Employing Neural Networks for Manipulability Optimization of the Dual-Arm Cam-Lock Robot

Abstract

Cooperative systems have been extensively investigated in literature. Herein the criteria and implementation for employing neural networks artificial intelligence for finding the optimal configuration of the Dual-Arm Cam-Lock (DACL) robot manipulators at a specific point with the objective to optimize the applicable task-space force in a desired direction are addressed. The DACL robot manipulators are reconfigurable arms formed by two parallel cooperative manipulators which operate redundantly but they may lock into each other in specific joints to increase structural stiffness in the cost of losing some degrees of freedom. Obtaining the optimal configuration demands lots of computational time and is not practical in real-time applications. The neural network is trained to approximate the optimum configuration considering the geometrical constraints of the planar arms using genetic algorithm as a multi-objective optimizer.