Model Predictive Controller based Spatio-temporal Path Tracking Method for Trans-humeral Prosthesese.

Trans-humeral prostheses are worn by trans-humeral amputees to replace the missing upper limb segment between shoulder and elbow. Prostheses should be able to function as a natural limb for the user to gain the full advantage of wearing a prosthesis. When performing reach-to-grasp and pointing motions by the upper limb, the hand is capable of adhering to a straight line path with a bell shaped velocity profile. This paper proposes a dynamic path tracking method for trans-humeral prostheses to gain the capability of adhering to a straight line path. Proposed method uses Model Predictive Controller (MPC) developed based on the kinematic model of the prosthesis. Moreover, a shoulder matcher is proposed to match actual shoulder pose with the predicted shoulder pose and to select the best joint angles for the prosthesis for a particular instance. Furthermore, the proposed method is capable of dynamically updating the path if the human performs shoulder motions, which are not as planned by the MPC. Several experiments are conducted to validate the proposed method.