Model analysis and control of biped dynamic walker with fault steps in a gait cycle

Abstract

The biped robot has two identical legs that undergo alternating stance and swing phases during walking. The robot’s motion is facilitated by actuators positioned at the ankle and hip joints. The paper investigates various fault scenarios in dynamic walker models, specifically situations where one or both actuators experience failure while the robot is in motion. The first model assumes faults in both the hip and ankle joints, while the second model assumes a fault only in the ankle joint. The dynamic walker typically undergoes two steps in each cycle. This study specifically examines the different types of faults that may occur during the second step. The goal is to identify the potential walking gaits that can result from these faults. It is observed that walking with faults at both the hip and ankle joints exhibited greater stability and a wider range of stable states compared to walking with a fault only at the ankle joint. This suggests that incorporating hip joint faults or underactuation into robotic or prosthetic gait patterns could improve their stability and overall performance. The local and global stability of the dynamic walker were analyzed using Poincare’s method and the basin of attraction plot, respectively. Additionally, a control algorithm based on kinetic energy shaping is proposed for the second model, which resulted in an increased basin of attraction. Notably, the second model exhibited the ability to walk with variable step length and speed, akin to human locomotion.