Development of a vision aided reach-to-grasp path planning and controlling method for trans-humeral robotic prostheses

Abstract

This study proposes a reach-to-grasp path planning and controlling method for trans- humeral prostheses. Trans-humeral prostheses are used to replace the missing body part after the loss of upper limb (UL) above elbow. Reach-to-grasp paths refers to the paths taken by the human UL to reach towards an object with the intention of grasping. A trans-humeral prosthesis has been designed and fabricated with 5DOF. A simulation environment has been proposed using the design. Simulation environment consists of a virtual shoulder joint which can be actuated according to a natural human shoulder using an Inertial Measurement Unit (IMU). Prosthesis and the simulation environment has been used to experimentally evaluate the proposed path planning method. A reach-to-grasp path planning method combining Electromyography (EMG) signals and vision signals has been proposed. EMG Based Module (EBM) is capable of con- trolling prosthesis elbow motion e ectively with an accuracy of 92%. Visual Servoing Module (VSM) consists of a 2-1/2D visual servoing system to center the object of in- terest to the hand of the prosthesis and to correct the orientation. An object reaching algorithm has been proposed to reach towards the object. Later, the EBM and the VSM has been fused using an fusion lter. An improvement to the above method has been proposed to make the paths straight. It consists of a path generation module and a path tracking module. Path generation module is capable of generating a path towards the object. The object position is located and a path is generated from the current position of the prosthetic hand to the object position with the aid of vision. Path tracking module takes the prosthetic hand on the generated path considering shoulder motions. Two path tracking methods has been proposed: spatial path following method and Model Predictive Controller (MPC) based path tracking method. Proposed path planning method has been experimentally evaluated.