Human-adaptive control of series elastic actuators

SUMMARY Force-controlled series elastic actuators (SEAs) are the widely used components of novel physical human–robot interaction applications such as assistive and rehabilitation robotics. These systems are characterized by the presence of the “human in the loop” so that control response and stability depend on uncertain human dynamics. A common approach to guarantee stability is to use a passivity-based controller. Unfortunately, existing passivity-based controllers for SEAs do not define the performance of the force/torque loop. We propose a method to obtain predictable force/torque dynamics based on adaptive control and oversimplified human models. We propose a class of stable human-adaptive algorithms and experimentally show advantages of the proposed approach.