Perception Control and Planning Technologies of Upper-Limb Exoskeleton Robots

Abstract

To address key limitations of traditional upper limb rehabilitation exoskeleton robotic arms—including rigid motion planning trajectories, insufficient human-robot interaction safety, and poor adaptability to diverse patient rehabilitation needs—this paper presents an integrated perceptual control and motion planning framework tailored for clinical and home-based medical rehabilitation scenarios. The framework adopts a two-layer design: For perceptual control, it leverages compliant interaction and impedance control to achieve safe, comfortable human-robot collaboration, eliminating the risk of secondary injury from rigid mechanical contact. For motion planning, it combines IMU-based motion data acquisition with Dynamic Movement Primitives (DMP) and supervised learning, avoiding the complex inverse kinematics calculations required by traditional Cartesian space planning and significantly improving trajectory generation efficiency. Experimental validation demonstrates that the proposed system effectively aligns with natural human upper limb movement patterns, enabling real-time adaptive trajectory adjustment for patients at different rehabilitation stages. It meets personalized training needs, enhances user comfort, and provides a practical technical solution for rehabilitation exoskeletons, supporting their widespread adoption in both clinical settings and home environments.