Learning to Control Dynamic Systems

The article introduces two alternative approaches of self learning control of dynamic systems based on the framework of asyn-chronous dynamic programming BBS93]. The performance of the approaches is evaluated on a typical control task. The growing competence technique is presented as a useful method for robust learning. To demonstrate the capabilities of the self learning controller it is applied to learn exact control of a real cart-pole system.

[1]  Richard S. Sutton,et al.  Neuronlike adaptive elements that can solve difficult learning control problems , 1983, IEEE Transactions on Systems, Man, and Cybernetics.

[2]  C.W. Anderson,et al.  Learning to control an inverted pendulum using neural networks , 1989, IEEE Control Systems Magazine.

[3]  Michael I. Jordan,et al.  Learning to Control an Unstable System with Forward Modeling , 1989, NIPS.

[4]  G. Tesauro.  Practical Issues in Temporal Difference Learning , 1992 .

[5]  J. Sitte,et al.  A cartpole experiment benchmark for trainable controllers , 1993, IEEE Control Systems.

[6]  Martin A. Riedmiller,et al.  A direct adaptive method for faster backpropagation learning: the RPROP algorithm , 1993, IEEE International Conference on Neural Networks.

[7]  R. Riedmiller,et al.  Aspects of learning neural control , 1994, Proceedings of IEEE International Conference on Systems, Man and Cybernetics.

[8]  Martin A. Riedmiller,et al.  Using Neural Reinforcement Controllers in Robotics , 1995 .

[9]  Andrew G. Barto,et al.  Learning to Act Using Real-Time Dynamic Programming , 1995, Artif. Intell..