6 DOF manipulators absolute positioning accuracy improvement using a neural-network

An absolute positioning accuracy improvement method, using an artificial neural network, has been developed. In this method, an artificial neural network is used with a conventional inverse kinematics computation module in parallel. The network automatically learns the manipulator model errors. When a manipulator carries out tasks, both the network and the inverse kinematics computation module receive the commanded end effector location, described in a Cartesian coordinates. The network output is summed together with the inverse kinematics computation module output in a joint coordinate. Driving the manipulator joint by summed result, manipulator endpoint positioning accuracy is improved. The effectiveness of this method was investigated experimentally by using a 6 DOF (degrees-of-freedom) manipulator forward kinematic model with model error. Results show that the absolute positioning error has been reduced by 1/3. Also this method was effective in a work area which was not used for the network learning