Magnetic microrobots hold immense potential for biomedical applications such as targeted drug delivery. However, achieving precise and safe navigation remains a challenge due to the complex working environments. In this paper, a safety-critical autonomous navigation framework that integrates Reinforcement Learning (RL) with Control Barrier Functions (CBFs) is proposed to enable robust obstacle avoidance for magnetic microrobots. The CBF constraint is derived to define the safe admissible control space for microrobots navigation. The constraint is embedded as a safety filter layer within a RL policy network. This architecture allows microrobots to learn efficient navigation strategies in complex environments while strictly enforcing collision-free behaviors, effectively addressing the black-box safety concerns of RL. Simulations in cluttered environments demonstrate that the proposed method achieves successful navigation with zero collisions. Furthermore, real-world experiments using a helical microrobot achieve safe autonomous navigation with zero collisions, verifying the framework's feasibility and robustness against physical uncertainties.

Electromagnetic actuation systems (EMAS) for medical microrobotics are often constrained by trade-offs among workspace size, magnetic field gradients, and field modalities. This work presents a gradient-enhanced 3-D electromagnetic actuation system (GE3-EMAS) with an expandable workspace and multimodal capability. A structure-driven design paradigm is adopted, in which magnetic flux loop topology is explicitly treated as a design variable to enhance three-dimensional magnetic field gradients without relying only on excessive hardware expansion or current amplification. Based on this paradigm, the proposed system enables controllable gradient, uniform, rotating, and scattering magnetic fields within a unified platform. In particular, a gradient-enhanced scattering magnetic field is realized by regulating a minimum magnetic energy point, producing radially distributed gradients unattainable by conventional directional superposition. A human-in-the-loop interface is integrated to demonstrate the practical feasibility of tele-operation. Simulation and experimental results validate the enhanced gradient performance, multimodal field generation, and applicability of the proposed GE3-EMAS in anatomically constrained environments.

Microrobots hold significant potential for various applications, where targeted navigation is a basic requirement. Deep reinforcement learning (DRL) has recently emerged as a powerful paradigm for fully autonomous microrobot navigation. Yet, current DRL-based approaches pay limited attention to learning efficiency and effectiveness, requiring hours to days for model training. Consequently, this impedes both rapid practical deployment and parameter optimization. To address these challenges, we present a learning framework that enables effective microrobot navigation policies to be trained within minutes. In the proposed framework, we develop a fully vectorized simulator with more than 10,000 artificial vascular environments, parallelizing dynamics, LiDAR-inspired perception, and feasibility checks across thousands of environments to achieve roughly 190,000 transitions per second. To achieve effectiveness in the fast training, we propose a task-shaping-regularization (TSR) reward framework. The TSR framework accelerates convergence, improves final performance, reduces action variation by at least 33.7

Autonomous navigation of magnetic microswarms in dynamic and unstructured environments is essential for biomedical applications, such as targeted therapy and minimally invasive interventions. However, existing path planning methods struggle to simultaneously achieve real-time adaptability and path smoothness in dynamic obstacle environments. To address this, we propose a hierarchical Dynamic Rapidly-exploring Random Tree Star (D-RRT*) path planning framework that integrates dynamic step size adjustment, local target selection, and local planning that considers microswarms' turning capabilities and energy optimization. Comparative simulations and experiments validate the effectiveness of the proposed planning framework, and results show that it can significantly improve the planning efficiency, path smoothness, and collision avoidance in complex dynamic scenarios.

Automated navigation control of microrobots in complex environments is essential for applications such as targeted drug delivery and micromanipulation. Recently, machine learning (ML) has shown great potential for automated microrobot control but still lacks accuracy and smoothness. In this work, we propose a novel Learning-from-Demonstration (LfD)-based control and navigation framework to achieve precise and smooth motion control of microrobots. This work represents an early attempt to directly utilize expert-provided data for designing a learning-based microrobot controller. The framework begins by collecting a small dataset of expert demonstrations (several thousand episodes) , from which the controller learns compensatory behaviors and task-specific adaptability, eliminating the need for extensive exploration or parameter retuning. Based on this data, a time-series neural network is then developed to process the microrobot’s historical states and control actions, allowing the system to capture sequential dependencies and transitions for smooth and accurate path tracking. For demonstration, we take the magnetic microswarm as an illustrative example. Systematic simulations and comparative experiments validate the proposed framework, demonstrating its superior performance in tracking accuracy and smoothness, validating the efficacy of ML for low-level microrobot control.

Machine learning has become an emerging paradigm for microrobotics, enabling autonomous micro-/nanorobot navigation in complex and highly disturbed environments without requirements of precise models. However, the state-of-the-art learning-based methods adopt "black-box" neural networks (NNs), which raises critical concerns about the trustworthiness of the generated navigation policies, especially for biomedical scenarios. Motivated to address this issue, we propose a physics-constrained learning framework that embeds deterministic physical laws into network architectures to achieve trustworthy microrobot navigation. Instead of learning from scratch, we explicitly encode the monotonic relationships of kinematics and safety rules into NNs. These constraints serve as structural inductive biases, theoretically guaranteeing that navigation policies operate strictly within the trustworthy action domains. Experiments demonstrate that microrobots can autonomously reach random targets without collision with obstacles during long-time and long-distance navigation, proving our framework's reliability owing to its explainable nature. This work can bridge the gap between data-driven performance and rigorous safety requirements, constituting a meaningful step toward trustworthy artificial intelligence-empowered microrobotics.

Restoring haptic feedback remains a major challenge in robot-assisted minimally invasive surgery (RMIS), especially for localizing subsurface tumors and estimating their depth. This paper presents a compact, high-density tactile sensor array based on fiber Bragg grating (FBG) sensing for robotic palpation. The array uses a honeycomb topology to increase spatial sampling within an 8.5 mm footprint while preserving high sensitivity. Static and dynamic tests show a linear force-wavelength response across seven channels, an average force resolution of 8.43 mN, and \((<)\)1% full-scale dynamic error. To estimate tumor depth from palpation signals, we propose FBG-PatchFormer. Since depth annotations are scarce and costly to obtain, FBG-PatchFormer leverages contrastive self-supervised pretraining on unlabeled palpation windows to reduce the reliance on dense labels, and is then finetuned for depth classification. On phantom palpation with embedded inclusions, FBG-PatchFormer achieves 99.60% record-level accuracy on a ten-class depth task (blank and 2--10 mm). In vivo tests on porcine liver further demonstrate robust tumor localization under physiological motion and fluid interference, supporting the clinical potential of the proposed sensing system.

Endovascular interventions require fast access to affected regions, followed by effective treatment. Catheterizations are effective approaches for treating vascular diseases; however, they face challenges in accessibility, efficiency, and invasiveness in narrow, tortuous vascular systems. This study presents a submillimeter magnetically actuated soft rotatable-tipped microcatheter (MSRM) designed to access small blood vessels and provide efficient, minimally invasive therapeutic interventions for blood clot treatment. The MSRM’s rotatable tip design enhances accessibility and navigation speed through a rotation-assisted active steering strategy. Improved blood clot treatment efficiency is achieved through the MSRM’s multifunctionality: It can accelerate drug-blood clot interactions, mechanically break down blood clots, and retrieve clot debris. The low invasiveness is attributed to the soft material design and conservative actuation strategy. The performance of the MSRM is validated in both in vitro phantom studies and in vivo rabbit models, and the invasiveness is evaluated using a human placenta model.

Actively controllable microswarms have been a rapidly developing research field with appealing characteristics. Autonomous collision-free navigation of microswarms in confined environments is suitable for various applications, including targeted therapy and delivery. However, several challenges remain unaddressed. First, microswarms possess varying dimensions, and a path planning method suitable to swarms with different dimensions is essential to avoid obstacles. Second, studies on the environment-adaptive navigation of reconfigurable microswarms are limited. Therefore, the planning of the pattern distribution of microswarms based on the local working environment should be examined. This study proposes a deep learning (DL)-based environment-adaptive navigation scheme for swarms. The controller provides reference moving directions for swarms of different sizes in static and dynamic scenarios. Moreover, a pattern-distribution planner was designed to navigate transformable swarms in unstructured environments. To validate the proposed scheme, we applied Fe3O4 nanoparticles swarms as a case study. The proposed scheme enables motion and pattern planning for microrobots of multiple sizes and reconfigurability in various working environments, which could foster a general navigation system for reconfigurable microswarms of different sizes.

Magnetic fields exhibit excellent penetrability and biocompatibility with human tissues, enabling the actuation of magnetic microscale devices for minimally invasive medical tasks. However, conventional electromagnetic coil systems are limited by small working spaces and complex wiring. In contrast, parallel mobile coil systems offer significantly larger working spaces, sufficient to cover the human torso. This paper proposes a five-axis parallel mobile coil system, referred to as PentaMag. The kinematic model of the coil-end motion for PentaMag is derived, enabling real time tracking of target objects. Additionally, magnetic field and gradient models are established, and a method for achieving three-dimensional combined head and force control is proposed. Experiments are conducted to validate the large working space and the ability to control magnetic targets for motion tasks. Future work will focus on improving visual positioning methods, avoiding control singularities, and exploring automated control strategies.

Motion control of magnetic microswarms has attracted extensive attention due to its significance in microrobots-based biomedical applications such as targeted drug delivery. However, such reconfigurable microswarms are subject to complex interactions between individuals and environments which make accurate modeling challenging. These complexities of microswarms poses challenges for precise motion control, as traditional controllers often rely on precise mathematical models and manual parameter tuning that limits their scalability and efficiency. Learning-based methods, such as Deep Reinforcement Learning (DRL), offer an alternative but require large datasets (usually on the order of millions) and extensive exploration which may cause the microswarms instability in physical environments due to unreasonable actions during early training therefore results in the sim-to-real gap. Moreover, traditional DRL focuses on instantaneous state-action mappings, neglecting the sequential dependencies critical for accurate motion control, leading to low tracking accuracy in complex scenarios. To address these challenges, we propose a Learning from Demonstration (LfD)-based motion control framework, which inherently encode compensatory behaviors and task-specific adaptability into neural networks, enabling adaptive performance even under unmodeled disturbances. Furthermore, the neural networks consider a time series of microswarm states to determine the future control actions, enabling the system to learn sequential dependencies and transitions between states so as to ensure smooth and accurate motion control. Simulations and comparative experiments validate our framework’s effectiveness and demonstrate superior control accuracy and adaptability to microswarm’s shape changes.

Inspired by bacterial motility mechanisms, Magnetic Helical Miniature Robots (MHMRs) exhibit promising applications in biomedical fields due to their efficient locomotion and compatibility with biological tissues. In this review, we systematically survey the basics of MHMRs, from propulsion mechanism, magnetization and control methods to biomedical applications, aiming to provide readers with an easily understandable overview and fundamental knowledge on implementing MHMRs. The MHMRs are actuated by rotating magnetic fields, achieving steering and rotation through magnetic torque, and converting rotation into forward motion through the helical structure. Magnetization methods for MHMRs are reviewed into three types: attaching magnets, magnetic coatings, and magnetic powder doping. Additionally, this review discusses the control methods for MHMRs, covering imaging techniques, path tracking control—including classical control algorithms and increasingly popular learning-based methods, and swarm control. Subsequently, a comprehensive survey is conducted on the biomedical applications of MHMRs in the treatment of vascular diseases, drug delivery, cell delivery, and their integration with catheters. We finally provide a perspective about future challenges in MHMR research, including enhancing functional design capabilities, developing swarm-assisted independent control mechanisms, refining in vivo imaging techniques, and ensuring robust biocompatibility for safe medical use.

Micro light-emitting diode (MicroLED) displays possess exceptional advantages including rapid response speed, autonomous light emission, high contrast, and long service life. The technology is emerging alongside rapid advancements in wearable devices, virtual reality, augmented reality, and TV displays. Due to these strengths, MicroLED displays are widely recognized as the most disruptive and revolutionary next-generation display technology. However, the miniaturized characteristic of MicroLED chips poses significant challenges for efficiently, accurately, and cost-effectively transferring millions of these chips from the donor substrate to the receiver substrate. Over the past two decades, numerous innovative mass transfer strategies have been developed. These strategies aim to overcome the limitations of traditional transfer techniques. Such advancements are driving the commercialization of MicroLED displays. Herein, we review the development of mass transfer strategies for MicroLED chips and classify these strategies into two primary categories: pick-and-place technique and fluidic self-assembly method. The former is further classified based on different adhesion modulation mechanisms, while the latter is classified based on different driving forces. Furthermore, this review provides an in-depth analysis of the working mechanisms, along with a comprehensive evaluation of the advantages and disadvantages associated with specific strategies.

Magnetic miniature robots hold great potential for minimally invasive biomedical applications due to their small size and biocompatibility. However, single-modal robots face challenges in complex environments, while existing multimodal robots that rely on elastic materials lack kinematic stability and are sensitive to environmental factors. This work introduces a novel Multimodal Magnetic Miniature Robot (MMMR) which combines helical swimming with bipedal walking. The MMMR can adaptively navigate in diverse amphibious environments, including open liquids, shallow liquids, liquid surfaces, and solid surfaces. Owing to the structure-driven locomotion mechanism, the MMMR exhibits excellent kinematic stability. The helical swimming is optimized through simulations and enhanced with a closed-loop control system for height and directional correction. Bipedal walking is designed for adaptive locomotion on liquid and non-liquid surfaces. Experiments confirmed precise multimodal control, including swimming, walking to targets, and retrieval, offering a novel solution for miniature robots in diverse environments.

Minimally invasive surgery (MIS) has become increasingly favored by both patients and surgeons owing to its advantages such as shortened recovery times and reduced surgical trauma. To enhance intraoperative feedback from surgical instruments while minimizing harmful radiation exposure, a wide range of electromagnetic tracking systems (EMTS) has been developed at micro scales for medical applications. This review provides a comprehensive summary of advances in the field over the past five years, with an emphasis on the working principles of EMTS, system architecture, current research progress, and clinical applications. In comparison to other review papers, this article focuses specifically on EMTS for medical micro-devices, such as robotic catheters, endoscopes, and capsule robots. Moreover, Representative research studies and commercial systems are presented along with their clinical implementations, placing greater emphasis on the translation of EMTS into medical applications. Finally, this review outlines and discusses future research directions, highlighting major challenges and potential opportunities for advancing the integration of EMTS into routine clinical workflows.

Magnetic microswarms hold significant potential in minimally invasive medical procedures, owing to their reconfigurability, wireless actuation, and high biocompatibility. However, conventional microswarm deformation strategies are hindered by slow response rate, making them unsuitable for rapid obstacle avoidance. Moreover, Traditional model-based control methods rely heavily on precise modeling of microscale systems, limiting their applicability in uncertain scenarios. Meanwhile, data-driven approaches face challenges such as limited hardware adaptability. To address these issues, we propose a Fast Deformation Strategy of Swarm (FDSwarm), which enables rapid shape transformation through coordinated modulation of multiple magnetic field parameters. Furthermore, we introduce Fast Deformation and Motion control based on Imitation Learning (FDM-IL), which is a framework that addresses challenges in microswarm dynamics modeling, deformation-motion coupling, and environmental adaptability. By integrating expert demonstrations with real-time environmental perception, our approach achieves the efficient co-optimization of microswarm shape adaptation and locomotion. Experimental results show that, compared with conventional deformation strategies, FDSwarm achieves a 58.0% improvement in deformation speed. In unstructured environments, FDM-IL successfully guides the microswarm to perform autonomous deformation and navigation. This work provides an robust and efficient control framework for coordinated deformation and navigation of microrobot swarms. Videos link: Supplementary videos

Monocular 3D object detection aims to identify objects’ 3D positions and poses with low hardware and computation power costs, which is crucial for scenarios like autonomous driving and deep space exploration. While the corresponding research has developed rapidly with the integration of transformer structures, features in 3D are still simply transformed from visual features, resulting in a mismatch between the detection results and the reality. Moreover, most existing methods suffer from the slow convergence speed. To address these issues in monocular 3D object detection, a framework, named geometry‐guided monocular detection with transformer (GG‐Mono), is proposed. It consists of three main components: 1) the mix‐feature encoder module that incorporates pretrained depth estimation models to enhance convergence speed and accuracy; 2) the geometry encoding module that supplements hybrid encoding with global geometry data; 3) the GG decoder module that utilizes geometry queries to guide the decoding process. Extensive experiments show that the model outperforms all existing methods in terms of detection accuracy, and achieves 26.88% and 30.65% in average precision of 3D detection box (AP 3D ) on the validation dataset and test dataset, respectively, which is 1.88% and 1.81% higher than the baseline, and significantly improved the convergence speed (from 184 to 90 epochs). These facts prove the advantages of the proposed method for monocular 3D object detection.

Micro/nanorobots have gained increasing attention worldwide owing to their promising potential in biomedicine. Benefiting from their small size and controllability, micro/nanorobots are ideal candidates for applications including targeted therapy, minimally invasive surgery, and drug delivery in physiological environments. However, the micro/nano-scale dimension hinders the ability and future application of miniature robots in the meantime. In recent years, swarm micro/nanorobotics has emerged as a rapidly developing interdisciplinary field. By simultaneously manipulating multiple micro/nanorobots, a micro/nanoswarm possesses larger delivery dose, better adaptivity to external environments, and better imaging contrast. Unlike macroscale robotic systems, implementing sensors or power supplies on micro/nanorobots is hard to achieve, which brings challenges for the control, feedback, and interagent communication of swarm micro/nanorobotics. In this review, we summarize state-of-the-art research about micro/nanoswarm, including actuation, imaging, and automatic control. Effective driving strategies and feedback methods provide the foundation for practical application. With the assistance of advanced control algorithms, micro/nanoswarms are able to exhibit computational intelligence. Compared to manual control, micro/nanoswarm systems with high-level autonomy is able to conduct bio-tasks with better efficiency and precision. Moreover, the future challenges and directions for micro/nanoswarms are discussed. With this review, we aim to provide a comprehensive understanding and valuable guidance for swarm micro/nanorobotics researchers.

Untethered microrobots possess a promising perspective for micromanipulation applications. With specifically designed morphologies and structures, microrobots are able to perform controllable delivery of target objects. However, the manipulation process still lacks autonomy, to achieve which the mechanism of picking, transporting, and releasing behaviors needs further investigation. In this article, we propose to achieve automated microrobotic manipulation using magnetic microswarms with multimodal morphology. The microswarm is composed of around 11-21 million Fe(3)O(4 )nanoparticles (1.0 -1.8 mu L particle suspension). When exposed to different dynamic magnetic fields, the swarm could exhibit corresponding forms. We realize precise and controllable cargo picking and releasing by exploiting the fluid fields of different swarm forms. In order to quantitatively describe these behaviors, we design a finite-state machine. A super-twisting sliding-mode controller has been formulated for the motion control of swarms. The disturbances are compensated via a disturbance observer. To enable automated micromanipulation in obstructed scenarios, a path planner inspired by rapidly exploring random tree algorithm is designed for path planning when obstacles exist. We also propose an enhanced-genetic algorithm to optimally transport multiple objects to the target position. Experiments demonstrate that our method could effectively transport micro-objects with different sizes and shapes. The precise selectivity of the method is validated when multiple objects exist in the working environment. Finally, the long-distance delivery ability and adaptivity to various friction situations of our strategy are demonstrated. This work explores a concise, untethered, and automated micromanipulation strategy, provides a new automatic tool for micromanipulation tasks, and extends the application potential of swarm microrobotics.