Adaptive Collaborative Position Control of a Tendon-Driven Robotic Finger

Abstract

This paper presents a collaborative control scheme that adaptively combines two different control strategies in order to optimize the trajectory-tracking performance and the angular-position response of the motor, installed at the metacarpophalangeal joint of each finger, in a robotic hand. The scheme contains an adaptive neuro-fuzzy inference system (ANFIS) in the feed-forward path, and a proportional-integral-derivative (PID) controller in the feedback path. The ANFIS supervises the motion-control and efficiently tracks the reference trajectory. The PID controller improves the transient response and robustly nullifies the effects of exogenous disturbances. The final control output is taken as the weighted sum of individual outputs from the ANFIS and PID controller. A hyperbolic tangent function (HTF) of the error dynamics is used to adaptively vary the weightages in the controller combination. This feature dynamically varies the contributions of individual control effort, delivered by ANFIS and PID controller, with respect to the variations in the angular-error dynamics. The PID parameters and the variation-rate of HTF are meta-heuristically tuned via an adaptive particle swarm optimization (APSO) algorithm. The collaborative controller effectively tracks the desired trajectory and improves the overall system response. The results of real-time experiments are presented to validate the efficacy of the proposed controller.