Terrain Exploration, Stability Control and Trajectory Planning of Humanoid Robots employing Several Navigational Strategies

Abstract

Research investigation on the navigation of humanoid robots is a prevalent aspect in robotics nowadays because scientists develop robots to function together with maximum cooperation in their workspaces. The robot’s potential to maintain stability while standing, moving in the environment, evading barriers, stepping over a barrier is essential to protect itself and the operator from percussion. This research aims to address a portion of the trajectory mapping challenges for achieving steady movement. To address the challenges, the present research work is focused on the concept, formulation and application of several state-of-the-art algorithms for smooth, percussion-free and efficient locomotion of humanoid robots in various terrains. Humanoid NAO is preferred as a robotic platform for evaluating the designed trajectory mapping algorithms. The kinematic analysis of humanoid NAO is carried out using the DH (Denavit-Hartenberg) parameter strategy to better understand mobility limitation criteria. The dynamic analysis of humanoid NAO is performed by employing Whole-Body Control aided Simulated Annealing approach to integrate human-like intellect into designing and represent it accurately. Here, Particle Swarm Optimization (PSO) tuned Proportional-Integral-Derivative controller (PID), Dynamic Window Approach (DWA) and Teaching-Learning-Based Optimization (TLBO) Approach and Multi-Objective Sunflower Optimization (MOSFO) tuned Modified Multiple Adaptive Neuro-Fuzzy Inference System controller (MANFIS) are outlined for trajectory mapping of single and multiple humanoid NAOs in terrains with several challenges. The navigation of humanoid NAO on unstructured terrains is controlled using a novel 3D-Multilinked Dual Spring-Loaded Inverted Pendulum integrated with BFGS Quasi-Newton tuned Artificial Potential Field (APF) controller. For multiple humanoid NAOs navigation, dining philosopher controller are designed and integrated into the base strategies. The established trajectory mapping strategies are evaluated on simulated humanoid NAO in WEBOT and accredited in actual humanoid NAO under real-time scenarios. Their results are compared, and the margin of error is obtained under the 6%. The formulated strategies are also compared with various conventional trajectory mapping algorithms that display considerable performance enhancements between 5% to 13%. Ultimately, the applicability of present research investigations is discussed, along with prospective expansion prospects.