Generation of evaluation function for robot force control using genetic programming

摘要: Force control is one of the most important and fundamental tasks robot manipulators. It known that a neuro-fuzzy method best methods for force control. Usually, controller trained to minimize error function. However, unexpected response such as overshooting or oscillation may occur when only evaluated, since dynamics environment not reflected in evaluation In this paper, we propose an effective function generation using genetic programming. The effectiveness proposed was evaluated by simulation.