ABSTRACT Vision‐based three‐dimensional (3D) reconstruction plays an important role in robotic grasping and manipulation by providing accurate 3D information for target localization and motion planning. These tasks are usually performed in confined environments, where the vision‐based measurement systems are required to provide higher measurement accuracy while occupying a smaller volume. However, traditional stereo vision systems struggle to simultaneously achieve high precision and small volume, which limits their deployment in confined environments. In this paper, we developed a rotatable‐lens monocular vision system (RLMVS) consisting of a monocular camera, a wedge‐shaped lens, and an automatic rotation mechanism for the wedge‐shaped lens. A flexible learning‐based calibration method was proposed for the RLMVS. Furthermore, we proposed a rotatable lens‐based 3D reconstruction (RLR) method for RLMVS, which is the first time using the refraction of the light induced by a rotating wedge‐shaped lens to enable 3D reconstruction. Experimental results demonstrated that, compared with the traditional stereo vision system, the RLMVS reduced the RM measurement error by 61.2%, and reduced the overall volume of the 3D sensing modules by 46.5%. These advantages make the RLMVS particularly suitable for robotic manipulation in confined environments. We integrated the RLMVS with a robotic arm and performed grasping of small objects in a confined environment, demonstrating that the RLMVS provided a compact, low‐cost, yet accurate 3D sensing modality for robotic manipulation in confined environments.
ABSTRACT Owing to their high power‐to‐weight ratio and inherent flexibility, tendon–sheath mechanisms (TSMs) have been extensively adopted in robotic systems that demand compact designs and adaptability to complex environments. However, nonlinear friction between the tendon and the sheath inevitably induces tendon‐elongation and backlash‐like hysteresis, making accurate position transmission control in TSMs highly challenging. This short communication critically reviews representative strategies developed to address this issue, identifying their limitations in meeting the practical demands of TSM‐driven robotic systems. Based on experimental findings, we further propose a promising solution that simultaneously fulfills the requirements of compactness and adaptability. This work offers direct and valuable insights into the fundamental limitations of existing strategies for TSM position transmission control, whereas the proposed remedial solution demonstrates considerable potential for broad application across various TSM‐driven robotic systems.
ABSTRACT In recent years, rapidly advancing artificial intelligence has injected new momentum into advanced robotics, enabling its application across numerous domains. However, traditional computing architectures struggle to support robots in achieving complex environmental perception and autonomous decision‐making capabilities. Neuromorphic electronic devices, which mimic biological nervous systems, offer advantages such as low power consumption, parallel computing, adaptive learning, and event‐driven processing. These attributes show considerable promise in autonomous robot perception, and human–machine interaction. This review presents a comprehensive analysis of the material systems used in neuromorphic devices. Importantly, performance metrics such as memory characteristics and plasticity are employed to evaluate device performance. Building on this hardware foundation, the article further explores the software and hardware implementations of neuromorphic computing based on such devices, with a focus on recent applications in the field of embodied intelligent robotics. Finally, current technological challenges are discussed, and future research directions are proposed, with the aim of paving the way for next‐generation intelligent robotic systems.
ABSTRACT Small‐body sampling robots are crucial for asteroid and comet exploration as well as in situ resource utilization, yet they face unique challenges of microgravity, irregular terrain, and uncertain surface properties. Although existing studies have advanced sampling, mobility, and anchoring technologies, current reviews often treat these technologies in isolation, failing to reveal their intrinsic coupling relationships and hinder holistic research and development. To address this gap, this review provides a unified review of small‐body sampling robots, focusing on the mechanical aspects of the three core functional modules: sampling, mobility, and anchoring. It surveys state‐of‐the‐art techniques demonstrated in missions, highlights emerging approaches and compares performance trade‐offs in efficiency, adaptability, and technology readiness. The review further highlights critical challenges, including the coupling effects among the three functional modules as well as additional challenges arising from other factors. Finally, this review outlines the prospective development trends of small‐body sampling robots. By consolidating lessons from past missions and emerging innovations, this work aims to serve as both a reference for ongoing research and a road map for the design of next‐generation autonomous sampling robots.
ABSTRACT In this study, we present a method for the direction and value estimation of network latency in telesurgery systems using deep reinforcement learning (DRL). Network latency can impair surgical teleoperation because the delay in the command and feedback signals affect the system to not perform in real time. We utilized a deep Q‐network agent trained on a set of 60 features extracted from raw latency data. The framework was designed as a two‐stage process: first, by predicting the binary directional trend (up/down) of latency and second, by using the direction as an input to solve for the value estimation. The DRL agent achieved a direction prediction accuracy of 85.8% on a held‐out test set, which represents a 35.8% improvement over a random classifier. The subsequent value estimation achieved a low mean absolute percentage error (MAPE) of 7.23% and a 92.55% accuracy. Notably, the framework provides 85.8% directional accuracy, a capability absent in pure regression methods. While simpler regressors achieve marginally lower MAPE, they cannot predict whether latency will increase or decrease, limiting their utility for proactive control in telesurgery. This approach demonstrates that a DRL‐based direction and value estimation paradigm can provide a better and computationally efficient signal for latency mitigation, offering a practical pathway to improve the resilience of tele‐surgical systems operating over unpredictable networks.
ABSTRACT Microrobots, typically with dimensions between a millimeter and a few microns, have emerged as a transformative class of intelligent machines at the convergence of robotics, materials science, and biomedicine. Inspired by the motility and adaptability of microorganisms, these systems are designed to operate in low‐Reynolds‐number environments, where viscous forces dominate and conventional macroscopic actuation principles no longer apply. This perspective outlines the fundamental physical constraints governing microscale locomotion, reviews state‐of‐the‐art propulsion strategies, including magnetic, acoustic, chemical, optical, and biohybrid actuation, and discusses recent progress in micro/nanofabrication, functional materials, and embodied intelligence. Key biomedical applications, such as targeted drug delivery, minimally invasive diagnosis, microsurgery, and swarm‐assisted therapy, are examined with respect to their clinical promise and translational challenges. Finally, critical issues related to energy supply, control and imaging, manufacturing scalability, regulatory pathways, and ethical considerations are analyzed, and future directions toward autonomous, intelligent, and clinically deployable microrobotic systems are proposed. The continued integration of advanced materials, high‐resolution imaging, and AI‐driven control is expected to accelerate the transition of microrobots from laboratory prototypes to practical tools for next‐generation precision medicine.
ABSTRACT Minimally invasive surgery (MIS) and robot‐assisted MIS reduce access trauma but attenuate direct haptic cues at the tool–tissue interface. Consequently, interaction forces are often inferred indirectly from vision, which can increase uncertainty during force‐sensitive tasks such as grasping, suturing, palpation, and cannulation. Fiber Bragg grating (FBG) sensors are attractive for MIS instruments because they are compact, multiplexable, and intrinsically immune to electromagnetic interference, enabling distal force measurement in confined operative spaces. This review surveys FBG‐based force sensors reported for MIS and robot‐assisted MIS, covering sensing principles, mechanical transducer architectures, and interrogation strategies. We summarize representative implementations across ophthalmic microsurgery, natural‐orifice procedures, vascular intervention, and laparoscopy and compare key performance metrics reported in the literature. We further review force/temperature decoupling approaches from calibration‐matrix methods to learning‐based models and discuss practical constraints affecting translation, including packaging and sterilization, strain‐transfer drift, temperature compensation, dynamic bandwidth, and integration with surgical workflows. Finally, we outline research directions to improve robustness and validation toward clinically deployable optical force‐feedback systems.
Magnetically steerable guidewires show significant advantages in vascular intervention surgery by enabling precise and minimally invasive navigation. Guidewires with embedded magnetic materials are widely studied. However, when an external magnetic field directly acts on the distal tip, it often experiences high acceleration, which leads to unpredictable motion and potential vessel injury. In contrast, coil‐based magnetic guidewires allow precise adjustment of magnetic forces and torques through current, and immediate de‐energization eliminates this risk. Nevertheless, single‐coil designs provide limited deflection angles, which restrict their steering performance in complex vessels. To address these issues, this paper presents a dual‐coil magnetic guidewire (DCMG) that achieves dual‐attraction, dual‐repulsion, and spoon‐shaped modes under an external magnetic field. These three modes provide adaptable curvature control, that is, the dual‐attraction/repulsion mode enhances bending range and axial force transmission, and the spoon‐shaped mode facilitates compliant and safe navigation in tortuous vessels. Thermal characterization under static and flow conditions confirms safe operating temperatures under continuous energization. Furthermore, the deflection characteristics are experimentally measured under different current inputs. Steering performance is evaluated in a 90° glass channel under three actuation modes and validated through in vitro navigation experiments in a 3D heart aortic arch phantom. The results demonstrate that the DCMG achieves smooth curvature steering and controllable navigation, showing strong potential for future clinical application in vascular intervention surgeries.
Designer DNA‐based machines represent a rapidly emerging interdisciplinary field at the intersection of bionanotechnology, mechanical engineering, and computer science and has garnered increasing attentions from both academia and industry due to their transformative potential in various applications. In this article, we review the research of DNA‐based machines, covering design, analysis, fabrication, biomedical applications, and roles in nanomanufacturing; for the future developments, we further provide perspectives across the entire pipeline from design to actuation in terms of interdisciplinary and AI empowerment. We first outline the foundational principles of DNA self‐assembly and trace the evolution of DNA‐based machines from static motifs to dynamic robots. We then critically explore the integration of mechanical design into DNA nanotechnology for engineering kinematic joints, rigid‐body, and compliant mechanisms—critical components that enable the rational construction of increasingly complex DNA machines. Thirdly, we evaluate available design and analysis tools, highlighting the shift from shape‐based to motion‐based design and the critical trade‐off between multiscale simulation accuracy and computational efficiency. We also assess the laboratory‐scale fabrication methods and biomanufacturing strategies. Finally, we highlight diverse applications of DNA‐based machines, while offering future directions, including AI‐enhanced intelligent design, predictable mechanical characterization, and the transition toward autonomous, industrial‐scale fabrication. The in‐depth integration of DNA‐based machines with mechanical science, robotics, and artificial intelligence is poised to propel their evolution and expand their transformative roles in advancing precision medicine, atomic‐scale manufacturing, and information‐bionanotechnology convergence.
Magnetic soft robots have the characteristics of small size, noncable drive, and motion agility, which are suitable for medical operations in the gastrointestinal tract. However, multiangle folding and reconfigurable magnetization have yet to be fully investigated, and thus, the study of magnetic soft robots with morphological changes and medical functions is still challenging. To this end, we propose a magnetic soft sheet robot based on the magnetorheological fluids, which presents a remarkable capability for reversible folding motion, rapid real‐time reconfigurable magnetization, and targeted drug delivery functions. Furthermore, the robot has a fully soft sheet structure that is not magnetized in a zero magnetic field. After folding, the surface area of the soft sheet robot can be reduced to one third of its original area to cope with the complex gastrointestinal cavities. Five kinds of soft sheet robot prototypes with different magnetic driving abilities are fabricated, and the movement experiments of these robots are carried out on a smooth surface, a flexible fluff surface, a slope surface, underwater, and under load, respectively. The influence mechanisms of different magnetic field strengths and frequencies on robot movement are analyzed. The effectiveness of the proposed scheme is verified by ex vivo porcine stomach experiments and ultrasonic detection.
Human dexterity relies on rapid, sub-second motor adjustments, yet capturing these high-frequency dynamics remains an enduring challenge in biomechanics and robotics. Existing motion capture paradigms are compromised by a trade-off between temporal resolution and visual occlusion, failing to record the fine-grained hand motion of fast, contact-rich manipulation. Here we introduce T-800, a high-bandwidth data glove system that achieves synchronized, full-hand motion tracking at 800 Hz. By integrating a novel broadcast-based synchronization mechanism with a mechanical stress isolation architecture, our system maintains sub-frame temporal alignment across 18 distributed inertial measurement units (IMUs) during extended, vigorous movements. We demonstrate that T-800 recovers fine-grained manipulation details previously lost to temporal undersampling. Our analysis reveals that human dexterity exhibits significantly high-frequency motion energy (>100 Hz) that was fundamentally inaccessible due to the Nyquist sampling limit imposed by previous hardware constraints. To validate the system's utility for robotic manipulation, we implement a kinematic retargeting algorithm that maps T-800's high-fidelity human gestures onto dexterous robotic hand models. This demonstrates that the high-frequency motion data can be accurately translated while respecting the kinematic constraints of robotic hands, providing the rich behavioral data necessary for training robust control policies in the future.
Ionogels are polymer networks infused with ionic liquids. Ionogel actuators are thus devices that convert molecular-scale responses into flexible, reconfigurable, and stimuli-responsive macroscopic deformations or motions. Compared with hydrogels, ionogels combine low density, mechanical compliance, and large biomimetic deformation at low operating voltages with sustained ionic conductivity and mechanical and environmental stability, thereby offering significant potential for a wide range of applications. This review systematically examines the development and diversification of ionogel materials, outlining their progression from simple structural supports to advanced functional designs, as well as the synthesis and processing strategies involved. It further discusses the stimuli-responsive properties and actuation mechanisms of ionogels, together with recent representative cases under various external triggers such as chemical reactions, electromagnetic fields, temperature, humidity, light, and mechanical stress. Finally, the review highlights key actuation applications in artificial limbs, soft robotic grippers, microfluidic systems, lab-on-a-chip devices, sensors, interactive human–machine systems, as well as biomedicine. Despite significant progress, challenges remain in unclear multiscale mechanisms, limited quantitative links between material properties and device-level performance, limited synergy between materials and functions, unmet requirements for green design and biosafety, and weak system integration with closed-loop control. Overcoming these issues will broaden the applicability of ionogel actuators, facilitating their reliable integration into diverse soft device platforms and advancing the development of next-generation smart materials.
The rapid advancement of flexible electronic technology has enabled the creation of diverse innovative flexible devices, greatly facilitating the development of next-generation intelligent robots. Specifically, the integration of such advanced flexible electronics into robotic systems has significantly enhanced human–robot interaction, improved the level of intelligence of robots, and refined their operational performance. These breakthroughs span various aspects of robotics and reveal substantial application potential. In this review, we categorize flexible electronic devices based on their functional roles and systematically summarize the latest progress in the development of novel flexible electronic components. Furthermore, we conduct in-depth analyses of innovative applications of flexible electronic devices in robotic command input, intelligent decision-making, and the enhancement of manipulation performance. The review comprehensively demonstrates the considerable improvements that flexible electronics bring to intelligent robotic systems. It is hoped that the review can offer valuable insights and inspiration for the future development of flexible electronic devices and novel applications in intelligent robotics.
Achieving comparable mobility of natural creatures is always the pursuit of soft robots for better adapting dynamic environments and uncertain tasks such as field exploration, search and rescue, etc. However, most current soft robots still lag far behind natural vertebrate counterparts in both agility and speed, which mainly attributes to the limited degrees of freedom or morphing modes and performance of soft actuators. Here, we report a new methodology of electrohydraulic origami (EHO) for creating powerful and multimodal soft actuators with large strain, high speed, lightweight, flexibility, reconfigurability, and programmability. By leveraging the transmission and reconfiguration of origami structures, the simple and low-strain actuation of soft electrohydraulic actuators are enriched and amplified, and an ultra-large actuation strain (3300%) and strain rate (over 23,500% s −1 ) were achieved by the EHO actuators, as well as the high dynamic multimodal actuation (extension, rotation, and translation). We then demonstrate three types of electrohydraulic soft robots that perform high-speed bidirectional legless sliding, multidirectional jumping, and crawling based on the shape morphing and reconfiguration of EHO actuators. Moreover, untethered crawling robots were developed that achieve multidirectional crawling and higher average crawling speed than most existing soft robots driven by electro-active soft actuators. This study offers an effective strategy for creating high-performance and multimodal soft actuators and may also pave the way for electrohydraulic soft robots in real-world applications.
Humanoid robots, increasingly recognized for their potential to drive economic and social development, have garnered significant attention in recent years. This paper aims to provide a comprehensive overview of the progress, challenges, and future directions in humanoid robotics, with a particular emphasis on essential system components and key technological innovations. Through a review of historical milestones, this paper explores critical aspects such as the design of the head and body, and examines state-of-the-art technologies in areas like locomotion control, perception, and intelligent manipulation. By presenting a thorough analysis of the field, this work aims to serve as a valuable resource for researchers and inspire future innovations that will drive the continued evolution of humanoid robots.
Structural health monitoring is essential to ensure the safe operation of infrastructure such as construction machinery and bridges, which can effectively prevent accidents, extend service life, and reduce maintenance costs. However, traditional inspection methods mainly rely on manual visual inspection, which has limitations such as low efficiency, high cost, and high risk, making it difficult to meet the demand for efficient and accurate inspection in modern engineering. As an emerging technology, intelligent inspection robots provide a new solution for structural health monitoring by virtue of their autonomy, flexibility, and efficiency. This paper systematically reviews the technical characteristics of intelligent detection robots in structural health monitoring and their applications in different scenarios, with a focus on analyzing their performance in actual detection. At the same time, it summarizes their application progress in sensor technology, fault data processing and analysis methods, and fault localization technology, highlighting their ability to achieve efficient data acquisition, accurate defect identification, and real-time health assessment in complex environments. In addition, the main technical challenges faced by current inspection robots are summarized, such as stability in complex environments, data processing capability, and autonomous decision-making level. Finally, the future development direction of inspection robots is outlooked, which mainly focuses on the deep integration of artificial intelligence and machine learning, the optimization of multi-robot collaborative technology, and the improvement of lightweight and energy efficiency. Through this review, we aim to provide theoretical support and practical reference for further research and application of intelligent inspection robots in the field of structural health monitoring.
Designing soft electronic skins for tactile sensing facilitates natural human–machine communications. However, the nonlinear characteristics of electrical transductions and mechanics usually compromise precise tactile decoding and compliant interactions. In this work, we demonstrate a soft 3D-architectured pressure sensor featuring a PEDOT:PSS-PVA hydrogel lattice encapsulated within an origami-inspired elastomeric framework. Our 3D lattice sensor leverages the ultracapacitive principle to achieve wide-range linearity (0–220 kPa), fast responses, and high-resolution detection under extreme loading. The proposed 3D configuration also enables linear compression behaviors within ∼49.5% strain without sacrificing tissue-like compliance ( E = 127–404 kPa). Using this sensor as human–machine interfaces (HMIs), we facilitate accurate, timely, and stable pressure input for diverse signal waveforms in robotic teleoperation, as well as a deformable, intelligent fingertip for safely detecting soft tissue modulus. Our design provides a promising route to decode sophisticated tactile interactions by linearizing both electrical responses and mechanical behaviors.
Soft robots demonstrate remarkable potential in diverse environments because of their flexibility and compliance. Although various soft robots capable of independently responding to multiple external stimuli have been developed, challenges persist in the integration of multiple responses and the avoidance of interference among them. This study develops an amphibious soft robot with triple-response capabilities to temperature, humidity, and magnetic fields. Through the implementation of a strong alkali modification strategy on polyimide films, polyimide acid was successfully formed on the surface, enabling the processing of composite films. This composite film was endowed with dual-responsive characteristics to temperature and humidity. Additionally, by integrating magnetic particles, the composite film constructs a triple-response feature in conjunction with magnetic driving modules. The soft robot developed from the triple-responsive composite film can rapidly transition capabilities between submerged/terrestrial and submerged/surface environments, demonstrating exceptional environmental adaptability. This amphibious soft robot can achieve speeds exceeding 4 cm/s (∼12 body lengths/s) in various environments. The maximum speed attainable on the water surface is 9.6 cm/s (∼32 body lengths/s). This performance reached the normal moving speed of insects such as ants and whirligig beetles. Furthermore, by utilizing the cooperative interplay of multiple stimuli-responsive mechanisms, the soft robot achieves selective swarm manipulation, controllable cargo transportation, and targeted release in complex terrains. It can carry objects weighing up to 2.5 times its own weight. This multimodal actuation strategy reveals significant potential for smart robots' development.
The rise of embodied artificial intelligence (embodied AI) marks a pivotal shift in AI, moving it from the digital realm into the physical world. This transition aims to create autonomous robots capable of perceiving, reasoning, and acting in complex unstructured environments. Achieving this goal demands unprecedented capabilities for robots to comprehensively perceive both their external surroundings and internal states. However, traditional sensors cannot meet the requirement of robotic perception systems due to limitations in size and power consumption. In this context, micro-electromechanical system (MEMS) technology emerges as a critical enabler for advancing next-generation robotic perception capabilities. Its core advantages, including miniaturization, low power consumption, high integration, and cost-effectiveness, make it ideal for this role. This review provides a comprehensive overview of the latest advancements in MEMS sensing technologies specifically designed for embodied AI robots. By integrating diverse MEMS sensors, such as those for ranging, inertia, tactile, hearing, and olfaction, robots can achieve rich multimodal perception. These highly integrated sensing systems provide a robust technological foundation for robot applications in various fields, demonstrating the immense potential of MEMS technology in promoting autonomy, safety, and interactive capabilities in robots. In essence, the future of embodied AI will be built upon a powerful symbiosis: MEMS providing the rich semantic-aware 'sensory neurons' and AI models providing the 'cognitive brain'. This fusion promises to usher in an era of truly perceptive and intelligent machines.
Creating intelligent beings like humans is a long-standing goal in AI research, such as intelligent robots in science fiction. Classic AI technologies are disembodied, and insufficient to make robots intelligently behave in the real world. In contrast, embodied artificial intelligence (Embodied AI) enables artificial agents with physical embodiment to achieve intelligent behavior through interactions with environments. However, there are few comprehensive surveys on Embodied AI from the perspective of robot behavior within the AI domain. Thus, we provide a comprehensive survey on Embodied AI. According to the process of robot behavior, we categorize Embodied AI into three modules: embodied perception, embodied decision-making, and embodied execution. For each module, we review its tasks, methods, and challenges. We hope this survey can provide a structural framework for Embodied AI research. Besides, we also pay attention to large foundation models in Embodied AI.
