IEEE/ASME International Conference on Advanced Intelligent Mechatronics : [proceedings]

Robotic systems operating in unstructured environments are prone to damage, posing significant challenges for task continuity and system recovery. To address this, we propose a modular resilient robot designed to autonomously recover functionality following partial damage. MR-Robot features a novel "key-lock" docking mechanism implemented via male and female connectors, with the female connector incorporating a multilayer structure that supports docking, locking, disassembly, and joint mode switching. This architecture enables seamless transitions between full actuation and under-actuation modes. A kinematic model is developed and an inverse kinematics solution is formulated to support precise motion planning. To minimize wiring complexity and enhance scalability, a lightweight MQTT-based communication protocol is adopted for intermodule data exchange. Furthermore, closed-loop control using a PID algorithm ensures accurate joint positioning. A functional prototype has been developed to validate the proposed design. Experimental results demonstrate MR-Robot’s capability for autonomous docking and its adaptability through transitions between active and passive joint states, confirming its potential for resilient robotic applications in challenging environments.

Rotorcraft uncrewed aerial vehicles (UAVs) have been employed in coastal defense for surveillance tasks; however, the high energy consumption limits their operational range in wide-ranging environments. To address this challenge, a Lyapunov-based formation controller for fixed-wing UAVs is proposed to reduce energy consumption and further expand the surveillance coverage. Furthermore, the windy environment poses a considerable challenge to UAVs, especially those in close formation flight, as minor deviations from the desired position may lead to potential collisions. To mitigate these effects, the wind information estimated by the proposed sliding mode wind observer is incorporated into the formation controller. Finally, the effectiveness of the proposed controller is validated through the software-in-the-loop (SITL) simulation using the open-source PX4 autopilot.

In micromanipulation, modeling the dynamics and environmental interactions of micro objects is challenging. This study employs Imitation Learning (IL) to avoid explicit modeling, allowing stable and efficient autonomous operation in the Robotic Micromanipulation Systems (RMS) by learning strategies directly from human demonstrations. We propose Micro-ACT, an IL model tailored for RMS, based on the Action Chunking with Transformers (ACT) framework. Micro-ACT uses multimodal observations and latent variables to generate action sequences of the tool tip. Users can intuitively demonstrate actions by dragging the tool tip with a cursor on a graphical interface. Tool tip positioning relies on an enhanced self-calibration method, where calibration data are weighted by template matching similarity and the matrix is optimized using weighted least squares. Experiments confirm that this method accurately tracks the tool tip during continuous demonstration and inference operations. For a pushing task spanning approximately 6.79 mm, Micro-ACT performs comparably to expert operation, averaging 19.56 s with an 80% success rate and an IoU of 0.79. Incorporating motor velocity data further improves control over subtle movements. Moreover, Micro-ACT exhibits robust performance under artificial visual and physical disturbances, underscoring its reliability.

Muscle function monitoring and assessment play a crucial role in sports medicine, but traditional muscle monitoring methods have certain limitations. In this paper, we propose a flexible, stretchable, low-cost, and highly durable wearable ultrasound patch for real-time dynamic monitoring of muscle tissue. The device features an orthogonal cross-array arrangement of the transducer array, with an "island-bridge" layout for electrical connections. The single-layer serpentine traces endow the device with excellent flexibility, stretchability, and torsional capability, achieving a stretch rate of 30%, which allows the patch to closely conform to the surface of the skin. High-performance 1-3 composite ceramics are selected as the material for the ultrasound transducers, offering excellent electromechanical coupling and acoustic properties. The device is encapsulated and electrically isolated with low-modulus, highly elastic silicone resin elastomer. Studies show that the wearable flexible ultrasound patch can accurately and continuously monitor the forearm muscle dynamics using both A-mode and M-mode, demonstrating its broad application potential in muscle monitoring, human-machine interaction, and wearable devices.

Modeling of ElectroMechanical Brake (EMB) system is fundamental for the safety control of intelligent or autonomous vehicles. However, traditional physics-based EMB system modeling can be challenging due to requirement of precise parameters value and the switching among multi-stage working process of the EMB system. This paper proposes a hybrid physics-data driven modeling approach based on the Linear-Parameter-Varying (LPV) state space representation to describe the multi-stage working process of the EMB system in a unified model framework. The unified model starts with a traditional physical-based EMB system model, where multiple system parameters are presented, while the friction torque and the hysteresis braking force is modeled by switching functions; Then, the physical model is transformed into a quasi-LPV state space with motor speed appeared in the scheduling parameter vector; The system matrices are then identified through data, avoiding the precise value of each individual system parameters. The effectiveness of the unified model is then validated using experiment data, showing the smoothness and accuracy of the proposed model.