Transfer and Evaluation of Decentralized Reactive Swing-Leg Controls for Powered Robotic Legs

Animals and legged robots balance dynamically by placing their feet into proper ground targets [1], which can be identified with simple locomotion models [2][3]. Commonly, fully robotic systems realize identified foot placements using inverse dynamics and kinematics [4], which requires a highfidelity system model and accurate estimate of a robot's full state, whereas human-in-the-loop applications rely on extensive motion libraries extracted from healthy human gait [5]. Both approaches may fail in highly dynamic situations, such as trips when encountering unexpected obstacles.