Exploiting monotony on a genetic algorithm based trajectory planner (GABTP) for robot manipulators

This paper presents an optimization of a Genetic Algorithm (GA) based trajectory planner (GABTP), exploiting the inherent monotony in the trajectories described by serial robotic manipulators, implementing a Forced Inheritance Mechanism (FIM). The approach described in this paper successfully solves both, non-redundant and redundant planar manipulators of 2, 3 and 7 degrees of freedom. The workspace of the manipulators is initially considered to be homogeneous and free form any obstacle and it is later extended to consider static obstacles. The trajectory planning is performed directly into the workspace of the manipulator in order to take full advantage of the natural dexterity of the open kinematic chain configuration, as opposed to solving the problem in the joint space, where the solution is constrained to a desired configuration. The fitness function for the GA is based on a modified potential field representation of the obstacles by considering the reach of the manipulator. The GABTP finds the best sets of configurations which drives the manipulator smoothly to its specified goal in workspace whilst preventing it from colliding with the obstacles.