Chaos-based optimal path planning of humanoid robot using hybridized regression-gravity search algorithm in static and dynamic terrains

Abstract

The present work deals with the navigational analysis of humanoids in complex terrains by tuning the classical approach with reactive techniques. Here, the linear regression (LR)-based classical approach is merged with the gravitational search algorithm (GSA), called RGSA, and Chaos is added to achieve an optimum path. To improve upon the local traps, Chaos is introduced into the basic structure of the RGSA model to attain a global solution without effecting the higher convergence rate of the RGSA model. The Chaos further adds ergodicity to the dynamical system, gaining an optimal path length in the shortest time possible. The use of the proposed hybridization technique is found to be effective as compared to its counterparts. Various chaotic maps like the logistic map, Gauss map, piecewise linear chaotic map, and sinusoidal maps are used, and the results are compared. Also, the smoothness of the path is ensured during trajectory planning. The analysis is done in real-time lab conditions and simulation environments using single and multiple humanoids (Nao robots) with static and dynamic obstacles within the search space. The simulation for the path planning is done using the Webot software, while the choregraphe software is used for experimental path planning. The simulation and experimental work presented in the current paper shows good agreement and are within 6% acceptable limits. The proposed controller is then compared with a vision-based approach and another parallel sensor-based approach for optimum operational cost and smoother trajectory. The current work is a novel approach in which the humanoid’s interaction with the mobile robot is also analyzed. The proposed approach successfully avoided the dynamic obstacle in the form of the KHEPERA-II mobile robot. This concept can find potential applications in soccer scenarios employing mobile robots as opponent players to the Nao robots. Further, because of the lower operational cost, the proposed methodology can also be applied to other areas, such as multitasking, cooperative scheduling, power distribution systems, etc.