of 155 µm/s at 5 Hz (Fig. 2b). This behavior is consistent with prior experimentation in our group, which showed that translational speed scales with rotational frequency¹⁰.

This experiment shows that velocity can be programmed directly through frequency selection, enabling a simple method for adjusting speed and positioning accuracy without modifying the microdriller or control hardware. The relatively small variation between repeated measurements further suggests stable operation throughout the tested frequency range. Although step-out was not investigated, future studies will progress through higher frequencies to determine the operating limits of the system.

Equally important is the ability to navigate along nonlinear paths. To evaluate heading capabilities, trajectories for headings from 0° to 90° were generated in 10° increments through sequential field orientations (Fig. 2a). The microdriller successfully followed these programmed trajectories through the full turn, with a maximum measured heading error of 3° and an average error of 1.5°.

Precise heading control is essential as small errors in orientation can accumulate over longer headings, resulting in significant deviation from the intended target. This challenge becomes increasingly important in neurosurgical and other tissue-navigation applications where surrounding structures may experience severe damage from unintended contact. Accurate heading control not only improves navigation performance but also contributes to procedural safety by reducing the likelihood of hemorrhage or other adverse outcomes. The low angular errors observed in these experiments represent an important step towards reliable navigation in clinically relevant environments.

The demonstrated turning capabilities additionally highlight an important advantage of magnetic control over other methods. Because steering is achieved through reorientation of the magnetic field rather than onboard actuation, directional control may be implemented regardless of device size, provided that the minimum field magnitude is reached. This is particularly beneficial for devices operating in confined environments, for which minimum dimensions are a critical factor.

Conclusion

This work demonstrates a novel planar eight-solenoid magnetic field generation system capable of controlling micron-scale helical robots in viscoelastic media. By combining a large-workspace MFG with a trajectory-based control framework, the system enables programmable propulsion and accurate steering in viscoelastic media. The microdriller achieved velocities up to 155 µm/s and executed heading changes greater than 90° with low angular error. These results establish a path towards clinically relevant navigable microrobotics in tissue-like material.

Future work will extend these tests to biological tissue, investigate dimensional trajectory generation and control and explore functional mechanisms on the microdrillers for various medical operations. Ultimately, these systems aim to support minimally invasive interventional operations by opening access to inoperable regions deep in the body and neural tissue for targeted therapeutic operations.

Author contributions

This report was compiled by E. Z. Freeman, with contributions to figures from T. H. Wang, P. Chatterjee, and A. Ansari. K. Hoang provided medical advisement and assistance with custom artwork from Emory University. All authors took part in editing and finalizing the manuscript.

Competing interests

The authors declare no competing interests.