A Multi-Objective Genetic Algorithm Approach for Path Planning of an Underwater Vehicle Manipulator

Abstract

In this work the kinematic redundancy resolution based on Multi-Objective Genetic Algorithm (MOGA) has been exploited to path planning of a underwater vehicle-manipulator system (UVMS). Some objective functions are analyzed with the purpose to achieve the desired evolution of the configuration of the mobile manipulator when the position and orientation of the effector is imposed. In the proposed method, the generalized coordinates, additional kinematic constraints and relevant objectives functions are selected, in such a way that the base motion is implicitly limited while maximizing the whole system manipulability. Simulations for a ten dof vehicle-arm are developed, with the consideration of three objective functions for optimization. The results reveal that it is possible to choose among several solutions from the Pareto Front, according to the importance of each individual objective or in other way, there are an objective function that dominated the optimization process.