A neuromorphic cascaded context sensitive controller for robot manipulator

The application of cascaded context-sensitive neural networks to suboptimal control of industrial manipulators operating under a weighted minimum time-fuel criterion is described. The nonlinearity and the joints coupling in the complicated robot dynamics (a major hurdle in robot-control design) are handled using the average dynamics method of linearization along the trajectory. The most interesting feature of the proposed neuromorphic controller lies in the possibility of adopting the weights of the function network, in real-time depending on the averaged system parameters presented to the context network