Design of controllers for humanoid robots for stable motion with optimized parameters using soft computing techniques

Abstract

newline newline newlineIn this study, our prime objective is to develop efficient path planning techniques and design controllers for humanoid robots with stable motion using emerging bioinspired algorithms. To meet our goal, we further divided primary objective into sub-categories, which were subsequently organized as chapters in this thesis. firstly, the first chapter presents an outline of diverse approaches used for route planning in legged robots, wheeled robots, unmanned aerial vehicles (UAVs), unmanned ground vehicles (UGVs), and related devices. This chapter provides a concise explanation of dynamical systems and the mechanics of robots. newline newlineAn inclusive review of the present literature on controller design for various types of robots, including humanoid robots is presented in chapter 2. Various soft computing techniques viz., deterministic, heuristic, and metaheuristic were also discussed that were employed to achieve steady movement in presence of obstacles. newline newlineChapter 3 includes an outline of fuzzy logic controllers and their design methodologies. This study introduces an novice method of path planning for single and multiple NAOs in complex and changing terrain, while ensuring the avoidance of collision with obstacles. Webot simulator was used to conduct the experiments for both static and dynamic obstacles. The presented methodology integrates a fuzzy controller based on Reinforcement Learning with a petri-net model. newline newlineIn chapter 4, Grey Wolf Optimization algorithm and its several modified variants has been discussed. In this study, the Modified Grey Wolf Optimization (MGWO) method was developed that obtained the satisfactory results over the previously existing state of art methods. The developed method have been tested over several maps for autonomous robots using MATLAB. Furthermore, comparative analysis between MGWO and other bionic algorithms were carried out that proclaims the efficacy of the presented method in terms of trip time, path calculation, obstacle avoidance, and convergence. We substantiate