Robust Control of a Powered Transfemoral Prosthesis Device with Experimental Verification
American Control Conference (ACC)
This paper presents, compares, and experimentally implements two robust model-based controllers for transfemoral prosthetic walking: the robust passivity (RP) controller and the robust sliding mode (RS) controller. These findings constitute the first steps toward using model-based controllers for prosthetic devices as an alternative to commonly-used variable impedance and proportional-derivative (PD) control methods. The model upon which the controllers are based is a 5-link planar hybrid system (both continuous and discrete behaviors) with point feet, to represent a transfemoral amputee's body and limbs. A desired walking trajectory is generated through the framework of human-inspired control by solving an optimization problem. Smooth humanlike gait is achieved by combining model information with a desired trajectory. The stability of both controllers is proven for continuous dynamics within the framework of the Lyapunov stability theorem. Simulations show the proposed controllers are capable of meeting specific performance requirements regarding trajectory tracking of the prosthetic knee and convergence to a stable periodic orbit while walking on flat ground. Finally, both RP and RS controllers are experimentally implemented on AMPRO3 (the third iteration of Advanced Mechanical Prosthesis), a custom self-contained powered transfemoral prosthesis. Results show that both controllers provide humanlike walking and accurate tracking performance for a healthy human subject utilizing a transfemoral prosthesis.
Azimi, Vahid; Shu, Tony; Zhao, Huihua; Ambrose, Eric; Ames, Aaron D.; and Simon, Daniel J., "Robust Control of a Powered Transfemoral Prosthesis Device with Experimental Verification" (2017). Electrical Engineering & Computer Science Faculty Publications. 420.
V. Azimi, T. Shu, H. Zhao, E. Ambrose, A. D. Ames, and D. Simon, “Robust control of a powered transfemoral prosthesis device with experimental verification,” 2017, pp. 517–522.