Stability analysis and control of biped walking gait on slope and uneven terrains

Abstract

Robot locomotion can be broadly classified into wheeled or legged. Wheel mechanism provides fast, energy efficient and smooth locomotion on flat surface whereas legged locomotion is very versatile, capable of moving on rough terrain and can step over obstacles. Legged locomotion does not need continuous support surface and can use separated footholds, that provide better speed and efficiency, on uneven surface. Legged locomotion is more challenging in terms of control, energy consumption and complexity of dynamics and balance. There have been significant developments over the past three to four decades in the design of legged robots and their coordination, control, balance and stability during locomotion on flat terrain. The locomotion on uneven terrain proves to be a niche research topic for biped research groups around the world. The stability, robustness, controllability and trajectory tracking are the main topics to consider in biped locomotion. This thesis focuses on modelling, simulating, building and controlling a biped robot to follow a stable trajectory. The walking of a biped robot is analysed using Passive Dynamic walking and Zero Moment Point (ZMP) based walking. A passive walker can walk in a stable manner down a particular slope without any actuation; but walking on flat and any other slope needs active control inputs. In this work, the control of a fully actuated compass gait biped robot based on feedback linearization technique is carried out using hybrid Lagrangian model. Two cases namely flat terrain and slope are considered for stability analysis. The number of actuators and control effort can be reduced by limiting the actuation to either hip or ankle joint. The analysis of walking when the actuation is applied only at ankle or hip is also carried out with partial feedback linearization technique. As the number of actuators is less in under actuated walking, it is more efficient in terms of energy consumption. ..