Adaptive Whole-Body Dynamics: An Actuator Network System for Orchestrating Multijoint Movements

Abstract

Adaptive locomotion is crucial for robots that move across rough terrain and variable ground surfaces to complete various tasks [1]. Recent advances in numerical computation allow robots to operate in real environments by using complex control frameworks based on precise models of the robots' bodies and surroundings. However, it is not always feasible to obtain a precise model because factors such as the robot's movement, position, shape, and stiffness and surrounding objects affect the robot's motion. In particular, the interaction with other autonomous systems, such as humans, can result in significant disturbances in robot motion. Therefore, robots should possess robust mechanisms to react to unexpected forces from their surroundings [2], [3]. Examples are seen in nature. Animals can effectively react to such disturbances because of their elaborate reaction mechanisms (e.g., rapid reflexes arise from the intrinsic mechanical dynamics of the musculoskeletal system) [4].