3D‐Printed Metamaterial‐Based Soft Sensors: Materials, Design, and Fabrications

ABSTRACT

Additive manufacturing has enabled highly sophisticated three‐dimensional soft bodies through material distribution, opening new possibilities for architected soft matter systems. In parallel, rapid development in soft robotics has intensified the necessity of compliant, distributed, and deformation‐driven sensing solutions. Among possible solutions, 3D‐printed metamaterial‐based soft sensors are promising due to unprecedented mechano‐sensing performances. Specifically, by integrating material design and structural topology into a unified functional entity, these metamaterial‐based sensing solutions enable tunable stiffness, controlled instability, and efficient electromechanical transduction. With these promising findings in mind, this review aims to provide a comprehensive framework addressing additive manufacturing technologies, material systems, and sensing mechanisms in 3D‐printed metamaterial‐based sensors. In this work, available 3D printing technologies are discussed, highlighting trade‐offs in resolution, multi‐material capability, and structural fidelity. In parallel, a variety of 3D printing materials, including polymer‐based, functional composite, and smart responsive materials, is examined, emphasizing material–structure interactions determining sensing performance. Subsequently, metamaterial‐based transduction mechanisms are classified into resistive, capacitive, and inductive modalities, together with emerging multifunctional and multimodal sensing modality. Conclusively, by synthesizing fabrication technologies, material systems, and sensing architectures within an additive manufacturing perspective, this review provides design frameworks and outlooks for perceptive soft machines and their applications in real‐world scenarios.

3D‐printed metamaterial‐based soft sensors exploit architected unit‐cell structures to translate mechanical deformation into electrical signals. By programming compressibility, stiffness, and negative Poisson's ratio through structural topology, these sensors achieve extraordinary mechanical performance and enhanced sensing capability. The integration of additive manufacturing, functional materials, and resistive, capacitive, and inductive transduction mechanisms establishes a unified framework for intelligent soft robotic systems.

1. Introduction

Over the last decade, soft robotics horizon has rapidly expanded, allowing for highly sophisticated and intelligent performances in real‐world scenarios [1, 2, 3]. Among the robotic tasks, cognitive perception plays a vital role in allowing the soft robot to exhibit desired performance while interacting with surroundings [4, 5]. More importantly, proprio‐ and exteroceptive sensors are necessary to enable the robot to understand better on the environment and make decisions properly, which results in better safe and compliant interaction with humans [6, 7, 8].

Recently, such efforts have been achieved by perusing multidisciplinary approaches, including material and computer science, and engineering aspects [9, 10]. Particularly, soft sensors made of constitutive hyperelastic materials have shown strong potential in achieving high compliance and distributed sensing systems [11, 12, 13]. Herein, it is worth mentioning that one of pivotal aspects to realize advanced sensing architectures relates with seamless integration of sensing and actuated systems [14, 15]. More specifically, given that the entire soft body usually undergoes large deformations, conventional sensing units (e.g., accelerometers, Inertial Measurement Unit (IMU) sensors, Camera vision, etc.) are likely to encumber the desired movement of the soft robots, resulting in poor mechanical performances, which is far being matched with the maturity that the rigid bodied robots have achieved so far. Therefore, it is noted that by taking inherent softness of the sensor at the material level into account, embodiment should be addressed.

From a material point of view, achieving both bidirectional deformation (including both extension and compression) and mechanical continuity remains open challenges [16]. Given that most constitutive materials are nearly incompressible, soft bodies generally need specific morphological and/or geometrical designs to accommodate compressive strain [17]. As regards the mechanical continuity of the soft body, the mechanical discontinuity of sensorized soft bodies has often been induced by non‐homogeneous distribution of the materials used, which leads many researchers to suffer from incomplete distributed systems [9]. To this end, metamaterials have been highlighted as a promising solution. Indeed, metamaterials (also called by auxetic materials) represents a truly man‐made material that can be barely observed in nature and that shows unique mechanical characteristics, such as negative Poisson's ratio, compressibility, etc [18, 19]. Among such extraordinary behaviors, compressibility of the metamaterial has overcome the bottlenecks that conventional sensors have, opening new paradigms in designing transduction mechanism based soft sensors [20, 21, 22].

In general, metamaterials refer to a class of architected material systems whose effective properties are governed by the topology and arrangement of their unit cells, rather than by intrinsic material and chemical compositions [23]. Such architectural design enables extraordinary behaviors that are rarely observed in nature, including tailored stiffness, programmable compliance, and shape morphing. Depending on their functionalities, metamaterials can be broadly classified into mechanical, electromagnetic, and acoustic families [24, 25]. In addition to passive sensing, metamaterials support co‐design with actuation and computation toward embodied intelligence, which opens new avenues toward intelligent and perceptive soft grippers [26].

From a mechanical point of view, the metamaterials are hierarchically composed of plural unit cells. Depending on their topologies and arrangements, it is possible to determine different dimensional deformation modes (e.g., Isotropic, Orthotropic, Anisotropic deformation, etc.) [27]. Herein, topological transformation refers to re‐distribution of the materials used with binary code, which can be applied to two dimensional (surface tessellations) and three dimensional (volumetric tessellations) soft bodies [28, 29, 30]. Due to these promising mechanical characteristics of the metamaterials, many pioneering studies have worked at developing novel soft robots and their applications, e.g., actuators [31, 32, 33]. sensors [21, 34, 35], grippers [30, 36, 37]. locomotive robots [38, 39, 40], bioinspired mechanisms [29, 41], etc. Given that stiffness and strain can be tuned at both global and local levels by employing different unit cell dimensions and/their topological characteristics, it is possible to realize programmed motions without kinematic junctions or support, which allows for achieving underactuated systems while ensuring energy efficiency.

In a sensing perspective, such programmed deformability (particularly compressibility) has opened a new possibility in developing transduction mechanisms based soft sensing technology. More specifically, physical deformations of soft bodies induced by internal and/or external forces can be transduced into electric signals, in forms of resistance [42, 43], capacitance [44, 45], and inductance [46, 47]. Unlike conventional sensing solutions usually created by rigid‐bodied counterparts, deformation induced sensing is achievable, allowing the soft bodies to exhibit perceptive mechanical performances [48, 49]. On the other hand, given the morphology and geometry of the metamaterial, conventional fabrication approaches (e.g., molding and casting) are unlikely to realize such sophisticated structures. To this end, metamaterial based robotic systems, particularly employing volumetric tessellation, have been mainly addressed by additive manufacturing [50, 51, 52].

With these promising potentials of metamaterial in mind, as shown in Figure 1, this comprehensive work aims to summarize metamaterial‐based transduction mechanisms, focusing on resistive, capacitive, and inductive sensing solutions and their materials and fabrications, while providing insights and outlook of perceptive soft robotic application scenarios. To this end, this review is organized from the 3D printing technologies which enable metamaterial fabrication, through the material systems that determine their functional performance, to the sensing mechanisms that define their sensing capabilities, and finally to the challenges and future perspectives toward perceptive soft robotic systems.

FIGURE 1.

Overview of 3D‐printed architected material‐based soft sensors.

2. 3D Printing Technologies

3D printing technology, a subgroup of Additive Manufacturing (AM), enables the fabrication of geometrically complex structures that are difficult or impossible to achieve using conventional subtractive machining (e.g., milling, turning, drilling) [53, 54, 55, 56] or mold‐based manufacturing processes (e.g., injection molding or casting [57, 58, 59, 60]. Compared to these traditional methods, 3D printing technology offers several key advantages, including the ability to produce intricate internal architectures [50, 61, 62, 63], fabricate multi‐material components [64, 65, 66], reduce material waste [67, 68, 69, 70], and avoids the need for geometry specific molds by building parts directly from digital designs [71, 72, 73]. These capabilities are particularly valuable for manufacturing micro‐architected and lattice‐based structures with high affordability in design and fabrication. In general, 3D printing technologies allow for rapid prototyping by depositing or solidifying material layer by layer [74, 75, 76, 77]. This layer‐wise material accumulation enables the realization of highly sophisticated geometries.

The geometric freedom enabled by AM is especially critical for the realization of metamaterials. Such architected materials, with engineered unit‐cell architectures, have been extensively investigated for advanced functional applications, including actuators, sensors, and adaptive mechanical systems [18, 78], because their mechanical and functional responses are predominantly determined by precisely engineered unit‐cell architectures rather than bulk material properties. The realization of sophisticated geometries and structures, spanning multiple length scales, has been constrained by conventional manufacturing processes. Moreover, these are inherently limited in producing topologically tessellated structures with high geometric resolution and hierarchical arrangement of their unit cells [79, 80]. On the other hand, AM can overcome such constraints by enabling direct fabrication of unit‐cell topologies through layer‐wise material deposition, facilitating multi‐material integration within a single build, and allowing programmable control of structural parameters across different length scales [50, 65, 81]. These capabilities make AM particularly suitable for producing architected materials with precisely defined mechanical, electromagnetic, and acoustic responses that are difficult or impossible to realize using conventional manufacturing routes [24, 27]. However, despite these advantages, AM still faces challenges in scalability, including trade‐offs between fabrication resolution, build volume, and production throughput, particularly when fabricating hierarchical or multi‐scale structures [61, 77]. Recent advances in computational design tools — including topology optimization, generative design algorithms, and simulation‐driven material architecture engineering — have provided systematic frameworks for predicting structure–property relationships in metamaterials prior to fabrication [82, 83]. When coupled with the geometric freedom of AM, these digital design approaches enable rapid prototyping, iterative testing, and performance‐driven optimization of architected metamaterials, thereby accelerating their development for a wide range of engineering applications [27, 80]. With these inherent advantages in mind, it is worth mentioning that 3D printing can not only create single‐mode sensors but also fabricate multimode sensors with fully 3D‐printing process [84, 85, 86] or hybrid 3D‐printing process [87, 88, 89]. Fully 3D‐printed sensors refer to sensing systems in which all functional components — including the structural body, sensing elements, and electrical interconnects — are fabricated directly through AM without requiring post‐fabrication assembly of externally manufactured parts. In such systems, both geometry and material distribution are realized in a single or sequential AM process additive manufacturing [62, 90]. In contrast, hybrid 3D‐printed sensors combine AM with complementary fabrication techniques to achieve functionalities that are difficult to realize through AM alone.

In literature, many studies have implemented hybrid manufacturing by integrating conventionally fabricated components — such as commercial sensing chips, conductive wires, thin‐film electrodes, or microelectronic elements — with additively manufactured structural frameworks. This integration often involves post‐printing assembly, embedding prefabricated components during the printing process (print‐pause‐insert strategies), surface coating or deposition of functional layers, or combining AM with processes such as casting, lithography, or laser micromachining [91, 92]. Depending on the fabrication strategy, sensors can be manufactured as standalone devices and subsequently installed into a host structure, or they could be directly integrated into the host material through in situ printing or embedded fabrication during the build process [93, 94, 95].

More specifically, given that metamaterial‐based sensors rely on conductive networks (that respond to external stimuli), it is noted that their sensing performance is strongly influenced by the physical and electromechanical properties of the constituent materials. By far, a wide range of material classes are employed in AM of resistive sensors, including polymers [53, 96, 97], ceramics [98, 99, 100], metals [76, 77, 101, 102], and multifunctional composites [64, 88, 103]. Depending on the specific printing process, these materials can be processed in different physical states, such as liquid precursors [96, 104, 105], viscous inks [64, 103, 106], or solid filaments [107, 108, 109]. The selection of material therefore plays a vital role in determining sensing characteristics, mechanical compliance, process compatibility, and overall device functionality. Representative examples of materials used in 3D‐printed metamaterial sensors include silicon rubber, polymer composite [110], nanocomposite [111], polylactic acid (PLA) [112], acrylonitrile butadiene styrene (ABS) [113], exfoliated graphite [114], and hydrogels [115]. To provide a systematic overview, AM approaches primarily used to fabricate metamaterial‐based sensors are discussed (Figure 2). In parallel, a detailed overview of the 3D printing material is presented in the following sections.

FIGURE 2.

Working principle of 3D printing technologies. (a) Vat Photo Polymerization (VPP) technique. (b) Material Extrusion (MEX). (c) Powder Bed Fusion (PBF). (d) Material Jetting Technology (MJT). (e) Binder Jetting Technology (BJT).

Among the various AM approaches summarized above, vat photopolymerization (VPP), as illustrated in Figure 2a, is one of the most widely adopted techniques for fabricating high‐resolution metamaterial‐based sensors. Due to the promising abilities that the VPP can provide (e.g., high dimensional accuracy, and surface accuracy, etc.), robotic applications in real‐world that require precise structural definition have been extensively studied, including biomedical and wearable robotics [116, 117, 118], mechanical and thermal sensors [119, 120, 121, 122], and gas sensor [123, 124, 125, 126, 127]. Based on the VPP principle, a liquid photopolymer resin is selectively solidified through light‐induced polymerization to form three‐dimensional structures in a layer‐by‐layer manner. The curing process is controlled by spatially patterned light exposure within a resin‐filled vat, enabling the fabrication of complex micro‐architected geometries with high feature resolution. In literature, stereo‐lithography (SLA) represents the earliest implementation of the VPP concept [128]. VPP technologies have evolved into multiple system configurations, primarily distinguished by light‐source arrangement and exposure strategy [79]. This laser‐based technique scans and cures each layer sequentially by tracing linear patterns across the resin surface [104]. From a mechanical configuration perspective, VPP systems can be categorized into bottom‐up and top‐down architectures. Regarding the former, the light source is positioned beneath the resin vat, and the build platform incrementally moves upward as each layer is cured. In contrast, the latter, top‐down systems, expose the resin from above, with the building platform progressively descending into the vat during fabrication [79]. From an optical projection perspective, several advanced exposure strategies have been developed. Digital Light Processing (DLP) employs a digital micromirror device (DMD) to project an entire patterned image onto the resin surface, enabling simultaneous curing of each layer and significantly improving fabrication speed compared with scanning‐based methods [129]. Another notable projection‐based technique is Continuous Liquid Interface Production (CLIP), which introduces an oxygen‐inhibited “dead zone” near the curing interface. This uncured layer prevents adhesion between the part and the vat window, enabling continuous, rather than layer wise, fabrication and thereby substantially increasing production efficiency [105].

Material extrusion (MEX) represents an AM process in which material is continuously deposited through a nozzle and solidifies to form successive layers of a three‐dimensional structure (Figure 2b) [130], including Fused Filament Fabrication (FFF), Fused Deposition Modeling (FDM) and Direct Ink Write (DIW) [131]. The flow of the material is driven by different propulsion mechanisms, including screw‐based [101, 132, 133], pneumatic, plunger‐based, or filament‐fed systems [64, 103, 134, 135]. The final mechanical performance of printed parts strongly depends on printing parameters, e.g., layer thickness, filament width and orientation, and interlayer bonding quality [136]. Variations in interbead spacing and void formation between adjacent filaments further influence structural integrity and functional properties. From a sensing perspective, given that the MEX allows for controlled deposition of conductive and elastomeric materials with programmed architectures, the MEX is particularly affordable are useful in fabricating metamaterials. Due to these, it is possible to realize sensorization that exploit conductive material‐based transduction mechanisms. By tailoring printing parameters, it is possible to tune conductive network topologies, anisotropic mechanical responses, and strain‐dependent electrical pathways within the printed structure [61, 103]. Indeed, many of lattice‐based sensing architectures have been introduced, including flexible strain sensors, and soft robotic components with integrated sensing functionality (which is in form of embodied intelligent soft robotic systems). However, the relatively coarse resolution compared with photopolymerization‐based techniques and the dependence of electrical performance on interlayer contact resistance remain open challenges, particularly for micro‐architected metamaterial systems requiring high structural fidelity [61].

Powder bed fusion (PBF) indicates an AM technique in which a focused energy source, such as a laser or electron beam, selectively fuses regions of a powder bed to produce dense three‐dimensional components [137]. Primary process variants include Selective Laser Sintering (SLS), Selective Laser Melting (SLM), and Electron Beam Melting (EBM) [138]. Given that the PBF directly consolidates material through thermal fusion, it is possible to fabricate high‐strength metallic or polymeric lattice structures with high mechanical performance and reliable geometric precision (Figure 2c). This capability makes the PBF particularly valuable for load‐bearing metamaterial sensors and structural sensing systems operating under large mechanical stresses [76]. Architected metallic lattices produced by the PBF can serve simultaneously as mechanical frameworks and conductive sensing networks, enabling distributed strain monitoring and structural health sensing. Furthermore, the high spatial resolution and ability to precisely control internal lattice geometry in PBF allow the fabrication of sophisticated periodic and non‐periodic unit cells, whose topology and relative density critically govern deformation‐induced electrical behavior [76]. However, thermal gradients during processing, residual stress accumulation, and limited material compatibility with embedded soft conductive phases present challenges for fabricating multifunctional hybrid sensing systems [77].

Material jetting (MJT) represents an AM technique in which discrete droplets of liquid material are selectively deposited onto a substrate to build a structure layer by layer (Figure 2d) [130]. Droplet generation can occur through either drop‐on‐demand or continuous jetting modes. In drop‐on‐demand systems, thermal, electrostatic, or piezoelectric actuation generates pressure pulses that eject individual droplets [139, 140, 141]. Continuous jetting systems instead maintain a steady pressure in the material reservoir to produce a constant liquid stream that is subsequently modulated or deflected. The printing quality in the MJT is primarily governed by fluid dynamic and interfacial phenomena, including droplet size, jet velocity, viscosity, surface tension, and nozzle‐to‐substrate distance [142, 143, 144, 145]. These parameters determine droplet spreading, coalescence behavior, and curing uniformity, which in turn influence resolution and material distribution [146]. For metamaterial‐based sensors, the MJT offers a significant advantage in multi‐material integration and spatially selective property control. The ability to deposit different functional distribution single build process enables fabrication of graded conductive networks, embedded electrodes, and complex heterogeneous architectures [64]. This capability is particularly useful in developing multimodal sensing systems that require localized variation in stiffness, conductivity, or responsiveness [65, 88, 89, 147]. Nevertheless, the MJT is restricted to a narrow viscosity range for stable droplet generation and is therefore primarily compatible with low‐viscosity liquid resins rather than conventional thermoplastic polymers or metallic feedstocks. High‐viscosity inks hinder jet breakup and increase clogging risk, while low‐viscosity inks cause satellite droplets and poor dimensional control. This rheological limitation complicates the printing of highly filled conductive composite materials, which are necessary for high‐performance sensors [64]. In addition, the printed parts, typically based on photopolymer resins, may exhibit limited mechanical robustness and long‐term stability compared with parts fabricated from structural thermoplastics.

Binder jetting (BJT) is an AM process in which a liquid binder is selectively deposited onto a powder bed to form the cross sectional geometry of a part (Figure 2e) [148, 149, 150]. The building platform lowers incrementally as new powder layers are spread and selectively bonded. Final component properties depend strongly on binder chemistry, particle morphology, powder packing density, wettability, and post‐processing treatments such as sintering or infiltration. In the perspective of metamaterial‐based sensing, binder jetting is particularly suitable for fabricating large‐scale lattice architectures and porous structures with complex internal geometries [151, 152, 153, 154]. Since the printing process does not rely on localized melting, the BJT enables fabrication of highly intricate unit‐cell topologies with minimal thermal distortion, which is advantageous for architected mechanical metamaterials designed to control deformation pathways [151]. Post‐processing infiltration with conductive materials can further introduce functional sensing capability within mechanically optimized structures. However, in binder jetting (BJT), the as‐printed parts (green bodies) exhibit limited mechanical strength and electrical conductivity prior to post‐processing. Furthermore, shrinkage occurring during sintering can reduce dimensional accuracy and geometric fidelity, leading to deviations in micro‐architectural features that are critical for achieving reliable and repeatable sensor calibration [148, 151, 155, 156].

In nutshell, the AM techniques discussed above provide complementary fabrication pathways for realizing metamaterial‐based sensors with tailored geometries, hierarchical architectures, and integrated sensing functionalities. Each processing strategy offers distinct advantages in terms of resolution, material compatibility, structural complexity, and scalability, thereby enabling the design of sensors with tunable mechanical and electromechanical responses. Nevertheless, apart from a proper choice of fabrication technology, the material system is likely to govern the performance of the metamaterial‐based sensors, including electrical conductivity, mechanical compliance, environmental stability, and sensing reliability. Previous studies have demonstrated that the interplay between printable materials and processing conditions critically determines the achievable microstructure, interfacial bonding, and long‐term durability of additively manufactured sensing devices [143, 157, 158]. Therefore, a comprehensive understanding of material selection and material–process compatibility plays a vital role in optimizing sensor performance and expanding application potential [103]. The following section addresses the material systems commonly employed in metamaterial‐based sensors, with particular emphasis on their functional roles, processing requirements, and structure–property relationships.

3. Materials for 3D‐Printed Metamaterial‐Based Sensors

A choice of material selection plays a vital role in determining the functionality of 3D‐printed metamaterial sensors. Unlike conventional sensing systems, effective properties of the metamaterial are governed by the coupling between intrinsic material behavior and engineered structural geometry, as illustrated in Figure 3. Consequently, mechanical and electrical characteristics of the material used (i.e., elasticity, dielectric response, electrical conductivity, and multi‐physical coupling) strongly influence resonance characteristics, sensitivity, and tunability. AM enables the integration of a variety of material classes — including polymers [159], functional composite [160], and smart responsive materials [161]—each offering advantages for sensing applications in electromagnetic, mechanical, and multi‐physics domains.

FIGURE 3.

Some examples of materials for 3D‐printed metamaterial‐based sensors. (a) Flexible, cubic building blocks for Voxelated Mechanical Metamaterials. Reproduced with permission [162] Copyright 2016, Springer Nature. (b) A TPU array unit cells of perforated structures. Reproduced with permission [163] Copyright 2022, Elsevier. (c) 3D‐ printed conducting polymer mesh in hydrogel state. Reproduced with permission [164] Copyright 2020, Springer Nature. (d) A graphene and graphene/polymer architectures printed by DLP processing method. Reproduced with permission [165] Copyright 2023, MDPI. (e) PLA‐based 3D‐printed herringbone tessellated tube. Reproduced with permission [166] Copyright 2021, Wiley. (f) Strain recovery under infrared radiation. Reproduced with permission [167] Copyright 2007, Elsevier. (g) A locally resonant acoustic metamaterial. Reproduced with permission [24] Copyright 2016, American Association for the Advancement of Science. (h) A programmable bipolar sheet—quasi‐2D elastic slab of material patterned. Reproduced with permission [168] Copyright 2014, American Physical Society.

3.1. Polymer‐Based Materials

Polymer‐based materials represent the most widely used material class in 3D‐printed metamaterial sensors due to their excellent printability, low processing temperature, and tunable mechanical properties [53, 61, 62, 96, 169]. Their compatibility with diverse AM techniques enables the fabrication of complex architected lattices with precisely controlled geometric features, which are critical for tailoring effective material properties and resonance behavior [159]. This category includes thermoplastics, photocurable resins and elastomers polymers, each offering distinct advantages depending on the targeted sensing mechanism and fabrication technique [89, 105, 164, 170, 171].

• Thermoplastic Polymers, including polylactic acid (PLA), acrylonitrile butadiene styrene (ABS), and thermoplastic polyurethane (TPU), are commonly processed using material extrusion techniques. These materials offer structural robustness and design flexibility, making them suitable for mechanical and structural metamaterial sensors [61, 172]. PLA and ABS have been widely employed for load‐bearing lattice architectures due to their relatively high stiffness and dimensional stability [171, 173], whereas TPU provides enhanced elasticity and strain tolerance, which are advantageous in flexible and wearable sensing platforms [174, 175]. Furthermore, MEX technique enables rapid prototyping of unit cell geometries with controllable infill density and anisotropic mechanical properties, which directly influence stress–strain behavior and piezoresistive or strain‐sensing performance [176, 177]. In addition, to improve mechanical performance, thermoplastics can be functionalized through metallic coatings to enable electromechanical coupling within elastic lattices [178, 179]. However, interlayer adhesion, printing‐induced anisotropy, and limited resolution may affect structural uniformity and sensing performance in metamaterials [180, 181, 182].

• Photopolymers, or Photocurable Resins, employed in VPP processes (i.e., SLA, DLP, and TPP). VPPs are particularly advantageous in producing micro‐ and nano‐structured metamaterials due to their high precision and smooth surface quality, while reducing scattering and structural defects in electromagnetic and optical sensing applications [183, 184]. The availability of functional photopolymers (i.e., high‐refractive‐index resins, elastomeric resins, and nanoparticle‐reinforced composites) broadens their applicability in sensing platforms that require optical transparency, dielectric tunability [185, 186], or mechanical compliance [187]. Nevertheless, photopolymers present intrinsic limitations such as polymerization shrinkage, residual stress accumulation, and limited fracture toughness, which may compromise structural durability under cyclic mechanical loading [188, 189]. Photo‐initiator residues and long‐term UV exposure can also degrade material stability, affecting sensing reliability [109, 190].

• Elastomeric Polymers further enable large reversible deformation and nonlinear mechanical response, making them suitable for strain, pressure, and tactile metamaterial sensors [160]. In this view, elastomers typically exhibit large strain with nonlinear behaviors that can be predicted by hyperelastic constitutive formulations. Hyperelastic materials have become one of the most important material classes in 3D‐printed metamaterials, particularly for sensing applications due to large deformation, structural reconfiguration, and tunable response. Compared to elastic materials, constitutive hyperelastic materials undergo extremely large reversible deformation. Their mechanical response is typically described using strain energy density models such as Neo‐Hookean, Mooney–Rivlin, or Ogden formulations [191]. Common hyperelastic materials used in AM include silicone elastomers, rubber‐like photopolymers, and soft polymer networks, etc. These materials are especially compatible with MEX and VPP, which enable the fabrication of highly compliant elastic lattice and multi‐material soft structures [160]. The importance of hyperelastic materials in metamaterial sensors arises from their strong coupling with structural geometry. Mechanical metamaterials derive many of their unusual properties—such as negative Poisson's ratio, programmable stiffness, controlled buckling, and snap‐through instability—from large deformation of structural elements (Figure 3a,b) [162, 163, 168, 192, 193, 194]. Hyperelastic constituents allow these deformation mechanisms to occur without structural failure, enabling reliable sensing upon cyclic loading and large strain conditions [195]. In sensing applications, hyperelastic metamaterials enable several key functionalities. For instance, large reversible deformation significantly enhances strain‐dependent property variation, improving sensitivity in piezoresistive, capacitive, and resonance‐based sensors [174, 196, 197, 198, 199]. Moreover, nonlinear mechanical behavior allows strain‐dependent tuning of effective stiffness, wave propagation, and electromagnetic resonance. In parallel, elastic lattices enable conformal and wearable sensing systems capable of interacting safely with biological tissues and flexible surfaces [160, 195]. Hyperelastic materials have been widely exploited in tunable electromagnetic metamaterials, where mechanical deformation modifies resonator spacing or unit‐cell geometry, thereby shifting resonance frequency or effective permittivity [200]. This mechano‐electromagnetic coupling enables adaptive sensing and reconfigurable functional devices [159]. Despite these advantages, hyperelastic materials present several challenges. The viscoelastic behavior of the elastic lattice can introduce hysteresis and time‐dependent creep, which often affect sensing accuracy and repeatability [201, 202, 203, 204, 205]. Material response is also sensitive to temperature and strain rate, and nonlinear constitutive modeling often requires high computational load. Furthermore, fatigue resistance upon long‐term cyclic loading remains an important reliability concern for practical applications [195].

However, polymer‐based metamaterials typically exhibit limited thermal stability, mechanical creep, and relatively high dielectric losses at high frequencies [206, 207, 208, 209]. These limitations can reduce resonance sharpness and sensitivity. To overcome these constraints, polymers are frequently combined into multiphase composites by incorporating functional particulate or nanostructured fillers (i.e., conductive carbon nanomaterials, metallic particles, high‐permittivity ceramics, magnetic inclusions) in order to enhance electrical conductivity, mechanical stability and field‐responsive functionality [174, 210, 211, 212, 213].

3.2. Functional Composite Materials

To overcome the inherent limitations of monolithic polymers, functional composite systems have been widely developed in 3D‐printed metamaterial‐based sensors. In these systems, distinct functional phases are integrated into a polymer matrix to produce electrical conductivity, electromechanical coupling, magnetic reactivity, or multifunctional behavior. Unlike monolithic polymers, the sensing performance of composite systems arises primarily from interfacial interactions and permeable networks created by fillers, rather than from intrinsic properties of the polymer.

• Conductive Composites: Conductive polymer composites combine the processability of polymers with the electrical functionality of conductive fillers (i.e., metal particles, carbon black, multiwalled carbon nanotubes (MWCNTs), graphene, or conductive fibers, etc.) [214, 215, 216, 217]. These materials provide a balance between mechanical flexibility and electrical performance, making them particularly suitable for flexible and wearable metamaterial sensors. Electrical conductivity and dielectric properties can be tuned by controlling filler concentration (referring to the volume or weight fraction of conductive particles dispersed in the polymer matrix), distribution, and orientation, which results in precise modulation of resonance characteristics and strain‐dependent electrical response. AM further enables spatial control of material composition, allowing graded conductivity and multifunctional sensing architectures (Figure 3c,d) [164, 165]. Conductive nanocomposites based on copolymer or MWCNT systems enabled high conductivity at low filler loading and reliable strain sensing fabricated using FDM method [218]. Similarly, 3D‐printable resin and/or carbon nanotube (CNT) composites have shown stable performance in wearable strain sensors with high cyclic durability [219, 220, 221]. Anisotropic piezoresistive responses have been reported in ABS and/or MWCNT based pressure sensors [222]. Conductive TPU or MWCNT composite materials have also demonstrated both capacitive and piezoresistive sensing in soft pneumatic actuators [223]. However, electrical performance strongly depends on filler dispersion and percolation behavior. Nonuniform distribution or poor interfacial bonding can lead to unstable conductive pathways and reduced sensing reliability [224].

• Piezoelectric Composites: Piezoelectric composite systems integrate piezoelectric particles (e.g., ceramic or piezoelectric polymers) into a polymer matrix to achieve electromechanical coupling within flexible printed architectures [225, 226, 227]. The mechanical deformation of these composites induces variation in the piezoelectric phase, enabling direct strain‐to‐electrical signal transduction [228, 229, 230]. For 3D‐printed piezoelectric composites, optimization of formulation and structural design (e.g., auxetic networks) allowed for an enhancement of the piezoelectric coefficient and the haptic sensing performance [231]. Recent developments in flexible piezoelectric composites fabricated by additive processes have shown strong potential in accomplishing integrated sensing and actuation for robotics and wearable systems [232]. Custom‐designed composites containing functional piezoelectric phases within the polymer matrix have improved piezoelectric response and sensitivity, expanding applications in haptic sensing and self‐powered structures [233, 234]. For these reasons, many studies on piezoelectric composite metamaterial‐based sensors have been explored [235, 236].

• Magnetoactive Composites: Magnetic composite materials consist of magnetic filler phases embedded in a flexible polymer matrix, exhibiting mechanical and functional properties that react to magnetism [237, 238]. These composite materials are crucial realizing proximity for sensing and remote control when integrated into metamaterials [239, 240]. An advanced full‐printed magnetoelectric layered composites combining a magnetorheological elastomer (MRE) and a piezoelectric polymer have recently been developed using DIW and BJT method, exhibiting measurable magnetoelectric effects at the nanometer scale [241]. Active magnetic composites have been extensively investigated for actuation in soft robotics, their application in sensing remains limited [242, 243].

• Multi‐Functional Hybrid Systems: Multi‐functional hybrids integrate two or more functional phases within a single composite system to enable coupled sensing methods [174, 244]. A representative study exploited magnetoelectric composites that combine magnetoactive and piezoelectric behavior in a fully printed layered structure [241], as well as a multimodal sensing of external pressure and an electromagnetic field. Such hybrids can achieve enhanced tunability of effective conductivity, permittivity, stiffness, and resonance behavior [23, 245]. However, the increased structural and functional complexity of hybrids system introduces challenges in maintaining mechanical stability, interfacial compatibility, and long‐term reliability, which requires careful choice of materials and process design in AM systems [246, 247].

Functional composite materials provide enhanced electrical, electromechanical, and magnetic responses in 3D‐printed metamaterial sensors through careful integration of fillers and polymer matrices [214, 215, 217, 225, 226, 227, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248]. However, reliance on passive interaction between filler and matrix limits adaptability to various external stimuli and restricts dynamic performance [224, 247]. This realization drives the shift toward intrinsically active and stimulus‐responsive materials, where the material composition and structure are designed to adaptively respond to mechanical, electrical, magnetic, and/or environmental stimulus.

3.3. Smart Responsive Materials

Smart polymers that intrinsically respond to external stimuli (i.e., temperature, humidity, or light) have been explored for AM [249, 250, 251, 252, 253]. These materials can undergo controlled changes in shape, stiffness, or other properties when triggered, offering new routes for adaptive sensing components. Smart responsive materials represent an advanced extension of functional composite systems, in which engineered multiphase architecture enable active or adaptive responses to external stimuli. These systems introduce dynamic coupling mechanisms, resulting in enhanced sensing performance. Electromechanically active composites integrate piezoelectric phases into polymer matrices to enable direct electromechanical transduction [245, 246, 247, 254]. In 3D‐printed piezoelectric composites, the formulation and mechanical design of the structure can strongly influence sensitivity and active response upon dynamic stimuli [255, 256, 257]. An optimized 3D‐printed piezoelectric polymer, a ceramic composite with auxetic materials, developed to achieve to exhibit enhanced piezoelectric coefficients and high output open‐circuit voltages, making them effective as flexible tactile sensors and self‐powered devices [258, 259, 260]. Furthermore, multifunctional piezoelectric composites with integrated electrodes have been fabricated using multi‐material FDM method to achieve both pressure sensing and energy harvesting in wearable robotic platforms (Figure 3f) [88, 89, 167]. Hybrid polymer nanocomposite sensors with improved piezoelectric responses have also been developed, showing increased voltage output by incorporating nano‐scale fillers into the polymer matrix using AM [255, 261, 262]. In addition to mechanical and electrical stimuli, sophisticated systems can be designed to respond to a variety of different external signals. Hybrid piezoelectric‐magnetic actuators, fabricated by AM, exhibit both piezoelectric and magnetic effects, enabling self‐sensing operation and adaptive performance under varying external conditions (Figure 3e) [166, 245, 263]. These multi‐stimuli hybrids illustrate how coupling intrinsic material properties with engineered elastic lattice leads to advanced sensing capabilities. Although 3D printing of such polymers remains challenging due to material formulation and resolution limitations, ongoing developments highlight their promise for future smart metamaterial sensors [61, 73, 160, 161, 264].

3.4. Material–Structure Interaction in Sensing Performance

In metamaterial‐based sensors, performance stems are not solely determined by intrinsic material properties or purely from geometric design; rather, it results from their inseparable combination [159, 265]. While the abovementioned sections discussed a variety of 3D printing materials (ranging from monolithic polymers to functional composites and smart responsive materials), the following section discusses how sensing performance in metamaterials can be enhanced by integrating intrinsic material properties into the unit cell lattice geometry.

In contrast to conventional materials, whose properties are primarily determined by chemical composition and microstructure, the metamaterial generates efficient mechanical, electromagnetic, and multi‐physical responses through geometrically induced field redistribution [266]. A lattice, consisting of topologically tessellated unit cells, can produce distinctly different sensing responses depending on the elastic modulus, dielectric constant, electrical conductivity, viscoelastic damping, and interface coupling characteristics of the constituent materials (Figure 3g) [24, 25, 27]. These intrinsic properties influence the resonance frequency, Q factor, strain location, energy dissipation, and signal stability, thus directly determining the sensitivity and repeatability of the sensor [27, 50, 267]. More importantly, the elastic lattice can amplify, redistribute, or localize physical fields [268, 269, 270]. For example, lattices dominated by stretching or bending can cause strain amplification in specific structural members, significantly increasing electromechanical sensitivity [21, 271]. Similarly, auxetic geometry alters the transverse strain behavior, resulting in a variation of effective electromechanical coupling while providing enhanced mechanical tunability [271]. In electromagnetic metamaterials, the strain of the unit cells alters the spacing and symmetry of the resonator, thus effectively tuning the dielectric, permeability, and resonant frequency [199, 272]. These phenomena illustrate that sensing performance is a response tuned by the topology and intrinsic material characteristics.

The elastic lattice can be understood as an effective material, from the viewpoint of mechanical engineering and material science. Indeed, once fabricated, the lattice no longer behaves as a polymer or composite material. Instead, it exhibits extraordinary behaviours such as effective modulus, negative Poisson's ratio, bandgap properties, or programmable stiffness that cannot be directly predicted from the underlying material alone (Figure 3h) [168, 273, 274]. In this sense, the lattice becomes a material system whose properties are encoded by the architecture. However, despite extensive numerical and experimental studies, a universal theoretical framework capable of analytically describing the quantitative relationship between lattice geometry and effective sensing properties remains unresolved. Recent approaches rely heavily on finite element models [275], efficient environment approximations [276], or data‐driven optimizations [277], highlighting the lack of closed‐form predictive models for general‐architecture sensing systems. AM enables simultaneous control of geometry and material distribution, allowing co‐optimization of architecture and material composition. Multi‐material printing and topology optimization have been widely exploited to tailor spatially varying properties within metamaterial structures, enabling enhanced sensing performance and multifunctionality [160].

4. Metamaterial‐Based Sensing Horizons

With the fabrication technologies and material systems established in the preceding sections, this section examines how metamaterial architectures translate physical stimuli into measurable electrical signals across resistive, capacitive, and inductive sensing modalities.

4.1. Performance Metrics

Establishing fundamental performance metrics is essential for evaluating soft sensors. Unlike conventional rigid‐body sensors that are likely to encumber robotic movements, soft sensors are inherently designed to undergo large deformations. Therefore, evaluating how reliably they transduce applied mechanical stimuli into electrical signals is of paramount importance. The performance of metamaterial‐based soft sensors is primarily characterized by several interdependent key indicators.

A primary metric is represented by sensitivity, which indicates the performance index how well mechanical stimuli (i.e., strain or pressure) can be transduced into measurable electrical signals [278]. Depending on the specific transduction mechanism, this is generally expressed as relative changes in resistance (OR/R), capacitance (OC/P) [278], inductance, or resonant frequency (OL/P, Of/P) [279]. Another critical factor is related to hysteresis, which refers to the discrepancy observed between response curves during the loading and unloading phases. Because this phenomenon—mainly induced by energy dissipation within the dielectric layer or 3D‐printed architectures—can provoke unpredictable responses, minimizing it is crucial for ensuring accurate control [280]. In metamaterial‐based sensors, the Signal‐to‐Noise Ratio (SNR) is a vital parameter that quantifies the capability to discriminate actual mechanical stimuli from electrical and environmental noises, directly correlating with overall measurement accuracy and reliability [281]. Sensors are also evaluated by their detection limit and dynamic range. While the detection limit defines the minimum measurable pressure required for capturing subtle bio‐signals [282], the dynamic range refers to the entire measurement span over which the sensor operates reliably [283]. In parallel, the dynamic performance of these sensors is heavily dependent on their response and recovery time. Achieving fast response is critical for applications demanding real‐time responsiveness, minimizing delays during signal processing and active control tasks [284]. Last but not least, durability is a fundamental requirement for long‐term operation. The integrated materials and structural designs must ensure sustained mechanical compliance and electrical conductivity, preventing critical failures such as structural damage or layer delamination under repeated mechanical deformations [285].

4.2. Resistive Sensors

Among the various conversion methods used in robotic systems, resistive sensors have emerged as a particularly versatile and widely applicable devices due to their simple operating principle, easy signal acquisition, compatibility with a wide range of functional materials, and scalable fabrication processes [286]. The working principle of a resistive sensor is to measure the variation in the electrical conductivity of a material with respect to the stimuli from the environment. The electrical resistance of a material is closely related to its dimensional and intrinsic properties through the formula R = ρd/A (where ρ is the electric resistivity, d and A are the length and cross sectional area, respectively). In general, the resistivity of the material is fixed under designed working conditions; thereby, changes in electrical resistance arises from variations in either the length (d) or the cross sectional area (A). Metamaterial‐based resistive sensors are usually composed of two or more materials. Specifically, conductive layers, deformable materials such as carbon nanotubes (CNTs) [287], graphene [288], metallic nanowires (NWs) [289], or MXene [290] are integrated within a polymer‐based substrate. Conventional resistive sensors (with a positive Poisson's ratio) often lead to a diminished net electrical signal [291], and undesired mechanical and/or cross‐channel interferences between neighboring elements, especially those in high‐density sensor arrays [289]. On the other hand, the metamaterial‐based resistive sensors (that undergo a negative Poisson's ratio or auxetic behavior) have shown strong potential, as next‐generative sensing solutions. Numerous studies have shown that such auxetic behavior can be achieved through specifically engineered geometric architecture rather than intrinsic material properties [289, 292, 293]. Among them, a pioneered study has successfully analyzed electrokinetic properties of 3D‐printed conductive lattice, demonstrating that architected deformation can actively regulate charge transport pathways and significantly enhance electromechanical coupling within the sensing network [294, 295].

Meanwhile, such metamaterials can be formed in 2D layer, which are called planar 2D auxetic materials, including re‐entrant structures (i.e., inverted honeycombs [300, 301, 302] or double‐arrowhead units [299, 303, 304]. Sensing performance can be controlled and/or adjusted through geometric parameters of the structure, e.g., cell angle, rib thickness, and unit aspect ratio [305]. Herein, rotating‐unit cells are composed of rigid shapes such as squares, triangles, or rhombus cells connected at hinges or compliant joints that allow them to rotate under loading [271, 306, 307, 308]. In parallel, Poisson's ratio of these structures can also be controlled through geometric parameters or the merging between them [309]. A representative study on planar 2D auxetic materials introduced four patterns of 3D‐printed auxetic lattice‐type structures (i.e., truss (TR), honeycomb (HN), chiral truss (CT), and re‐entrant (RE), acting as multifunctional compressive strain sensors [296], as shown in Figure 4a. All four patterns were fabricated using an identical material formula consisting of NinjaFlex TPU‐based filament and additives. Finally, the auxetic materials are coated with a 5–6 nm thick graphene (GNP‐UC) layer to improve conductivity and strength. While the TR and HN structures show positive Poisson's ratios, and CT and RE show negative Poisson's ratios, all four patterns have resistance sensing with deformation. Similarly, Fitzgerald et al. [297]. exploited re‐entrant auxetic architectures to develop a biomimetic interlocking interface, as illustrated in Figure 4b), a fully 3D printed multimodal sensor integrated directly onto the fingertip of a robotic hand. By using a standard 11 × 11 tactile sensor array, the working range of this sensor is from 0.1 to 0.26 MPa, and it has sensitivity in both normal and shear directions. The sensor substrate was made of Graphene/CNT/Silicon rubber, while the conductive component was composed of micro‐copper‐silicone composite which were fabricated by means of FDM printer.

FIGURE 4.

3D printed metamaterial‐based resistive sensors. (a) Auxetic lattice CT patterns. Reproduced with permission [296] Copyright 2022, Springer Nature. (b) Multidirectional biomimetic and auxetic structure sensor on the vertebra plateau. Reproduced with permission [297] Copyright 2023, Springer Nature. (c) Prototype of the bike grip sensor with a twist deformation structure embedded at the base of the grip. Reproduced with permission [117] Copyright 2023, ACM. (d) An elastic lattice printed by DLP technology. Reproduced with permission [298] Copyright 2025, American Association for the Advancement of Science. (e) Constriction‐resistive sensors for multidirectional proprioception. Reproduced with permission [43] Copyright 2020, American Chemical Society. (f) Re‐entrant honeycomb structure resistive pressure sensors. Reproduced with permission [299] Copyright 2022, IEEE. (g) Illustration of the CNT‐coated AMM (C‐AMM) auxetic strain sensor. Reproduced with permission [312] Copyright 2025, Wiley. (h) 3DGP samples. Reproduced with permission [42] Copyright 2019, Elsevier.

Another pioneered works on metamaterial based resistive sensors can be found in [310]. In this work, the Auxetic Bilayer Conductive Mesh Strain Sensor (ABSS), composed of multi hardness silicones, has shown strong potential in using them for wearable and biomedical applications. Specifically, each unit cell was integrated into a deformable conductive elastomer with a crack‐sensitive nanotube network, enabling strain‐dependent modulation of conductive pathways through geometry‐induced structural deformation. This work highlights how auxetic architecture can amplify electromechanical coupling and improve sensing responsiveness beyond conventional soft resistive sensors. Moreover, by combining multi‐material AM with complementary deposition techniques, the study provides an effective strategy for fabricating highly compliant, wearable sensing systems with strong potential for physiological monitoring and human–machine interaction (Figure 4c) [117]. In addition to demonstrations of auxetic architectures for wearable sensing, extended studies have shown metamaterial‐based resistive sensors toward more advanced fabrication strategies and diversified functional applications. For example, Zhu et al. introduced a fully 3D‐printed metamaterial sensor with an architected U‐shaped geometry (Figure 4d) [298]. This structure with the wall thickness of 320 µm, demonstrated how multi‐process AM can be used to integrate structural frameworks, conductive pathways, and embedded electronic functionality within a single sensing platform. By combining vat photopolymerization and material extrusion–based printing, their work illustrates a practical route toward monolithically fabricated soft sensing systems and validates the feasibility of architected metamaterial sensors in interactive applications such as touch‐based remote control. Beyond wearable interfaces, metamaterial resistive sensing concepts have also been extended to soft robotic proprioception. Mousavi et al., (Figure 4e) [43], developed an architected tactile sensing system capable of selective multidirectional strain detection with high gauge factor, highlighting the potential of metamaterial geometries for directional deformation sensing in compliant robotic bodies. Similarly, Oliveira et al. reported a flexible auxetic metamaterial pressure sensor based on a double‐arrowhead unit‐cell design exhibiting near‐zero Poisson's ratio and rapid response (Figure 4f) [299]. Their work demonstrates how tailored metamaterial kinematics can be engineered to achieve stable pressure sensing performance with minimized lateral deformation coupling.

While the studies discussed above primarily focus on functional integration and application‐specific implementations of architected sensors, another major research direction concerns the transition from planar (2D) architectures to fully three‐dimensional (3D) metamaterial. In contrast to 2D structures, which are only suitable for thin‐film or membrane‐type sensors as discussed above, 3D structures directly exploit volumetric deformation to exhibit more complex mechanical behaviors such as multidirectional bending, rotational motion, and curved surface deformation, fundamentally different from the common folding or hinge mechanisms in 2D architecture [308, 311]. The spatial interconnection of unit cells in three dimensions — such as re‐entrant foam, Schwarz P‐D surfaces, and Gyroid geometries are interconnected in 3D space, resulting in improvements in load distribution and isotropy during deformation compared with planar designs. Several representative studies demonstrate how these 3D architectures fundamentally enhance sensing functionality. Kang et al. [312] presented a periodic cubic lattice material with embedded central spherical voids for compressive load sensing. A fully 3D printed multi‐material strain sensor was fabricated by DLP process and coated with CNTs to form a pressure‐responsive conductive network, capable of sensing in both capacitive and resistive modes (Figure 4g). Their work illustrates how architected geometry can actively regulate mechanical–electrical coupling, rather than merely serving as a passive structural support. In the field of soft robotics and tactile sensing, Huang et al. presented a microstructure of three‐dimensional graphene‐PDMS (3DGP) with design parameters including diameter of filaments‐D, interaxial angle‐θ and interlayer space‐L (Figure 4h) [42]. The 3DGP structure demonstrated superior properties as a strain sensor with ultrahigh sensitivity, specifically excellent stability and a gauge factor have been reached. A key insight from this work is that sensing behavior arises from coupled multiscale deformation mechanisms: macroscopic compression modifies layer spacing, while microscale structural rearrangement of the conductive network dominates resistance evolution. This multiscale interaction produces non‐monotonic resistance responses, demonstrating how 3D architecture enables sensing phenomena unattainable in conventional planar systems. At the system level, Sakura et al. presented a 3D lattice‐based for deformation sensing [117]. These materials are fabricated using an FDM printer and can work with three‐dimensional compression, twist and large shear deformation. Importantly, their work advances the concept of structural sensing, in which the load‐bearing architecture simultaneously functions as the sensing element. This approach blurs the traditional distinction between sensor and host structure and suggests a new design paradigm for soft robotic bodies and adaptive mechanical systems.

The most important parameter for a resistive sensor is the sensitivity S, or Gauge factor, which is typically defined as the relative change in resistance value per unit input pressure S = (OR/R)/P. However, in some cases, the sensitivity S can also be defined through the current output value instead of the resistance value S = (OI/I)/P [292]. To date, studies have focused on improving the Gauge factor [43, 313] Other parameters, including operating range and linearity [314, 315], hysteresis, and response time [316, 317] have also been considered and improved in recent studies.

4.3. Capacitive Sensors

Capacitive sensors have emerged as one of the promising mechanisms for soft robotics and wearable applications due to their advantages of high sensitivity, fast response and scalability on sensor array implementation [49]. The ability to organize capacitive sensors into array though spatially defined electrode patterning enables distributed pressure mapping and tactile images with spatial resolutions suitable for object and texture recognition (Figure 5a) [318].

FIGURE 5.

3D printed metamaterial‐based capacitive sensors. (a) 3 axis detecting sensors, 1 lower electrode, 4 top electrodes. Reproduced with permission [318] Copyright 2023, Springer Nature. (b) 3D printable metamaterial structures by fabricating specific cell walls from conductive filament. Reproduced with permission [44] Copyright 2021, ACM. (c) Fully 3D printed soft sensor with tilted dielectric layer. Reproduced with permission [319] Copyright 2025, Wiley. (d) Fabricated capacitive tactile sensor composed of an AMM sandwiched between Cu/Ecoflex electrodes [312] Copyright 2025, Wiley. (e) Capacitive pressure sensor with a designed microstructure fabricated by DLP 3D printing [320] Copyright 2025, MDPI. (f) Copper electrodes on a flexible polyimide, separating normal and shear force by lattice structures. Reproduced with permission [283] Copyright 2024, Springer Nature. (g) Flexible ITO PEN film electrodes and 3‐axis alignment airgap dielectric layer. Reproduced with permission [278] Copyright 2024, MDPI. (h) Array of multi‐layered blocks. Reproduced with permission [34] Copyright 2025, Springer Nature.

The fundamental principle of capacitive sensing relies on the measurement of capacitance variance in response to external stimuli (C = ε ₀ ε_rA/d where ε ₀ representing the permittivity of vacuum, ε_r representing the permittivity of dielectric materials, A representing the overlapping area of two electrode layers, and d representing the distance between two electrodes) [321]. When architected materials are integrated into a capacitive sensor, mechanical deformation induced by forces and strains changes one or more of these parameters which cause capacitance change [322].

Materials of electrodes should satisfy requirements of high electrical conductivity, mechanical compliance, and durability on repeated deformation. Nano wire [323, 324], liquid metal [325, 326], conductive fabric [327, 328], and conductive composite materials [44, 319, 329] provide these characteristics when composing electrodes. Each material offers distinct advantages for specific applications.

The fabrication of metamaterial‐based capacitive sensors can be divided into two distinct approaches that differ in manufacturing steps and architecture of resulted device. Multi‐step method is the first approach of manufacturing. It involves the independently fabricated electrodes and dielectric layer, integrating by chemical bonding, lamination, and mechanical assembly processes [2]. In this method, the dielectric layer is mostly fabricated by 3D printing of photopolymer resins via stereolithography (SLA) [278], digital light processing (DLP) [283], or casting of silicone elastomers (PDMS, Ecoflex) using 3D‐printed mold made from acrylonitrile butadiene styrene (ABS) [330], polylactic acid (PLA) [329], and photopolymer resin. In this view, it is worth noting these printing methods allow for selecting proper materials, as electrodes and dielectric layers can be optimized independently to achieve the desired electrical and mechanical properties [323]. On the other hand, it could be vulnerable to delamination and/or material failures under repeated mechanical loading. Therefore, to ensure consistent sensor performance in multiple unit cells, accurate assembly tolerances should be taken account [331].

In recently, monolithic sensor fabrication through multi‐material addictive manufacturing process is emerging, which fabricate electrodes and dielectric layer in a single printing operation [44, 319]. This method requires multi‐material 3D printers that can switch between conductive and non‐conductive filament during one operation [332]. For instance, Jun et al., fabricated meta‐material sensors that are called MetaSense by selectively print layers from conductive filament on the dielectric filament layer, thereby creating electrodes that can measure the variation in capacitance induced by mechanical deformation of unit cells (Figure 5b) [44]. Similarly, Fei et al., fabricated fully 3D‐printed soft sensors having high toughness and measurement range, by exploiting silicone‐based ink with electrical properties, conductive electrodes and dielectric pyramidic structures (Figure 5c) [319]. The monolithic fabrication method reduces bonding failure and manufacturing complexity and enables consistent sensor performance [319]. On the other hand, this method has inherent limits on material selection, as both conductive and dielectric components must be suitable for a reliable fabrication process [331].

The primary design parameter is the compressibility of the dielectric layer, which is directly associated with the sensor's sensitivity by changing the stiffness of the sensor [52, 333]. Architected structures such as a pyramidic structure [327] enables highly sensitive detection of low pressure. On the other hand, excessive flexibility can lead to premature mechanical densification under low pressure, which leads to a lack of dynamic range of sensors which causes nonlinear response characteristics [278]. To address this challenge, Cui et al., developed the multi‐scale dielectric architecture that exhibited progressive changes of stiffness, which enables a larger range of detection [332]. Similarly, Xiao et al., fabricated a tilted plate dielectric structure to achieve four steps of deformation, which enabled a single sensor to detect a large range of external stimuli [319].

Architected structures also enable special resolution in capacitive sensor arrays. The application of negative Poisson's ratio structure in the dielectric layer can reduce crosstalk between sensing elements in arrays by reducing lateral expansion during compression, improving the accuracy of pressure mapping of sensor arrays [45]. Kang et al., fabricated a cubic lattice via DLP printer. The elastic lattice undergone a centripetal densification effect, where the material selectively concentrated toward thed center of the loading zone upon deformation (Figure 5d) [312]. Beyond geometry‐driven lattice designs, DLP 3D printing of ion composite photosensitive resins enables the precise fabrication of dielectric microstructures, where electric double layer formation enhances the initial capacitance and sensitivity (Figure 5e) [320]. Furthermore, dielectric layers can also be designed to discriminate normal and shear force components based on distinct capacitive response signature (Figure 5f) [283]. This modifiable design and rapid fabrication on complex dielectric structures enable change of sensitivity, detecting range, directionality, and spatial resolution through geometrical modifications, without changing characteristics of materials [45].

Recent applications of metamaterial‐based capacitive sensors have widely spread across soft robotics, wearable health monitoring, and human‐machine interaction [34, 45, 47, 278, 319]. In soft robotics system, Loh et al., integrated capacitive sensor arrays with universal jamming grippers enabled real‐time detection of reaction force to the object, preventing damage to delicate object while ensuring exact manipulation [45]. Moreover, as capacitive sensing system is applicable to curve and deform surfaces by array, it is well‐suitable for soft robotics skins that cover complex surfaces such as continuum manipulators or robot hands (Figure 5g) [279]. Inspired by human tactile sensing, multilayered mechanical metamaterial‐based soft tactile sensors exploit step‐by‐step locking and layer‐wise deformation to achieve multiple sensitivity regimes across distinct force sensing ranges within a single structure (Figure 5h) [34]. The development of 3D‐printed sensor integrated devices, such as personalized insoles fabricated in single printing operations [47], demonstrating the potential for specific customization of wearable sensing system [47]. As the architected designs and multimaterials printing technologies continue to advance, the capacitive sensor will expand the application space and performance [49].

4.4. Inductive Sensors

Inductive sensors have garnered significant attention in soft robotics, owing to their low hysteresis and wide bandwidth [334]. Fundamentally, these sensors operate on the principle of eddy currents for tactile and proximity perception. When an alternating magnetic field generated by a primary coil permeates a proximal conductive target, it induces eddy currents within the material. These currents generate a secondary magnetic field that opposes the primary flux, thereby modulating the effective impedance of the sensor coil (Figure 6a) [279]. This mechanism allows for the precise, non‐contact measurement of the separation distance, enabling the detection of conductive objects and the quantification of soft structure deformation [335].

FIGURE 6.

3D printed metamaterial‐based inductive sensors. (a) Schematic diagram illustrating the fundamental working principle of an inductive sensor based on magnetic coupling. Reproduced with permission [279] Copyright 2025, MDPI. (b) Single coil inductive sensor designed for short‐range requirements. Reproduced with permission [336] Copyright 2024, MDPI. (c) Battery‐free wireless origami pressure sensor integrated an insole. Reproduced with permission [47] Copyright 2022, Springer Nature. (d) Fabricated split‐ring resonator structures of various sizes for metamaterial sensing applications. Reproduced with permission [337] Copyright 2025, Springer Nature. (e) Micro‐CT image showing the internal structure of a 3D‐printed spiral induction coil. Reproduced with permission [338] Copyright 2023, Springer Nature. (f) 3D printable shape‐changing devices integrated with helix‐and‐ lattice structures. Reproduced with permission [339] Copyright 2024, ACM. (g) Self‐powered inductive sensor, with a TPMS based lattice structure. Reproduced with permission [340] Copyright 2024, Taylor&Francis. (h) Flexible origami‐inspired inductive sensor demonstrating inductance variations at different folding angles. Reproduced with permission [46] Copyright 2022, Wiley.

Coil geometries for inductive sensors integrated with 3D‐printed metamaterial structures can be categorized into planar and three‐dimensional configurations. Two‐dimensional planar coils are fabricated by conductive wires coiled on flexible printed circuit board (FPCB) [341], mainly providing noncontact measurement. For applications requiring highly localized detection, Feldkamp investigated a single‐coil inductive sensor specifically tailored for short‐range measurements (Figure 6b) [336]. Kim et al., integrated these two‐dimensional planar coils with origami structure, serving as a compressible dielectric layer that changed its height and geometry in response to external mechanical stimuli (Figure 6c) [47]. In addition to continuous coil designs, researchers fabricated split‐ring resonator (SRR) structures of various sizes, utilizing their distinct electromagnetic resonance characteristics for advanced metamaterial sensing applications (Figure 6d) [337]. Three‐dimensional helical coil architecture provides volumetric sensing techniques; by changing the distance between individual coil layers, this mechanical compression alters the magnetic flux linkage, resulting in a measurable shift in inductance [14]. Zhang et al., utilized micro‐CT imaging to visualize the precise internal architecture of 3D‐printed spiral induction coils, ensuring the structural reliability required for stable volumetric sensing (Figure 6e) [338]. When these coils are integrated with the lattice, they can provide real time inductance feedback as they deform with lattice framework, effectively translating structural strain into electromagnetic signals (Figure 6f) [339].

The performance of inductive sensors is characterized by their transduction mechanism, evaluated through sensitivity, hysteresis, and SNR similar with resistive and capacitive sensors. Sensitivity in inductance sensors is typically defined as the inductance (L) or resonant frequency (f) change per pressure on unit (OL/P, Of/P) [279], representing the sensor's fundamental ability to transduce mechanical stimuli into electrical signals. Similar with energy dissipation in capacitive dielectrics, hysteresis in metamaterial‐based inductive sensors is primarily defined by mechanical properties of the 3D‐printed architecture [280]. Inductive sensing offers inherent advantages in SNR due to its robust electromagnetic coupling, which remains immune to environmental factors such as humidity, temperature or parasitic capacitance that affect the signal quality of capacitive and resistive sensors [342]. Furthermore, self‐powered inductive mechanism provides distinctive advantage of lowering the detection limit, without the need of the external power source which causes parasitic noises (Figure 6g) [340].

The integration of inductive sensor and 3D‐printed metamaterials architecture enables a wide range of applications by applying both electromagnetic variations and physical deformation on coils. Lee et al., integrated a helical coil with lattice structure with tunable stiffness properties, applied in inductive joystick which detected tilting of the sensor. Mechanical compliance and sensing functionality were simultaneously achieved through metamaterial design [339]. Wu et al., fabricated triply periodic minimal surface (TPMS) lattice with self‐powered operation which was suitable for wireless motion sensing where external electrical source is impractical, utilizing electromagnetic induction for battery‐free operation [340]. Furthermore, Huang et al., developed a foldable self‐inductance sensor based on multi‐fold planar coils, which translated the physical deformation of the origami structure into measurable inductance variations at different folding angles (Figure 6h) [46]. Similarly, origami‐based wireless inductive sensors applied in insoles, offer deployment in environments where battery replacement is limited, utilizing geometric transformation of origami structure for passive sensing [47].

4.5. Other Sensing Solutions

While resistive, capacitive, and inductive sensors rely on variation in the electrical properties of conductive materials, several alternative transduction mechanisms have also been explored in conjunction with metamaterial architecture. These include piezoelectric coupling, triboelectric contact electrification, and vision‐based optical tracking, each of which operates on different underlying principles. Moreover, metamaterial architecture enhanced the performance of these mechanisms in ways that are distinct from their role in conductivity‐driven systems.

In literatures, Wu et al. developed a vision‐based tactile sensor by applying pyramid structures with markers tracked by machine learning based force prediction (Figure 7a) [343]. Furthermore, Piezoelectric sensors convert mechanical stress into electric charge through crystal polarization, suitable for dynamic force measurement. Metamaterial architecture enhances piezoelectric performance by amplifying strain and concentration stress where external stimuli apply [345]. Zhuo et al., developed a hexagonal auxetic lattice to harvest piezoelectric energy, demonstrating that the synclastic effect of the auxetic structure can amplify the strain compared to non‐auxetic structures (Figure 7b) [254]. Similarly, Wei et al., showed that 3D‐printed lead zirconate titanate (PZT) ceramics with reentrant hexagonal lattice can improve on voltage in comparison to bulk sensor by enhanced stress distribution [260]. Moreover, Pei et al., fabricated programmable poly vinylidene fluoride (PVDF) metamaterial via FDM 3D printing which enabled self‐polarization without requiring conventional poling processes [346]. In addition, metamaterials can also apply triboelectric sensors, which generate electricity through contact electrification and electrostatic induction when two different materials repeatedly contact and separate [347]. Metamaterial structures enhance triboelectric performance by increasing the effective contact area and controlling deformation patterns [348]. For instance, Liu et al., generated 3D‐printed pyramid kirigami metamaterials to achieve dual‐mode sensing (Figure 7c) [344]. The compressible structure provided triboelectric displacement sensing though contact and separation events, while integrated piezoelectric elements monitored pressure [349].

FIGURE 7.

Other sensing solutions using 3D‐printed Metamaterials (Piezoelectric, Triboelectric, and Vision based sensing), and toward metamaterial‐based multimodal and multifunctional sensing solutions. (a) Vision‐based tactile sensor using soft robotic metamaterial deformation. Reproduced with permission [343] Copyright 2024, Elsevier. (b) Piezoelectric sensor enabled by an auxetic metamaterial architecture. Reproduced with permission [254] Copyright 2023, Wiley. (c) Water containing elastomer based triboelectric sensor with an ionrich interface. Reproduced with permission [344] Copyright 2024, Springer Nature. (d) Resistive and capacitive multimodal tactile sensing. Reproduced with permission [312] Copyright 2025, Wiley. (e) Multifunctional piezoresistive tactile sensing for pressure, shear, and temperature. Reproduced with permission [297] Copyright 2023, Springer Nature.

Comprehensively, these alternative transduction mechanisms expand the functional scope of metamaterial‐based sensing beyond conductivity‐driven approaches, offering complementary capabilities such as self‐powered operation, dynamic force measurement, and contact‐free tactile perception. Their integration with metamaterial architecture suggests that geometry‐driven performance enhancement is not limited to a specific transduction principle but represents a broadly applicable design strategy across diverse sensing modalities.

4.6. Multifunctional and Multi‐Modal Sensors

In addition to single‐modal sensing, 3D‐printed metamaterial‐based sensors can integrate multiple transduction mechanisms to achieve enhanced functionality and performance. For instance, inductive‐capacitive configurations enable wireless, battery‐free sensing through resonant frequency modulation. Kim et al., fabricated 3D‐printed Miura‐ori origami metamaterial structures that integrated inductive coils and capacitive elements [349]. The multi‐layer Miura‐Ori structure, which was fabricated by conductive and flexible polymers, manifested as a predictable deformation structure. When external stimuli were applied, the origami metamaterial transformed the capacitance through gap decreasing and inductance through coil, shifting the LC circuit frequency. This battery‐free wireless architecture was adapted to a personalized smart insole for real‐time walking habit monitoring, with the metamaterial structure continuously providing mechanical support, cushioning, and wireless sensing [350].

Resistive‐capacitive sensing utilizes advantages of both sensing methods with a unified auxetic metamaterial structure [351]. fabricated a 3D auxetic metamaterial tactile sensing platform based on a cubic lattice with spherical voids, fabricated via DLP printing method. Two sensing methods were implemented by different methods in the same structure: capacitive sensor that utilized lattice as a dielectric layer, resistive sensor that utilized carbon nanotube coated lattice as a resistor. The auxetic‐driven deformation mechanism enhances sensitivity in both modes while minimizing crosstalk between adjacent sensing units (Figure 7d) [312].

In addition to RLC sensing, integrating with other sensing methods has been widely investigated. For instance, Kim et al., fabricated 4D‐printed kirigami metamaterials leverage shape memory polymers to achieve reconfigurable dual‐mode sensing combining triboelectric displacement detection and piezoelectric pressure monitoring [349].

In addition to multi‐modal mechanisms, multi‐functional sensing approaches exploit a single transduction principle to simultaneously detect multiple physical parameters, where signal decoupling is achieved through the structural anisotropy and geometric design of the metamaterial architecture [84]. While multi‐modal sensors rely on mechanically distinct operating principles, multifunctional sensors achieve parameter discrimination through structural anisotropy and geometric design of the architected unit cell. Wei et al., demonstrated a fully 3D‐printed piezoresistive tactile sensor integrating an auxetic re‐entrant lattice with a biomimetic interlocked papillae interface, enabling simultaneous detection of normal pressure, directional shear forces, and temperature variations within a single resistive sensing framework (Figure 7e) [297].

These multifunctional and multi‐modal metamaterial sensors extend sensor technologies beyond single‐modal sensing operation, enabling intelligent systems, detect several stimuli at once, and operating wireless within a single architected structure [312, 349, 351].

5. Challenge and Future Perspectives

Despite the remarkable achievements reported in abovementioned sections, the 3D‐printed metamaterial‐based soft sensor still presents open challenges, ranging from fabrication, and material to hierarchical sensing architectures. In the viewpoint of the fabrication, 3D‐printed metamaterial‐based sensors should be rigorously designed, considering a trade‐off between geometric resolution and material compatibility for different 3D printing technologies. Specifically, while the VPP enables highly sophisticated unit‐cell geometries and topologies with high‐resolution, constitutive hyperelas‐tic materials having conductivity and their mechanical properties (i.e., viscosity, Young's modulus, elongation at break, etc.) need to be studied further. Moreover, multi‐material composition through the VPP is barely possible, thus polylithic constructions (bonding different material layers) have been pursued. On the other hand, the MEXs allow multi‐material deposition of conductive and elastomeric components, yet due to their inherently coarse resolution, the interlayer void (of the topologically tessellated unit cells) could not be formulated successfully, resulting in mechanical and electrical discontinuity as well as poor structural fidelity to achieve micro‐architected sensing elements [75, 136]. In particular, the fabrication of the self‐supporting elastic lattice adds further complexity, as post‐processing steps and support removal are likely to delaminate or damage delicate conductive networks embedded within the structure. Moreover, achieving reliable bonding between dissimilar printed materials, particularly between conductive and dielectric phases, remains an open challenge with direct consequences for sensor durability under cyclic mechanical loading [331]. Based on these fabrication challenges, hierarchical arrangement of the unit cell allows for programmable instability, adaptive reconfiguration, and multi‐physical coupling, where material design and structural architecture are treated as a unified functional entity rather than separate considerations [352, 353, 354]. Therefore, the next frontier in metamaterial‐based sensors is not only in developing new polymers or composites, but also in understanding and designing materials structured as integrated material systems. Achieving predictive design capabilities will require advances in multiscale modeling, nonlinear mechanics, and data‐driven materials informatics, enabling the rational design of sensor networks with tuned and programmable responses [61, 73].

Meanwhile, in the viewpoint of sensing technologies, each transduction mechanism reported above carries its own set of unresolved limitations. Resistive sensors, despite their structural simplicity, struggle to simultaneously achieve high sensitivity and a wide operating range [43, 313]. For capacitive sensor arrays, crosstalk between neighboring sensing elements and interfacial delamination under repeated loading are persistent concerns, and these problems increase when electrode and dielectric layers are assembled separately rather than fabricated in a single printing process [331]. Inductive sensors are generally less sensitive to environmental interference, yet their operating principle inherently limits detection to conductive targets, which constrains their applicability in soft robotic systems where the objects of interest are often composed of non‐conductive materials [279]. To partially address these limitations, signal conditioning strategies including recursive filtering, exponential moving average methods, and baseline normalization have been explored to improve measurement stability and reduce noise [355, 356, 357, 358]. The selection of appropriate signal processing methods should therefore be guided by the required response characteristics and the nature of sensor output variability, rather than applied as a universal solution [358].

To this end, pioneering studies using artificial intelligence have been reported [84, 159, 160]. The integration of machine learning and data‐driven design tools with topology optimization frameworks enables inverse design of metamaterial unit cells with targeted electromechanical responses, leading to rapid development of sensors having well‐defined sensitivity, linearity, and dynamic range [82, 83]. At the system level, embedding distributed metamaterial sensing networks within soft bodies represents a pathway toward embodied intelligence in next‐generation soft robotic platforms, as the same architected structure can simultaneously provide mechanical support and proprioceptive feedback without requiring additional sensing components [49].

6. Outlook and Conclusion

3D‐printed metamaterial‐based soft sensors offer distinct advantages across material, mechanical, and electrical aspects. In particular, tunable stiffness through architected structures, programmable deformation directionality, and electromechanical transduction efficiency, collectively addressing the fundamental limitations of conventional rigid sensing solutions in soft robotics and wearable systems. This review work has classified these sensors according to their fabrication technologies, in material systems, and transduction mechanisms, and has examined how architected geometry and material properties control sensing performance across resistive, capacitive, and inductive modalities.

Among the AM technologies, their suitability for metamaterial‐based soft sensors can be evaluated by geometric resolution, compatibility with elastic and conductive materials, and multimaterial capability. In this view, vat photopolymerization has the strongest potential, which can achieve sub‐millimeter scale resolution for unit cells, support elastic and conductive resins, and provide smooth surfaces that suppress hysteresis. On the other hand, material extrusion remains valuable in fabricating multi‐material composite sensors, while powder bed fusion is suitable for high mechanical load systems. Therefore, while material extrusion and powder bed fusion have their roles in hybrid and structural sensing systems, vat photopolymerization is a promising process for 3D printed soft sensors, where multi‐material integration remains a challenge to be addressed.

Up to our knowledge, the multi‐material printing represents the most actively researched, which enables spatial programming of functional gradients within monolithic structures and eliminating interfacial delamination (that currently limits sensor durability) while resolving reproducibility and repeatability that occurs by man‐made processes. Recent advances in vat photopolymerization and material jetting technologies show strong potential in achieving simultaneous deposition of conductive, dielectric, and structural materials with high resolution. However, it is worth noting that achieving reliable bonding among different layers and preventing unexpected material mixing during printing remains challenges. Beyond performance enhancements, these architected sensors can integrate with soft robotics systems and wearable health monitoring scenarios. Smart clothes, soft gripper skins, and tactile array are representative examples of this integration. The development of distributed sensor networks embedded within soft actuator bodies, where metamaterial structures provide both locomotion and proprioception, could enable truly embodied intelligence in next generative soft robot paradigms.

In conclusion, 3D‐printed metamaterial‐based soft sensors have been developed from laboratory demonstrations to enabling technologies for next generative robotics, enabling human‐machine interaction. As multi‐material printing capabilities advance and design methodologies become more sophisticated, metamaterial sensors are expected to become components across applications on healthcare, human‐machine interaction, and soft robotic systems. We envision that the ultimate realization of soft machines with human like perceptive capabilities will depend on continued innovations in metamaterial sensing architectures and their seamless integration with actuation, control, and power systems.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The authors have nothing to report.