Abstract

A new methodology to design the motion control of robot manipulators is presented. The problem of tracking a trajectory, given in Cartesian coordinates, is considered. A constraint is imposed, which is unaware of manipulator's kinematics. The only available measurement will be the actual Cartesian position of the robot's tool. The strategy is based on Fuzzy Modeling. The controller is trained with a set of samples from the robot direct kinematics, although the more involved inverse kinematics is needed for trajectory control. The controller is trained in N-1 dimension space, where N is the dimension of the path to be followed. Motion problems near singular configurations are tackled during the controller design and training phase. Simulation tests are presented for a 6-DOF industrial robot.