Applications of neural networks for trajectory control of robots

The use of artificial neural networks is investigated for application to trajectory control problems in robotics. In particular, non-adapting networks which have been trained off-line are investigated to reduce the computational requirements of local control of robot position and trajectory. A representative two-link robot is investigated using an artificial neural network which has been trained to compute the components of the inverse of the Jacobian matrix. An illustrative trajectory following task, moving the end effector in a circular path, is implemented and the performance of traditional RMRC and neural-based control is compared.<
>