Angle Control for a Pneumatic Bionic Hand With Tendon-Driven Mechanisms Based on Digital Twin-Driven Motion Planning

Abstract

In this paper, a digital twin (DT)-driven motion planning and control framework is proposed for a pneumatic bionic hand with tendon-driven mechanisms. A DT model with virtual sensors is developed to generate multi-sensor data for motion planning under limited physical sensing. A multiquadric radial basis function (MQ-RBF) based optimization method is used to generate smooth reference trajectories. A fixed-time extended state observer (ESO) is designed to estimate friction and coupling disturbances, a backstepping sliding mode controller is developed to improve tracking accuracy. Experimental results on a pneumatic bionic hand validate the effectiveness of the proposed method. Note to Practitioners—This work is motivated by the practical need for smoother, faster and more accurate motion execution in pneumatic bionic hands used in collaborative manufacturing, service robotics and rehabilitation, where strong nonlinearities, coupling disturbances and limited physical sensing often degrade performance. By using an MQ-RBF based motion planning method, this paper presents a DT-driven strategy to generate reference trajectories for a pneumatic bionic hand although sensor information is limited. The reference trajectories are tracked by the pneumatic bionic hand, where a fixed-time ESO estimates the friction and coupling disturbances, a backstepping sliding mode controller improves tracking accuracy. Experimental results on the pneumatic bionic hand validate the effectiveness of the proposed method. Current limitations include reliance on industrial personal computer computation instead of an embedded real-time platform. Future work will focus on multi-hand grasping tasks and the integration of the pneumatic bionic hand into embedded real-time platforms suitable for industrial.