Intelligent Hybridization of Regression Technique with Genetic Algorithm for Navigation of Humanoids in Complex Environments

Abstract

SUMMARYIn the current investigation, a novel navigational controller has been designed and implemented for humanoids in cluttered environments. Here, regression analysis is hybridized with genetic algorithm (GA) for designing the controller. The obstacle distances collected in the form of sensor outputs are initially fed to the regression controller; and based on the previous training pattern data, an intermediate advancing angle (AA) is obtained as the first output. The intermediate AA obtained from the regression controller along with other inputs is again fed to the GA controller, which generates the final AA as the desired final output to avoid the obstacles present in a complex environment and reach the destination successfully. The working of the controller is tested on a V-REP simulation platform. In the current work, navigation of both single as well as multiple humanoids has been attempted. To avoid inter-collision among multiple humanoids during their navigation in a common platform, a Petri-Net model has been proposed. The simulation results are validated through a real-time experimental platform developed under laboratory conditions. The results obtained from both the simulation and experimental platforms are compared against each other and are found to be in good agreement with acceptable percentage of errors. Finally, the proposed controller is also compared with other existing navigational controller and an improvement in performance has been observed.

Design of a hybrid controller using genetic algorithm and neural network for path planning of a humanoid robot

“…Based on the above discussed hybrid approach, a control scheme has been prepared for testing on a NAO humanoid robot (Kumar et al , 2018a, b, 2020c). V-REP software has been chosen as the simulation platform to be used for demonstrating humanoid navigation.…”

Dynamic Strategy Planning of Humanoid Robots Using Glowworm-Based Optimization

“…Table V. Comparison of recurrent neuro-fuzzy approach 24 with GO controller in terms of track stretch.…”

“…Clever and Mombaur 23 developed a motion transfer scheme from humans to humanoids using a 3D template model. Kumar et al [24][25][26] developed various humanoid motion planning models using computational intelligent techniques. Baskoro and Priyono 27 used zero moment point and inverse kinematics information for stability analysis of a humanoid model.…”

Humanoid navigation: A firefly based approach

“…Bin et al 39 proposed use of Riemannian manifold learning approach for analyzing motion capture data segmentation. Kumar et al [40][41][42][43] designed several motion control algorithms for efficient navigation in complicated terrains.…”