Motion control of exoskeleton arm with potential energy minimization

Abstract

The exoskeleton arm represents a type of rehabilitation or service robot designed to assist individuals in upper-limb motion restoration or manipulation tasks. Achieving efficient assistive paradigms for users necessitates precise motion planning and control of exoskeleton arms along desired paths. During the motion control process, potential energy exhibits significant oscillations at high levels, resulting in discomfort during motion reconstruction. In this study, we introduce an innovative strategy for motion planning that incorporates potential energy minimization to enable accurate control of the exoskeleton arm with minimal variation in potential energy. This motion planning method is framed as a constrained quadratic programming problem and optimized with a dynamic recurrent neural network to ensure solution convergence and accuracy. Simulation and experimental results confirm that the proposed motion planning scheme allows the exoskeleton arm system to accurately follow desired motion paths. Furthermore, our findings reveal a significant reduction in potential energy variation, with an average decrease of 84.87% in the sagittal plane (X-Z), 82.25% in the planar plane (X-Y), and 87.49% in 3D space. This reduction highlights the effective integration of gravity compensation in achieving precise motion planning and control.