Modeling and Control of Nonlinear Series Elastic Actuator

Abstract

Series Elastic Actuators are force actuators with applications in robotics and biomechanics. In linear Series Elastic Actuators, a large force bandwidth requires a stiff sensor (spring), but the output impedance puts an upper limit on this parameter, therefore selecting the proper spring is difficult in these actuators. In this paper, Series Elastic Actuator is modeled with a nonlinear, stiffening spring and controlled using the Gain Scheduling method. Simulations show that both linear and nonlinear models have similar force bandwidths, but the nonlinear one shows much lower output impedance. Hence, the choice of spring for actuator design is an easier task than that of the linear model. Also, as a force-augmenting device for the knee joint in normal human gait, the nonlinear model acts better in simulations.