A More Realistic Replica of a Walking Biped for Implying Dynamics and Control on Ascending and Descending Stairs by Considering Self-impact Joint Constraint

Abstract

The aim of this study was to bring forth a model-based control method for trajectory tracking of a normal human-like biped on descending and ascending stairs using the straight knee state and to verify the controller performance in the straight knee state periods. Contemplating a straight state in knee joints does not consent to each leg’s shank link to undertake angular rotations greater than thigh link’s rotations. For this purpose, among suggested methods, the self-impact joint constraint has been recommended for energy-efficient (normal) bipedal walking with the implementation of straight knee constraint. To reach this intention, dynamical motion equations are derived, developed and modified to comprise self-impact joint constraint at the knee joint. For gaining control over this complex dynamical system, accessible anthropometric stair gait cycle data are used to derive desired trajectories of thigh and shank links of self-impact biped. Furthermore, for taking the stair ascent and descent of the biped under control, a nonlinear intelligent controller with adaptive neural network control method was put forward. In proportion to simulation results on stair ascending and descending of biped robot, the biped could track the desired angular positions and velocities of the links very well despite having complex nonlinear terms in its governing dynamical equations due to the presence of self-impact constraints. Moreover, during the activation period of joint self-impact, an error of link position angles was evidently improved in comparison to unconstrained biped.