Model reference neurocontrollers based on feedback linearization

The authors report experimental results on learning of the feedback linearizing control laws for an inverted pendulum system based on an unsupervised learning control scheme. Only the case where no state transformation is required for linearizing the nonlinear system is considered. A method inspired by a direct adaptive control scheme was used to learn the linearizing law using artificial neural networks. The main advantage of feedback linearizing control laws is that after linearization, which is not exact due to errors in computation and learning of the control law, the nonlinearities lie in the range-space of the input. This is important since various robust control techniques can be implemented as an outer loop such that the desired performance is guaranteed