Novel series elastic actuator design and velocity control

Abstract

A novel series elastic actuator( SEA) was designed to implement physical human-robot interaction by embedding a new torsional elasticity between the gear train and driven load which made the actuator more integrated. The elasticity reduced the upper bound of impedance and enhanced the capacity while operating in constructed environment. In order to realize the stable velocity control of SEA,a BP neural network controller which can compensate nonlinear characteristics of the load gravity was proposed. Experiment results show that this actuator has great velocity control performance and its low impedance makes it possible to adapt to variable environments. This kind of actuator is desirable in many applications including legged robots,exoskeletons and industrial manipulators.