Robot path planning using neural networks and fuzzy logic

A new approach for path planning of robotic manipulators using neural networks and fuzzy logic is proposed. These alternative computing techniques are evaluated for high level control of robots. Neural networks are used to predict in real-time the trajectory of a moving object to be caught by a serial three-degree-of-freedom manipulator. An inference engine controlling the joint motion with fuzzy logic rules is described. Collision avoidance between the object and robot members is also considered. Simulation results are presented to illustrate the performance of the algorithm both in predicting the object´s movement and planning the robot´s trajectory