Keywords: Micro/Nano robotics, Advanced materials and sensors for robotics, Cyber-Physical bio-system
Abstract: Optical microrobots actuated by optical tweezers (OT) offer great potential for biomedical applications such as cell manipulation and microscale assembly. These tasks demand accurate three-dimensional perception to ensure precise control in complex and dynamic biological environments. However, the transparent nature of microrobots and low-contrast microscopic imaging challenge conventional deep learning methods, which also require large annotated datasets that are costly to obtain. To address these challenges, we propose a physics-informed, data-efficient framework for depth estimation of optical microrobots. Our method augments convolutional feature extraction with physics-based focus metrics—such as entropy, Laplacian of Gaussian, and gradient sharpness—calculated using an adaptive grid strategy. This approach allocates finer grids over microrobot regions and coarser grids over background areas, enhancing depth sensitivity while reducing computational complexity. We evaluate our framework on multiple microrobot types and demonstrate significant improvements over baseline models. Specifically, our approach reduces mean squared error (MSE) by over 60% and improves the coefficient of determination (R^2) across all test cases. Notably, even when trained on only 20% of the available data, our model outperforms ResNet50 trained on the full dataset, highlighting its robustness under limited data conditions.

Keywords: Biomimetic robotics
Abstract: Flow field perception in underwater environments remains a significant challenge in robotics, particularly in the identification and detection of complex flow phenomena such as piping erosion. This paper proposes a piping flow field perception method by a bionic underwater robot with lateral line systems, which achieves precise identification of complex underwater flow fields by mimicking the lateral line sensing system of stingrays. First, an underwater robot electromechanical system is designed mimicking stingrays. Second, simulation analysis of piping flow fields is conducted to develop an optimal sensor array positioning algorithm. Finally, a GRU-based piping flow field feature recognition method is developed. Experimental results demonstrate that this method effectively accomplishes flow field data acquisition, analysis, and recognition, detecting piping phenomena. This research provides a new technological pathway for underwater flow field perception and holds significant application value in hydraulic facility maintenance, effectively replacing manual operations.

Keywords: Micro/Nano robotics, Bio-Cell assembly and tissue fabrication, Biosensors and bioactuators
Abstract: Magnetic helical microswimmers represent a promising technology for biomedical applications including targeted drug delivery and minimally invasive surgery. However, autonomous navigation in dynamic and unstructured environments characteristic of biological systems remains a critical challenge. Traditional path planning algorithms struggle with real-time adaptation to unpredictable environmental changes and dynamic obstacles. This paper presents a novel Deep Reinforcement Learning approach for autonomous obstacle avoidance in magnetic helical microswimmers. We formulate the navigation problem as a Markov Decision Process and develop a physics-based virtual environment that accurately simulates microswimmer dynamics and obstacle interactions. A Deep Reinforcement Learning agent is trained using Proximal Policy Optimization with domain randomization to enhance policy robustness and generalization across diverse conditions. Comprehensive evaluation through simulations and real-world experiments demonstrates superior performance across static, fluctuating, and flowing obstacle scenarios. Our Deep Reinforcement Learning-based controller achieves navigation success rates of 75% in environments with 30 static obstacles and 60% with 30 dynamic obstacles, significantly outperforming traditional navigation methods in both success rate and path efficiency. The results validate the approach’s potential for practical biomedical applications requiring autonomous microrobot navigation in complex in vivo environments.

Keywords: Other related topics
Abstract: A promising effective human-robot interaction in assistive robotic systems is gaze-based control. However, current gaze-based assistive systems mainly help users with basic grasping actions, offering limited support. Moreover, the restricted intent recognition capability constrains the assistive system’s ability to provide diverse assistance functions. In this paper, we propose an open implicit intention recognition framework powered by Large Language Model (LLM) and Vision Foundation Model (VFM), which can process gaze input and recognize user intents that are not confined to predefined or specific scenarios. Furthermore, we implement a gaze-driven LLM-enhanced assistive robot system (MindEye-OmniAssist) that recognizes user’s intentions through gaze and assists in completing task. To achieve this, the system utilizes open vocabulary object detector, intention recognition network and LLM to infer their full intentions. By integrating eye movement feedback and LLM, it generates action sequences to assist the user in completing tasks. Real-world experiments have been conducted for assistive tasks, and the system achieved an overall success rate of 41/55 across various undefined tasks. Preliminary results show that the proposed method holds the potential to provide a more user-friendly humancomputer interaction interface and significantly enhance the versatility and effectiveness of assistive systems by supporting more complex and diverse task.

Keywords: Cyborg intelligence, Neuro-Control and communication, Neural-machine interface
Abstract: Robotic systems often encounter difficulties in distinguishing between instances of the same category when instructed via natural language, particularly under attribute-based constraints such as size, color, or spatial position. In this paper, we propose a novel framework that integrates open-vocabulary instance segmentation with language-driven reasoning to accurately identify the object referenced by attribute-rich user instructions—for example, "the largest red box" or "the second cup from the left". Our method first generates multiple candidate object masks from a static RGB image and assigns each instance a unique visual label. These labeled instances, together with the language command, are then processed by a large vision-language model to resolve the correct target. Unlike prior approaches, our framework does not require explicit pretraining for visual-language alignment and demonstrates strong generalization to unseen attribute combinations. We validate the effectiveness of our approach through robotic grasping experiments, which confirm its robustness in cluttered scenes containing multiple visually similar objects.

Keywords: Cyber-Physical bio-system, Cyborg intelligence, Neuro-Control and communication
Abstract: In order to investigate the response limitations of human skeletal muscles in generating movement bypassing the bottleneck of neural delays, we develop a closed-loop electrical stimulation system for elbow joint control using high-speed visual feedback. The proposed system integrates high-speed visual sensing (up to 1,000 Hz), high-frequency stimulation generation (up to 1,000 Hz), and a fuzzy logic-based control algorithm. It enables closed-loop feedback control of one-degree-of-freedom elbow motion through biceps and triceps stimulation. As an early stage study, experiments with three healthy subjects were conducted using the developed system for tracking a dynamic target, with tracking performance evaluated via mean absolute error. Preliminary results reveal the tendency that tracking performance improves with higher feedback frequencies.

Keywords: Micro/Nano robotics, Bio-Cell assembly and tissue fabrication, Biosensors and bioactuators
Abstract: 磁性微型机器人引起了相当大的关注 在生物医学应用中，由于其独特的 优势。然而，传统的磁控制系统 生成全局统一字段，限制独立 控制多个微型机器人，因为固有的 耦合效应。这种约束阻碍了它们在 需要协调复杂任务的情况。在 本文介绍了一种新型的局部磁场 系统采用电磁铁阵列，实现独立 多微型机器人控制并克服 均匀的现场方法。通过系统性表现 电磁铁的分析和优化，以及 开发单独的电磁场模型和 六边形电磁铁阵列平台的构建， 我们显着增强了磁场的产生。 此外，改进了 A* 算法 实现微型机器人移动

Keywords: Bio-Cell assembly and tissue fabrication, Regenerative medicine
Abstract: The in vitro reconstruction of an authentic liver tumor microenvironment holds significant promise for drug screening and clinical research. However, current 3D tumor models, such as those cultured in U96-well plates or via bioprinting, exhibit limitations in recapitulating dynamic cell-matrix interactions. This study presents a stepwise fabrication approach: liver tumor cells are first cultured via microcontact printing to form spheroids, followed by digital micromirror device-based microfluidic channel photopolymerization to encapsulate the spheroids within a tumor-stromal compartment, thereby constructing a physiologically relevant liver tumor microenvironment. This model enables systematic investigation of tumor-stroma crosstalk and high-throughput anticancer drug screening, providing a controllable in vitro platform for precision oncology and metastatic mechanism analysis.

Keywords: Micro/Nano robotics, Biomimetic robotics
Abstract: Untethered microrobots hold significant potential for navigating confined spaces and performing complex tasks, such as targeted drug delivery and environmental sensing, owing to their compact size and wireless operation. However, current magnetically actuated soft microrobots still face multiple technical challenges in complex, dynamic liquid environments. Their limited degrees of freedom and inferior flexibility and maneuverability compared to natural aquatic organisms, such as fish, restrict their ability to swim effectively in intricate hydrodynamic conditions. In this work, we present a soft microrobotic swimmer capable of multimodal locomotion, biomimicking both tadpole-like undulatory and sperm-like helical propulsion. This dual-mode actuation enables efficient navigation through diverse challenging environments, including simulated vasculature, narrow channels, and countercurrent flows. Furthermore, we systematically optimize the robot’s structural design by quantitatively investigating key parameters, such as the head-to-tail ratio, thickness profile, and magnetization orientation, to achieve superior swimming performance. Our findings provide insights into the development of advanced untethered microrobots for biomedical and environmental applications.

Keywords: Micro/Nano robotics, Biomimetic robotics, Medical surgical robotics
Abstract: Magnetically actuated microrobots have shown great promise in medical diagnosis and therapeutic applications due to their miniature size and non-invasive controllability. However, conventional rigid microrobots face significant challenges in navigating natural orifices, exhibiting limited adaptability and struggling to perform complex biomedical tasks. Flexible magnetic microrobots, while offering better compliance with biological environments, often suffer from poor environmental adaptability, limited controllability in multimodal locomotion and difficulties in integrating actuation and functional operations. To address these limitations, this study proposes a novel design strategy combining heterogeneous magnetization and soft-material fabrication techniques to develop two functional microrobots: a quadruped crawling microrobot and a starfish-inspired grasping microrobot. Under the control of a triaxial Helmholtz coil system that generates programmable dynamic magnetic fields, the microrobots demonstrate versatile locomotion modes, including rolling, and crawling. A real-time vision-based servo system is integrated to monitor the microrobot's motion and enable autonomous decision-making at branched pathways via predefined motion judgment points. Furthermore, the proposed system supports adaptive mode transitions in response to complex and unstructured environments, paving the way for intelligent operation of flexible microrobots in biomedical scenarios.

Keywords: Micro/Nano robotics, Bio-Cell assembly and tissue fabrication, Cyborg intelligence
Abstract: Cellular microbots with autonomous propulsion and navigation, a frontier in biomedical engineering, leverage their autologous origin and intrinsic cytotoxicity to evade immune clearance and accumulate efficiently in tumor regions for therapeutic action; However, functional inactivation caused by destructive effects of conventional fabrication methods often leads to premature loss of activity during ex vivo preparation or in vivo circulation. To address this bottleneck , we report a CD5+ dendritic cell based microbot (DC microbot) constructed by engineering natural CD5+ dendritic cells to phagocytose magnetic nanoparticles coated with tumor cell membranes and loaded with drugs, where cancer cell membrane camouflage enhances phagocytosis efficiency and specifically activates the immune-stimulatory functions of CD5⁺ dendritic cells. Under external rotating magnetic fields, DC microbots exhibit directed collective movement, precisely aggregating at intestinal target sites; they then navigate along tumor chemokine gradients,penetrating the extracellular matrix barrier via positive chemotaxis to infiltrate deep into tumor tissues and trigger cascading activation of resident immune cells. Distinct from traditional cellular microbot therapies that rely on self-mediated cytotoxicity, this dual-responsive DC microbot system activates durable antitumor immunity by reprogramming innate immune cells within the tumor microenvironment. Meanwhile, it preserves the biological functions of natural dendritic cells, thereby overcoming the premature inactivation typically observed in ex vivo engineered immune cells during in vivo applications. This strategy offers a novel pathway toward precision-targeted therapy with excellent biocompatibility and functional stability.

Keywords: Advanced materials and sensors for robotics, Micro/Nano robotics, Medical surgical robotics
Abstract: Small-scale continuum robots hold promise for interventional diagnosis and treatment, yet existing models struggle to achieve small size, precise steering, and visualized functional treatment simultaneously, termed an “impossible trinity”. This study introduces an optical fiber-based continuum robot integrated imaging, high-precision motion, and multifunctional operation abilities at submillimeter-scale. With a slim profile of 0.95 mm achieved by microscale 3D printing and magnetic spray, this continuum robot delivers competitive imaging performance and extends obstacle detection distance up to ~9.4 mm, a tenfold improvement from the theoretical limits. Besides, the robot showcases remarkable motion precision (less than 30 μm) and substantially widens the imaging region by ~25 times the inherent view. Through ex vivo trials, we validate the robot’s practicality in navigating constrained channels, such as the lung end bronchus, and executing multifunctional operations including sampling, drug delivery, and laser ablation. The proposed submillimeter continuum robot marks a significant advancement in developing biomedical robots, unlocking numerous potential applications in biomedical engineering.

Keywords: Micro/Nano robotics, Other related topics
Abstract: Cancer has become the second major health threat endangering human survival and represents a core challenge requiring urgent resolution in global public health. Immunotherapy is a novel therapeutic strategy that combats cancer by modulating the immune system. Although several immunotherapeutic agents have been clinically applied for cancer treatment, the presence of "cold" tumors in some patients leads to low response rates to immunotherapy. Potential off-target effects and other adverse events remain significant challenges in tumor immunotherapy. In this study, to address the limitations of immunotherapy for lymphoma, a highly heterogeneous and aggressive cold tumor, we designed chemokine (C-C motif) ligand 5 (CCL5)-loaded macrophage-based microrobots to remodel the tumor microenvironment. This strategy synergistically integrates the inherent phagocytic functionality and tumor-tropic properties of macrophages with the biosafety advantages, tetherless operational capacity, and spatiotemporal precision of magnetically navigated platforms. CCL5 mediates immune cell chemotaxis and enhances tumor cell infiltration and metastatic potential. We conjugated cytokines to the surface of magnetic nanoparticles using the carbodiimide method, achieving excellent biocompatibility and magnetic responsiveness. The transwell assay confirmed the engineered macrophages’ innate tumor tropism and magnetic field-guided targeting capability toward EL-4 cells. These magnetically responsive macrophages demonstrate potential for advancing safer and more effective clinical translation of immunotherapies. Our strategy facilitates immune cell reprogramming and recruitment, establishing a novel platform for tumor immunotherapy.

Keywords: Biomimetic robotics
Abstract: Constrained by their size and limited actuation, small-sized quadrupedal robots often struggle with mobility on uneven terrain. To address this challenge, we investigate the use of spinal flexion as a compensatory mechanism to enhance platform-climbing ability using a rat-like robotic platform. During such tasks, the posture of a rat-like robot is influenced by environmental conditions, limb motion, and spinal flexion. We first model the relationship among these components by analyzing the robot's kinematics affected by spinal flexion. Building on this kinematics model, we employ spinal flexion to increase hind leg lifting height, enabling the robot to overcome higher platforms. Experimental results show that the robot successfully climbs a platform reaching 83% of its own height with spinal flexion, demonstrating that lateral spine flexion improves the robot's ability to overcome higher obstacles.

Keywords: Wearable robotics, Biomimetic robotics, Prosthesis and exoskeleton robotics
Abstract: This work presents a soft hybrid finger that combines principles of bio-inspired design with compliant mechanical structures, aiming to enhance adaptability and dexterity in robotic grasping. The proposed solution consists of a soft- continuous finger, inspired by the Fin Ray effect, with embedded flexible living hinge joints. Finite element simulations were used to explore and inform the structural design with a parameter search. The final prototype was 3D-printed with compliant materials, resulting in a lightweight and adaptable finger. The finger is underactuated, with all three joints connected to a single servo motor. The DIP and PIP joints were tendon-driven, while the MCP joint is part of a four-bar mechanism linked to the motor. A preliminary evaluation was conducted using two soft hybrid fingers arranged in a parallel gripper configuration, and comparing them with conventional Fin Ray effect fingers. Kinematic data were collected while grasping nine different objects positioned at various heights, with the objective of evaluating joint behavior and deformation patterns under differing contact and force scenarios.

Keywords: Biomimetic robotics
Abstract: The spine of quadruped animals plays an essential role during movement. We used biological quadrupeds as our bionic reference. We developed a flexible biomimetic quadruped robot. Our design considers the spine’s bending motions, compliance, and connective functions. Our design features a two-degree-of-freedom (2-DoF) continuous flexible spine structure. For the leg mechanism, we implemented an anti-parallel four-bar linkage with parallel elastic components to significantly improve the robot's mobility, flexibility, and compliance. A functional prototype and corresponding testing platform were constructed for experimental validation. Test results demonstrate the spine's bidirectional bending capability, with measured improvements of 2.0 times in horizontal foot-end motion range and 0.91 times in vertical range. Locomotion and steering experiments successfully verified the spine's turning functionality and biomimetic performance.

Keywords: Micro/Nano robotics, Bio-Cell assembly and tissue fabrication
Abstract: Efficient manipulation of microscale fluids in open environments remains a significant challenge due to dominant viscous forces, risks of evaporation, and limited integration with robotic systems, all of which impede rapid mass transfer critical for applications such as diagnostics, drug discovery, and nanomaterial synthesis. Conventional approaches, including enclosed microfluidic channels or macroscopic robotic handling, often restrict sample accessibility and lack the precision required for sophisticated fluid control at the microscale. This study introduces a novel acoustic-actuated robotic end-effector designed for versatile open-environment microfluidics manipulation. By employing acoustically driven microbubble oscillations within a micropipette tip, the system generates intense, localized micro-vortices streaming that overcome the constraints of low Reynolds number flows. Through detailed flow analysis and experimental validation with high-viscosity liquids and nanoparticle dispersions, we have confirmed a significant enhancement in mass transfer efficiency. The system provides precise control over fluid dynamics, governed directly by the input voltage. Its open environment ensures easy access for sampling, broad compatibility with standard optical microscopy, and a low risk of contamination. These characteristics establish it as an accurate, cost-effective, and adaptable platform for applications like droplet assays, nanomaterial preparation, and the regulation of cellular environments.

Keywords: Micro/Nano robotics, Biomimetic robotics, Medical surgical robotics
Abstract: Abstract—The efficacy of endovascular interventions is often significantly influenced by procedure time, which is closely tied to the surgeon’s experience and the operational controllability of the endovascular devices. In recent years, millimeter scale flexible continuum robots have garnered increasing attention as promising alternatives aimed at reducing the uncertainty associated with manual manipulation. However, most existing continuum robots are solid structure and hindered by microscale fabrication, navigation in blood vessels, and operational efficiency during multivascular interventions. To address these limitations, we present a magnetic submillimeter dual-channel flexible continuum robot with hollow structures that enable bimodal applications for endovascular procedures. The robot features dual-channel hollow structures (outer diameter: 300 μm; inner diameter: 80 μm), with each channel independently actuated via a rotating gradient magnetic field (workspace: 180°). Facile fabrication is achieved via an efficient singlestep thermal forming process with parameters optimized by the OptiLOWESS-BO algorithm, ensuring high structural consistency and patient-specific dimensions for precise therapy. In vitro demonstration verifies the robot’s capabilities in vascular navigation, multivascular intervention, targeted drug delivery, and liquid biopsy within vascular phantoms. The maximum speed of the robot’s tip is 35 mm/s (~174.4 mm(Channel 1)/313.6 mm(Channel 2) × 300 μm), showcasing its potential for multivascular interventions. The proposed dual-channel microscale continuum robot system introduces a breakthrough design and fabrication approach for multivascular interventions, potentially accelerating the clinical translation of submillimeter continuum

Keywords: Biomimetic robotics
Abstract: Soft actuators, as the core components of soft robots, play a crucial role in the entire robot system. However, due to the limited output force and response speed of soft actuators based on smart materials, soft robots have difficulty achieving effective movement in amphibious scenarios. Inspired by the hydraulic drive mechanism of leaping spiders, a flexible and controllable amphibious robot module is developed by designing a new type of joint actuator and introducing an efficient underwater encapsulation method. This new type of joint electro-hydraulic actuator can achieve a large swinging angle and a fast dynamic response speed. In addition, the crawling robot developed by using this joint actuator can not only realize fast crawling speed (3.3 BL/s) but also perform climbing motion at an angle of 30°. Moreover, it can be applied underwater and achieve rapid propulsion by splashing water, demonstrating excellent amphibious motion performance. This work not only contributes to the development of high-performance amphibious soft actuators but also provides essential references for the design of soft robots.

Keywords: Prosthesis and exoskeleton robotics, Wearable robotics, Rehabilitation robotics
Abstract: Assistive robotic devices, such as ankle-foot orthoses (AFOs), have shown great potential in enhancing mobility. Conventionally, optimization of these devices has focused on single-objective approaches, primarily minimizing metabolic costs. However, this strategy often overlooks critical metrics related to user adaptation. Achieving optimal assistance requires balancing energy efficiency with the user’s ability to adapt to the device. This study investigates relationships between metabolic cost, symmetry, and user adaptation metrics to inform a multiobjective optimization strategy. Through controlled human subject experiments with a robotic ankle-foot orthosis, we examined the relationship between gait symmetry and user adaptation. Our results indicate that gait symmetry can serve as a promising indicator of effective device usage and user adaptation. A heart rate metric, root mean square of successive differences (RMSSD) exhibits a strong correlation with both metabolic cost and gait symmetry, suggesting its potential as a proxy for adaptation. Perceived effort and comfort are influenced by both metabolic cost and gait symmetry, while effort-tocompress (ETC) shows minimal correlation. These findings suggest adaptation metrics in the optimization of robotic AFOs to enhance both energy efficiency and user adaptability.

Keywords: Micro/Nano robotics
Abstract: This paper focuses on precisely manipulating microparticle motion using an optoelectronic tweezer system. By developing a parameterized pattern-interaction framework, a series of experiments were conducted, demonstrating multi-modal trajectory generation for a single particle, circular rotation, and parallel rotation of multiple particles within microgrooves (single or double particles per groove), as well as stepwise "rotation–translation" control. Leveraging light-induced dielectrophoresis and a non-uniform electric field gradient designed via an internal gear-shaped optical pattern, the system enables multi-parameter control of polystyrene microparticle behavior, including trajectory shape, rotational angular velocity, and coordinated motion among multiple particles. The results verify the system’s capability in regulating complex multi-particle dynamics, expanding the application scope of OET-based micro/nano manipulation and providing technical support for biomedical particle handling and micro/nano-assembly.

Keywords: Cyber-Physical bio-system, Advanced materials and sensors for robotics
Abstract: Physiological barriers in the human body, such as the skin, lungs, intestines, and blood-brain barrier, effectively protect against the invasion of pathogenic microorganisms and harmful substances, thereby maintaining a stable and healthy internal microenvironment. Recent studies have focused on simulating the functions of these barriers by developing porous membranes with tunable pore structures for barrier simulation and related applications. Although polydimethylsiloxane (PDMS) is commonly used in the fabrication of porous membranes due to its excellent transparency, flexibility, and biocompatibility, existing methods for producing PDMS-based porous membranes are often limited by expensive equipment and complex processes. To address the challenges in current membrane fabrication techniques, we propose a method for manufacturing PDMS porous membranes on the water surface. This involves the addition of a mixture containing PDMS solution, curing agent, n-hexane solvent, and polystyrene (PS) microspheres onto the water surface to form a thin film, which is then treated with ethyl acetate to dissolve the PS microspheres, resulting in a PDMS porous membrane. The proposed method does not require expensive equipment and is simple to implement; furthermore, its excellent biocompatibility provides effective support for in vitro physiological barrier simulation and related research.

Keywords: Cyber-Physical bio-system, Neuro-Control and communication, Other related topics
Abstract: Haptic display technologies are diverse, encompassing fixed, movable, portable, wearable, and encounter-based systems, each capable of rendering a subset of known haptic modalities. In this paper, we introduce a new category: implantable haptics. This concept envisions haptic display systems as permanent, biocompatible implants integrated into the human body, leveraging existing tactile sensory pathways. Implantable haptics are designed to deliver haptic feedback or enhance perceptual capabilities by interfacing directly with nearby nerve endings. We explore this concept specifically through the use of subdermal magnetic vibrator implants and demonstrate how such implants can be used to convey haptic information or transduce non-haptic data into tactile sensations.

Keywords: Biosensors and bioactuators, Neural-machine interface, Advanced materials and sensors for robotics
Abstract: The quality of electrophysiological acquisition highly depends on the performance of electrodes, which are also considered as a type of biosensor. Traditional wet electrodes have limits such as gel drying and skin irritation, while flat dry electrodes are easily interfered by high interface impedance and motion artifacts under dynamic conditions. Microneedle array electrode (MAE) is a type of dry electrode, where the microneedle structure can penetrate the stratum corneum of skin and establish a stable electrical contact with the highly conductive epidermal layer for signal recording. This study proposes a type of silicon-based MAE for high-quality electrophysiological acquisition. By evaluation of the electrode-skin interface impedance (EII) at six body parts (forearm, upper arm, lower leg, thigh, back, and waist) under three conditions (rest, walking, and cycling), it is found that MAE shows advantages in the whole testing frequency range of 20-2000 Hz, where the low-frequency impedance is reduced by more than 90% and the impedance curve is much smoother under dynamic conditions compared to the flat dry electrode. For surface electromyography (sEMG) recording, MAE exhibits signal-to-noise ratio (SNR) as high as the commercially available wet electrode, demonstrating the stable electrode-skin interface established by the microneedles. For sEMG-based dynamic muscle fatigue assessment (a task of 80-second bicep curl) by MAE, the attenuation trend of median frequency (MDF) is highly consistent with the result by wet electrodes, and the frequency domain change rate (ΔMDF) also has no significant difference, which confirms the feasibility of using MAE to collect high-quality sEMG for clinical applications.

Keywords: Neural-machine interface, Rehabilitation robotics
Abstract: Electroencephalography (EEG) signals provide neural information for brain-computer interfaces and clinical monitoring applications, offering high temporal resolution for brain-state assessment and motor intent recognition. However, these signals frequently suffer from missing data due to electrode displacement and wireless transmission failures, compromising brain-computer interface reliability. Traditional signal completion methods fail to capture complex spatiotemporal dynamics, while existing tensor completion approaches require offline processing and cannot handle 2D EEG streams.We propose a sliding-window EEG completion framework integrating Variational Mode Decomposition (VMD) and High Accuracy Low Rank Tensor Completion (HaLRTC). The method transforms streaming 2D EEG data into 3D tensors through VMD decomposition, enabling online tensor completion across spatial, temporal, and spectral domains.Experimental evaluation demonstrates 13-54% RMSE reduction and SNR values of 4.3-7.5 dB compared to conventional methods. This advancement enables practical tensor completion for neural signal processing, addressing critical barriers to EEG system reliability.

Keywords: Cyborg intelligence, Biomimetic robotics, Cyber-Physical bio-system
Abstract: Advanced spatial understanding and decision making are critical for medical robots, particularly for quadrupeds in hazardous rescue scenarios. However, Large Language Models (LLMs) cannot effectively reason over raw sensor data due to its lack of semantic structure. To address this, we propose an intelligent navigation and control system for quadruped medical robots, centered on a cognitive map. Our system constructs a sparse, structured map that integrates multi-level topology, multimodal sensor data, and inter-object relationships. This approach bridges low-level perception with high-level LLM reasoning. Experiments in simulated rescue environments demonstrate that our framework significantly improves path optimization, task completion rates, and causal reasoning accuracy. These results validate the system’s ability to enhance the cognitive capabilities of LLM-driven medical robots in complex scenarios

Keywords: Biomimetic robotics, Micro/Nano robotics
Abstract: Gait control in soft crawling robots is commonly based on manually designed control rules, which often results in limited adaptability, high parameter dependency, and unsmooth transitions between locomotion modes. These limitations hinder the robot’s ability to operate effectively in unstructured or dynamic environments. To address this, we propose a biologically inspired gait control strategy for the designed soft crawling robot, based on a central pattern generator (CPG) network composed of Hopfield oscillators. Each oscillator is mapped to an actuator of the soft robot. The CPG network autonomously generates rhythmic control signals, which are transformed into pulse-width modulation (PWM) signals through a discretization method. Experimental validation is conducted across multiple locomotion scenarios including forward crawling, turning, and rolling. The results show that the gait controller enables fast convergence from arbitrary initial states, smooth gait transitions through parameter modulation, demonstrating a scalable and effective solution for generating adaptive and continuous locomotion patterns in soft crawling robots.

Keywords: Medical surgical robotics, Advanced materials and sensors for robotics, Other related topics
Abstract: Magnetic resonance (MR)-guided robotic systems are promising tools for high-precision neurosurgical interventions, leveraging real-time imaging and enhanced soft tissue contrast. Recently, a number of MR-compatible hydraulic actuators have been developed. However, accurate linear position encoding remains a significant challenge due to constraints in structural space and limited availability of miniaturized sensors operating in MR environment. In this study, we present an MR-safe optical linear encoder comprising a spring substrate integrated with an optical fiber embedded with distributed fiber Bragg grating (FBG) sensors. A static mechanical model is established to relate the actuator’s linear displacement to the corresponding wavelength shifts in the FBG signals. Furthermore, a learning-based decoupling algorithm is introduced to estimate the actuator's linear position from the multiplexed FBG responses. The device is experimentally validated in a 1.5 Tesla MR environment, demonstrating both superior MR compatibility and high sensing accuracy. The proposed sensor has a compact form factor using fully optical design, making it well-suited for MR-safe robotic applications.

Keywords: Micro/Nano robotics, Biomimetic robotics, Medical surgical robotics
Abstract: Magnetic flexible robots show great promise in biomedical applications due to their shape reconfigurability, locomotion capability, and multifunctionality. For oral administration, miniaturizing these robots is essential. However, existing designs, such as origami robots, often face challenges like incomplete or slow unfolding transitions. To address this, we propose a magnetic multilayer flexible robot for gastric applications. Our design is inspired by the morphological and mechanical structures of petals during flower blooming. In the unfolded state, the robot is a petal-inspired four-layer patch with Polydimethylsiloxane (PDMS) edge reinforcement (≈1 mm thick, ≈1052 mm² area), magnetized along the z-axis so that its top and bottom surfaces bear opposite polarities. In the folded state, the robot assumes a bellflower-bud morphology, with petals arranged side by side without overlap, and opposing petals carrying identical magnetic polarities. The folded robot can be encapsulated in a No. 000 capsule (≈25.6 mm length, ≈9.8 mm diameter) and guided by an external magnetic source to the target location in the stomach. The rapid and successful unfolding is driven by three synergistic forces: asymmetric elastic recovery from its multilayer structure, edge elasticity from its PDMS cladding, and magnetic repulsion between the petals. Once deployed, the patch can be manipulated to adhere to the gastric wall, acting as a physical barrier for mucosal wound protection and delivering sustained drug release from its drug-loading layer.

Keywords: Medical surgical robotics, Micro/Nano robotics, Advanced materials and sensors for robotics
Abstract: This study systematically quantifies and compares manipulation forces during magnetic and passive guidewire navigation in a vascular model. Results show that magnetic guidewires significantly reduce both translational and torsional forces, especially above safety thresholds, suggesting improved safety and usability for minimally invasive vascular interventions.

Keywords: Cyborg intelligence, Cyber-Physical bio-system, Neuro-Control and communication
Abstract: Strabismus represents a frequently occurring visual disorder, for which gaze training constitutes a key component of the rehabilitation process. To enhance the effectiveness of gaze training for strabismus patients, we propose a novel deep learning model for salient object detection. The proposed model is built upon a Transformer architecture with prompt tuning, enabling flexible adaptation to diverse gaze training tasks. We introduce a covariance tune that innovatively captures second-order interactions among patch embeddings through covariance operations, thereby strengthening the model’s capability in modeling both local and global relationships within images. In addition, the covariance features are integrated with embedding tune to construct a second-order adaptor, which is combined with the Transformer to form Second-order Prompt Tuning Transformer. Experimental results demonstrate that our method achieves superior performance on several public salient object detection datasets and exhibits enhanced applicability in gaze training scenarios for strabismus. The proposed approach provides a new technological pathway for personalized rehabilitation training of strabismus patients.

Keywords: Micro/Nano robotics
Abstract: 显微图像的深度信息对于 细胞识别和作等生物学任务。然而，目前的深度估计方法大多依赖于 关于影像系统的硬件升级，即 昂贵、复杂且概括性差。解决 上述问题，本研究采用DINOv2，一个大型 视觉模型在提取方面特别有效 细粒度图像特征，开发专用 深度估计的分类和回归方法 细胞图像。首先，使用我们定制的自动成像 系统中，我们收集了 2,713 张多焦点细胞图像，以 构建微调数据集。然后我们设计了 特定于任务的模型架构。实现的模型 在分类和回归方面均具有优异的性能 任务，实验结果在细胞中准确率为 98% 状态

Keywords: Micro/Nano robotics, Cyborg intelligence, Advanced materials and sensors for robotics
Abstract: Existing studies have demonstrated that the introduction of nanomaterials into carbon-based actuators can significantly enhance their photothermal conversion efficiency. However, excessively high photothermal conversion efficiency makes actuators prone to thermal damage under intense laser irradiation. To address this issue, this study constructs a vision-based dynamic monitoring platform to achieve real-time identification of the actuator's operational status and early warning of damage under photoexcitation. Actuators with enhanced photothermal conversion efficiency are susceptible to structural burnout due to localized heat accumulation under high-power laser exposure, typically characterized by abnormal high-brightness patches in the imaging area. The developed vision monitoring system employs image processing algorithms and machine learning models to conduct real-time analysis of photofield distribution characteristics and structural deformation patterns, enabling dynamic evaluation of actuator performance and prospective identification of abnormal states. Experimental data show that this monitoring platform can reliably capture deviations in operational parameters and effectively reduce the risk of laser thermal damage through a closed-loop feedback mechanism. This research integrates efficient light-driven actuation mechanisms with intelligent monitoring technologies, providing high-performance and high-reliability solutions for advanced soft robotics, biomedical devices, and precision optical manipulation applications.

Keywords: Biomimetic robotics, Advanced materials and sensors for robotics
Abstract: The use of compliant tensegrity grids offers significant potential for adaptive and lightweight systems for applications in soft robotics. This paper presents foundational investigations into a new class of auxetic tensegrity metamaterials. This work demonstrates that the auxetic behavior of a tensegrity grid can be realized and tuned solely through the mechanical properties of tensioned members, without altering the equilibrium geometry, offering a key advantage over conventional auxetic structures with fixed topologies. To validate the concept, 3D-printed demonstrators are manufactured, tested and a possible application as a cylindrically shaped soft robotic skin is shown.

Keywords: Rehabilitation robotics, Medical surgical robotics
Abstract: Complete and uniform laser irradiation of irregular skin lesions is essential for effective dermatological therapies, including photothermal treatment and photodynamic therapy. Conventional scanning patterns like raster and spiral fail to recognize sensitive "no-treat" zones such as nevi and blood vessels, causing under- or over-treatment. This paper presents an obstacle-aware cooperative path planning framework based on Heuristic Path Construction(HPC) for multi-beam laser systems to fully cover complex skin surfaces. The lesion area and forbidden regions are first reconstructed into a discretized 2D grid using medical imaging. Then, we formulate the cooperative coverage problem under mechanical turn-rate limits of galvanometric scanners and thermal accumulation constraints. A distributed heuristic is applied to directly plan paths for each laser head without prior subregion allocation, minimizing overlap and idle travel. Simulations show our method achieves over 98% coverage, and shortens treatment time by 20% compared to baseline scanning patterns. This statistically validated approach provides a practical solution for precise, efficient, and safe laser-based skin therapies.

Keywords: Advanced materials and sensors for robotics
Abstract: To address the requirements of wire damage detection during data center operations and maintenance, this paper designs a detection method based on an intelligent inspection robot and proposes an improved YOLOv8-BE algorithm. A custom dataset was constructed by capturing images of damaged wires with an onboard camera, and experimental validation was conducted based on this dataset. This improved method builds on YOLOv8 by incorporating a BiFPN feature fusion network and an EMA attention mechanism to enhance multi-scale feature interaction and focus on key damaged areas. Experimental results demonstrate that this method performs well in complex background and small object detection scenarios, achieving 93.5% accuracy, 91.6% recall, and 94.5% mAP@50. Furthermore, this model was deployed on an intelligent inspection robot hardware platform to verify its feasibility and robustness in a real-world environment. These results provide an efficient and reliable solution for wire damage detection in complex scenarios such as data centers.

Keywords: Advanced materials and sensors for robotics
Abstract: 解决航点过多和更长的问题 传统跳点搜索 （JPS） 中的路径长度 机器人路径规划算法，介绍 模拟退火（SA）算法，并提出了一种 综合方法，双向的集成 交替使用 JPS 和 SA 算法 （SA-JPS），以确保 全局路径最优。SA-JPS 算法优化了 跳点搜索策略，细化路径代价函数， 并结合随机扰动以消除 冗余点并减少路径长度。该算法 利用双向交替 JPS 算法来 在 结构化网格地图，从而通过 双向搜索策略。随机的介绍 扰动和模拟的退火策略 退火算法有助于最大限度地减少冗余点 路径并防止其收敛到局部最优。 实验结果表明，与 传统的JPS算法，所

Keywords: Micro/Nano robotics, Bio-Cell assembly and tissue fabrication, Biosensors and bioactuators
Abstract: High-precision three-dimensional imaging of live embryos is hindered by photobleaching, invasive labels, and the mechanical constraints of conventional rotation methods. To overcome these limitations, we introduce a fully integrated, noncontact acoustofluidic platform for controllable embryo rotation and reconstruction. A piezoelectric transducer drives microbubble resonance at the tip of a glass micropipette, generating programmable vortical flows that gently and stably spin embryos along both vertical and horizontal axes. We use a deep learning method (Mask R‑CNN) to automatically delineate key embryonic structures in real-time. Rotation angles are then estimated via the tracker on surface feature points, and a multi-view tomographic algorithm reconstructs the full 3D embryo contour without fluorescent markers. In mouse embryo reconstruction, our system achieves ~3 µm reconstruction accuracy and reduces 40% imaging time compared to manual methods. This acoustofluidic approach opens new avenues for noninvasive, high-throughput embryonic morphokinetic analysis and quality assessment in assisted reproduction and biology.

Keywords: Micro/Nano robotics
Abstract: Droplet mixing research provides critical technical support for developing efficient and controllable complex chemical reaction systems. However, traditional methods and platforms pose inherent risks including cross-contamination and low energy conversion efficiency. Emerging alternative approaches, such as microfluidic chips, often entail complex fabrication processes and operational challenges. To address these limitations, this study proposes a non-contact droplet mixing platform. By applying a vortex acoustic field to droplets on a glass substrate, this system achieves stable droplet capture, precise manipulation, and rapid mixing. The platform significantly reduces contamination risks, minimizes reagent consumption, and enhances energy conversion efficiency, while featuring straightforward fabrication and operational simplicity. Experimental results demonstrate the system’s efficacy in accelerating microvolume droplet mixing, offering a practical solution for advanced chemical research.

Keywords: Micro/Nano robotics
Abstract: 提出了一种基于微镊子的微组装系统，用于制造软磁微型机器人。该系统通过三自由度机械手实现精确的三维定位。末端执行器设计结合了微型镊子和镊子盒。专用的传动机构将步进电机机械地连接到微镊子上，从而实现可靠的抓取和释放作。镊子盒可容纳不同尺寸的微镊子，同时确保组装过程中的作稳定性。为了证明组装能力，成功地纵了直径约800μm的磁性颗粒来组装四足软磁微型机器人。实验结果证实，该系统可以制造出具有优越运动性能的磁性微型机器人。这种组装方法在精密制造和生物医学作方面৻

Keywords: Micro/Nano robotics
Abstract: Ultrasonic field characterization plays a crucial role in various scientific and industrial applications, including acoustic manipulation, biomedical engineering, and material processing. However, conventional ultrasonic measurement platforms suffer from low positioning accuracy, complex manual operation, and poor repeatability, limiting their effectiveness for high-resolution acoustic field mapping. In this work, a fully automated ultrasonic acoustic intensity acquisition system is developed to address these challenges. The system integrates a computer-controlled three-axis translation stage, real-time signal acquisition modules, and adaptive scanning algorithms to enable precise probe positioning, flexible scanning path definition, and synchronized data collection. A dedicated host computer interface allows users to easily configure scanning parameters, establish communication with hardware components, and monitor real-time measurement status. The system performance is validated through two sets of experiments under predefined acoustic field conditions. The experimental results demonstrate that the system can accurately capture both focused and complex random acoustic fields, with good consistency to the expected distributions, thereby confirming its feasibility, accuracy, and adaptability. The proposed system offers a reliable and efficient solution for advanced ultrasonic measurement and acoustic field analysis.

Keywords: Cyber-Physical bio-system, Other related topics, Advanced materials and sensors for robotics
Abstract: Flexible ultrasound patches face the challenge of interfacial coupling between the device and the skin. The coupling material is required to possess not only excellent acoustic performance but also the ability to accommodate the modulus changes on both sides to achieve device support and adhesion. However, the currently widely used coupling materials fail to meet these demands. This study overcomes the rigid-soft conflict by developing a gradiently crosslinked polydimethylsiloxane (PDMS) coupling layer, containing a stiff top layer for probe support and an adhesive bottom layer for conformal contact. This design eliminates air gaps and enables stable coupling, achieving long-term, high-quality wearable ultrasound imaging.

Keywords: Micro/Nano robotics, Bio-Cell assembly and tissue fabrication, Biomimetic robotics
Abstract: Encapsulation of droplet particles and cells is essential in biomedical engineering and materials science. However, the environmental impact of commonly used Materials raise concerns. Existing research has primarily focused on encapsulating particles or cells in Newtonian liquids. This study investigates the encapsulation process using sodium alginate as the suspending medium and fish oil as the continuous phase. We first examined the droplet formation mechanism with fish oil, demonstrating that the droplet characteristics are suitable for particle encapsulation applications. Viscoelastic encapsulation Experiments identified conditions where the encapsulation Efficiency exceeded the stochastic limits predicted by Poisson statistics.

Keywords: Biomimetic robotics, Other related topics
Abstract: Autonomous navigation in complex 3D underwater environments is crucial for biomimetic robotic fish. However, the traditional Artificial Potential Field (TAPF) suffers from limitations like susceptibility to local minima, target unreachability when obstacles are near the goal, and path oscillations unsuitable for biomimetic locomotion, hindering practical applications. To address these deficiencies, this paper proposes the Dynamic Logarithmic Artificial Potential Field with Adaptive Virtual Guidance (DL-APF-AVG) method, which integrates four key innovations: (1) a logarithmic attraction potential function ensuring target convergence even near obstacles; (2) a dynamic, state-aware repulsion potential considering relative velocity and angle for smoother, proactive avoidance; (3) an Adaptive Virtual Guidance (AVG) strategy to deterministically escape local minima traps; (4) a dynamic step adjustment mechanism to actively suppress path jitter. The synergy of these components enhances target reachability, generates smoother trajectories appropriate for biomimetic locomotion, and provides robust escape from various local minima traps. Finally, the effectiveness of the proposed method over TAPF are demonstrated through 3D simulations.

Keywords: Biomimetic robotics, Other related topics
Abstract: This paper presents a novel biomimetic robotic fish prototype incorporating a 2-degree-of-freedom Spherical Parallel Mechanism (SPM) to enable bioinspired caudal fin actuation. The SPM architecture enables orthogonal flapping kinematics that synergistically enhance thrust production and maneuverability through coupled pitch-yaw motions. We systematically present (a) a mechanical system design methodology, (b) a comprehensive 6-DOF hydrodynamic-structural coupling model, and (c) numerical simulations verifying the system's propulsion performance in both forward swimming and turning maneuvers. Simulation results validate the effectiveness of the SPM-based propulsion concept and demonstrate the potential of the proposed biomimetic robotic fish for agile underwater locomotion.

Keywords: Biomimetic robotics
Abstract: The hydrodynamic performance of bio-inspired underwater robots is fundamentally governed by their undulatory kinematics. This study investigates body and/or caudal fin (BCF) propulsion in robotic fish, examining how kinematic parameters (tail-beat frequency, amplitude, wave number) and hydrodynamic conditions jointly influence thrust generation. By incorporating an inextensibility constraint into the conventional BCF undulatory model and employing computational fluid dynamics (CFD) simulations, we systematically analyzed the propulsion performance under the coupled effects of multiple undulatory parameters. Additionally, two distinct turbulent inflow conditions were established to evaluate the impact of incoming vortex streets on fish-like propulsion efficiency. The results indicate that increasing tail-beat frequency enhances mean thrust generation, particularly under conditions of large amplitude and small wave number gain. However, the effects of undulation amplitude and wave number on thrust production exhibit more complex dependencies. While mean thrust generally increases with greater amplitude gain and reduced wave number gain, certain conditions lead to opposing trends. Inflow with relatively low Reynolds numbers was found to enhance rapid-start swimming capability, whereas appropriately developed turbulent vortex streets could provide auxiliary propulsion under specific circumstances. These findings underscore the necessity of optimizing the coordination between frequency, amplitude, and wave number to achieve peak swimming performance, as well as the importance of adaptive adjustments in response to hydrodynamic conditions. The insights derived from this study offer valuable practical references for enhancing the propulsive performance of bio-inspired robotic

Keywords: Biomimetic robotics
Abstract: This study investigates swimming control strategies for biomimetic robotic fish using central pattern generator (CPG) control. Traditional CPG-based methods face challenges in achieving optimal C-turning performance with a single CPG parameter set, necessitating the use of multiple parameter sets. However, the complex fluid structure coupling effects complicate the establishment of the robotic fish’s dynamics model, the switching of CPG parameters using traditional dynamics modeling, and the acquisition of accurate physical parameters and modeling processes. This limits the applicability of these methods across different robotic fish. To address this, the study proposes an end-to-end control framework incorporating a perception neural network and an action strategy network. This framework enables real-time output of CPG parameters based on global camera data, thereby enhancing C-turning performance without relying on robotic model accuracy. Unlike current frameworks that depend on simulations, this strategy trains in a real pool environment, thus avoiding the simulation-reality gap. Experimental results demonstrate that the proposed end-to-end control framework significantly improves the C-turning performance of biomimetic robotic fish compared to single CPG parameter control. Additionally, the framework exhibits good scalability and adaptability to diverse application scenarios.

Keywords: Biomimetic robotics
Abstract: Locusts exhibit stable jump-flying locomotion with remarkable environmental adaptability. However, prior designs failed to integrate jumping and flying modes effectively due to insufficient kinematic model inspired by the locust, leading to motion instability. In this study, we utilized the Kinovea and MATLAB software to calibrate and analyze the jump-flying data of locusts, revealing that its linear velocity profile exhibits three distinct linear phases. Before flapping their wings, locusts experience a rapid increase in jumping linear velocity. In the ascending stage after flapping their wings, the resistance increases, and the linear velocity decreases slowly. Due to the jumping speed being much greater than the speed required for level flight, the linear velocity further decreases. Subsequently, we conducted simulations grounded in locust motion data. The results demonstrate that the robot's linear velocity profile similarly manifests three distinct linear phases, establishing direct kinematic correspondence with locust locomotion. This discovery will enable locust-inspired robots to more accurately replicate biological prototypes, achieving stable and long-time jump-flying locomotion.

Keywords: Biomimetic robotics, Micro/Nano robotics
Abstract: Micro flapping-wing vehicle (MWV) has emerged as a prominent research focus recent years due to its efficient aerodynamic properties and flexible maneuverability. Inspired by bumblebees, this paper presents a novel wing-body joint structure that allows for periodic flapping under piezo actuation with high frequency. Utilizing the multi-physics simulation platform ANSYS, we construct a coupled model that incorporates a piezoelectric actuator alongside an indigenously designed underdriven microwinged joint. The study explores the dynamic relationship between the kinematic characteristics of this joint and various actuating parameters, such as voltage and frequency. Through both static and dynamic simulation analyses, results indicate that the proposed structure exhibits a high degree of rationality and reliability, thereby fulfilling the requirements for application in micro-flapping wing vehicles.

Keywords: Wearable robotics, Prosthesis and exoskeleton robotics, Rehabilitation robotics
Abstract: 医疗级表面肌电图（sEMG）手势识别系统存在个体特征差异敏感性和嵌入式部署效率瓶颈，提出了一种基于深度可分离卷积和信道注意力协同的轻量级网络架构（DCA-Net）。该方法采用多尺度特征解耦机制，采用分组时域卷积捕捉肌电图信号的瞬态响应和稳态收缩模式，结合逐点卷积实现信道维度特征融合，参数量较传统CNN减少72.6%，较CNN-LSTM模型减少94.9%。为了进一步优化特征表示，引入了高效的通道注意力模块（ECA）。通过全局时域平均池化和门控卷积对关键通道的权重进行动态校准，抑制噪声干扰。实验使用自建的健康Ö

Keywords: Biosensors and bioactuators, Neural-machine interface, Cyborg intelligence
Abstract: Surface electromyography (sEMG) is widely used for gesture recognition and prosthetic control, yet conventional systems often rely on multi-channel acquisition and high-performance computation, limiting their applicability in wearable and low-power scenarios. To address this issue, we propose a compact, low-cost gesture recognition system based on a single-channel sEMG sensor. The system employs a custom-designed wristband that integrates a high SNR analog front-end, a 24-bit ADC, and an ESP32-S3 microcontroller capable of executing quantized neural network inference in real time. A lightweight 1D convolutional neural network with residual blocks is developed and deployed on-device to classify six common hand gestures. To support data acquisition and model management, a PyQt-based graphical user interface is also implemented. Experimental results with five participants demonstrate that the proposed system achieves over 95% classification accuracy with an average inference latency of 35 ms. These results validate the feasibility of single-channel sEMG-based gesture recognition for practical, real-time applications in wearable and assistive devices.

Keywords: Rehabilitation robotics
Abstract: This study proposes MFE-MTL, a multi-task learning framework based on low-channel surface electromyography (sEMG) signals for stroke rehabilitation. The method combines manual feature extraction with a joint classification-regression model to simultaneously perform hand gesture recognition and grasp force estimation. Compared with three widely used neural networks (FCN, CNN, and LSTM), MFE-MTL achieved the best performance under low-channel conditions, reaching a Pearson correlation coefficient of 0.89 for force prediction and 95.41% precision for gesture classification. The integrated multi-task design significantly improved recognition performance, demonstrating the effectiveness and practicality of MFE-MTL in resource-constrained rehabilitation scenarios.

Keywords: Biosensors and bioactuators, Advanced materials and sensors for robotics, Neural-machine interface
Abstract: Historical martial arts (HMA) represent a gradually growing yet not fully developed field of sports. This study looks into muscle activation patterns in HMA movements using surface electromyogram (sEMG). Three typical movements were analyzed across participants from different skill levels, with each movement performed with and without guided correction. The sEMG data were first collected from eight muscles, preprocessed, and then compared using both RMS and normalized metrics. Repeated movements were segmented and normalized to investigate average activation patterns of each participant. Distinct patterns of muscle activation in participants from different skill levels were revealed. Compensatory behaviour decreased after guidance was given. These findings highlight the potential of sEMG as a tool for identifying inefficient movement patterns and evaluating the effectiveness of technique training. This study supports the integration of sEMG into martial arts instruction frameworks, offering a data-driven approach to movements correction, skill development, and injury prevention.

Keywords: Cyber-Physical bio-system, Neuro-Control and communication
Abstract: This paper presents a miniaturized high-density electrophysiological acquisition system based on the STM32H723 microcontroller. The system achieves synchronous multi-channel biosignal acquisition, high-speed transmission, and real-time host decoding. By analyzing the amplitude-frequency characteristics of target electrophysiological signals, it dynamically configures the front-end’s on-chip programmable gain amplifier (PGA), sampling frequency (1 kHz – 20 kHz), and bandpass filter parameters. SPI protocol initializes front-end modules and establishes strict multi-channel synchronous sampling timing control. High-density electrode arrays are managed via multiplexing technology, with signals acquired in parallel through the SPI bus. A zero-copy DMA data channel transfers multi-channel data streams processed by power-line comb filtering to the host via USB 3.0 SuperSpeed interface (480 Mbps) for real-time decoding and visualization. Experimental results demonstrate:2.6 μv equivalent input noise (0.5–500 Hz bandwidth)；System power consumption < 0.3 mW at 2 kS/s/ch；Clear capture of surface electromyography (SEMG) (0–2〖mv〗_pp dynamic range) and local field potentials (LFP) (θ-rhythm at 4 Hz). The system outperforms conventional electrophysiological acquisition systems, providing a hardware-streamlined solution for high-density biomedical sensing applications.

Keywords: Cyborg intelligence
Abstract: Although mice possess a keen sense of smell and excellent local environmental perception and autonomous decision‐making abilities, they exhibit significant deficiencies in acquiring global map information and transferring learned strategies. Robots leverage sensors to construct global environmental models and perform precise path planning, yet remain inferior to mice in terms of flexible local perception and olfactory sensitivity. To integrate the advantages of biological and artificial systems, this paper proposes a "mouse-vehicle" hybrid intelligent robot, establishing a collaborative mode that synergizes biological free exploration with robotic auxiliary intervention. The system achieves real-time localization of the mouse and vehicle via an overhead camera, leveraging YOLO-based object detection and PID tracking algorithms to drive a differential-driven vehicle for retrieving the mouse during critical entrapment scenarios. Experimental validation in a 1.5×1.5 m maze demonstrates high target recognition accuracy, stable tracking performance, and effective retrieval functionality. This study introduces a novel paradigm for constructing bio-machine hybrid intelligent agents and expands their potential application space in complex real-world scenarios.

Keywords: Micro/Nano robotics, Medical surgical robotics
Abstract: To address the limitations of conventional control systems for miniature soft robots, which often depend on single permanent magnets or electromagnetic coils, resulting in weak electromagnetic field gradients and cumbersome drive systems, this study introduces a highly integrated hybrid actuation system combining electromagnetic coils and permanent magnets. By synergistically coordinating locomotion, deformation, and orientation, this innovative system significantly enhances the operational capabilities of miniature ferrofluidic microrobots (MFRs), thereby substantially expanding their potential for advanced applications in complex clinical environments.

Keywords: Micro/Nano robotics, Medical surgical robotics
Abstract: Navigating microrobotic swarms through complex vascular networks for targeted therapies remains a significant challenge. The dynamic bloodstream and opaque bio-tissue often hinders real-time, high-fidelity tracking. Here, we introduce a dual-modal framework that synergistically integrates B-mode and Doppler ultrasound. Preoperatively, we use B-mode ultrasound for high-resolution, automated 3D reconstruction of a phantom vessel. For intraoperative guidance, we employ color Doppler ultrasound to intuitively track micro-turbulent flow induced by the magnetically actuated rotation of the microrobotic swarm, enabling robust real-time localization. A Kalman filter then fuses the Doppler measurements with a motion model to achieve precise closed-loop control. This integrated approach provides a robust solution for navigating microrobotic swarms in tortuous environments, paving the way for advanced in-vivo diagnostics and therapies.

Keywords: Cyborg intelligence, Biosensors and bioactuators, Biomimetic robotics
Abstract: Climbing is highly desirable for insect-scale terrestrial robots operating in complex environments, especially for real-life applications such as search and rescue missions. This paper demonstrates the ability to control locomotion in ZoBorg, a cyborg beetle created from a living Zophobas morio, during inverted climbing. The unique combination of ZoBorg's flex-rigid structure, flexible footpads, sharp claws, and embedded sensors enables agile locomotion with exceptional adaptability on inverted surfaces, all at low power and low cost. Electrical stimulation of the antenna induces contralateral turn, while stimulating both elytra drives the ZoBorg forward. These advancements highlight ZoBorg's potential for navigating complex environments, making it a promising tool for future search and rescue missions.

Keywords: Cyber-Physical bio-system, Biomimetic robotics, Cyborg intelligence
Abstract: Cyborg insects have emerged as potential robotic platforms for search and rescue missions due to their small size, naturally agile structure and functions of the living insects. In post-disaster environments, collapsed debris poses a significant risk to these cyborg insects, as physical impacts may inadvertently damage body parts. This paper demonstrates the ability to control Zoborg, a cyborg beetle from living Zophobas morio, even when its middle legs are damaged. Electrical stimulation of left and right antenna could induce right and left turns, respectively, in Zoborg with leg impairment. Graded response in turning control was partially preserved after removing the middle legs. The results highlight the potential of cyborg insects for practical applications by mitigating the effects of physical damage and demonstrating their capability to adapt, making them promising platforms for search and rescue operations.