A neural network model for trajectory formation of a redundant manipulator

Proposes a neural network model for trajectory formation of a redundant manipulator, which has the ability to control the speed profile of its end effector to make it bell-shaped. The model consists of two networks, a trajectory planning network and a trajectory forming network. These networks are similar to Hopfield networks, but the difference of these networks is that the input capacitance in the trajectory forming network varies with time. At first, the manipulator moves from the initial posture to the target point using the trajectory planning network, and the input capacitance of the trajectory forming network is calculated for the information of the planned trajectory. The trajectory forming network controls the manipulator using the calculated input capacitance. The authors apply the proposed model to the trajectory formation of a planar manipulator with 3 joints. The authors get the result that the proposed model can make the speed profile of the end effector be bell-shaped.