Optimized parallel joint springs in dynamic motion: Comparison of simulation and experiment

Abstract

Inspired by human tendon function, parallel springs in the joints of bipedal walking robots allow for a significant reduction of the cost of locomotion in simulation. An experimental setup is presented here which enables to verify the simulation results. The beneficial effects of parallel springs are investigated for a four-link robot arm equipped with a parallel spring at the last link. The desired motion results from simultaneous optimization of spring properties and motion which was shown to achieve the best reduction of actuation effort. Spring utilization is considered for two types of motion, free space motion and motion including contact with the environment at the end-effector. It is shown that the absolute reduction of effort through the utilization of an optimal spring is independent of friction. Hence, influence of the spring on effort is more relevant at higher speeds where the motion itself requires more actuation effort. Furthermore, for motion with environmental contact, a spring can reduce actuation effort more than in free space motion, especially for stiffer environments.