Abstract
Remote control technology is a core means of ensuring personnel safety in high-risk scenarios such as deep-sea exploration and nuclear facility maintenance. However, existing exoskeleton teleoperations suffer from drawbacks such as discretized arm length adjustment, insufficient dynamic control adaptability, and limited execution accuracy. This study proposes a reproducible unilateral upper limb exoskeleton teleoperation system, employing a 7-DOF active drive architecture (3 DOF for the shoulder, 1 DOF for the elbow, and 3 DOF for the wrist) to match the kinematic characteristics of the human upper limb. The system integrates four key designs: low-inertia, high-reaction joints (weight reduction using carbon fiber and aluminum alloy), an ergonomic alignment mechanism adapted to upper arm circumferences of 74.8-105.7 mm, a 6-DOF passive compensation module for scapulothoracic wall motion, and a digital adaptive arm length adjustment mechanism (achieving stepless adjustment and dynamic updating of the URDF model through a sliding rheostat). The control strategy adopts a hybrid scheme of "master-end position impedance control and slave-end force-based impedance feedback," modeled based on the Lagrange equation, introducing adaptive weighted coefficient balance trajectory tracking and force feedback compliance, and verifying global asymptotic stability using Lyapunov functions and the LaSalle invariance principle. The experiment used the Franka Panda robot as a platform, based on a Linux real-time kernel and ROS architecture, with joint position data transmitted via 500 Hz Ethernet. Master-slave teleoperation achieved a mirrored reproduction of a spatial figure-eight trajectory, and gravity compensation effectively offset the load. Ablation experiments showed that the peak contact force in group C3 decreased to 7.41 N (a 68.8% reduction from baseline); the sum of squared residuals in arm length measurement was [Formula: see text], and the deviation after multiple wearing cycles was [Formula: see text] mm. The experiment confirmed that the system overcomes the adaptability and robustness deficiencies of traditional systems, achieving high-precision master-slave synchronization and load compensation. It meets the teleoperation needs of high-risk scenarios and can also provide a rehabilitation training platform for hemiplegic patients, demonstrating broad application prospects.