Active Disturbance Rejection for Bipedal Walk of a Humanoid Robot Using the Motions of the Arms

Abstract

A control system for the walking of a redundant biped robot in the swing phase is considered. The biped is a humanoid with 6DOF per leg and 3DOF per arm. The controller will be based on a full kinematic model of the robot to depict a more accurate behavior of the robot. The arms of the robot are used to compensate for disturbances the robot may experience during walking. Instead of controlling the robots ZMP, keeping it within the support polygon, all six foot support reaction components are controlled. First, a “shoe” with force sensors detect the forces and moments on the foot for feedback. The feedback from the joint servos provide position and velocity information. The support reaction and the joint position/velocities are fedback to a sliding mode controller, which makes adjustments to the arm links’ acceleration to compensate the shift in the reaction components. Simulations show the comparison of the ZMP shift when disturbances are applied with and without controlling the reaction forces to prove the effectiveness of the approach.

The Impact of Simplifications of the Dynamic Model on the Motion of a Six-Jointed Industrial Articulated Robotic Arm Movement

“…Daniel and Soloniaina's work on robot modeling for reinforcement learning control emphasized the computation of inertia tensors [49]. Hill's research on bipedal robots highlighted the use of arms in disturbance rejection, considering the inertia matrix of each link [50]. This study aims to bridge the gap in understanding the implications of model complexity on robotic arm dynamic models, a subject of growing relevance in the context of increasingly refined industrial automation demands, and indicate a trend toward merging advanced computational methods and control theories in dynamic modeling.…”