Design of autonomously learning controllers using FYNESSE

With the growing number and difficulty of control problems there is an increasing demand for design methods that are easy to use. Fynesse fulfils this requirement: without knowledge of a process model the system learns a control policy. Optimization goals like time-optimal or energy-optimal control as well as restrictions of allowed manipulated variables or system states can be defined in a simple and flexible way. Fynesse only learns on basis of success and failure of former control interactions and, thus, learning can be carried out directly at the real process. A priori knowledge about the control policy as, for example, a fuzzy or linear control law considerably improves the learning process. Moreover, the learned policy can be interpreted as fuzzy control law which allows for easily checking the plausibility.

[1]  R. Bellman.  Dynamic programming. , 1957, Science.

[2]  C. D. Boor,et al.  On Calculating B-splines , 1972 .

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

[4]  Didier Dubois,et al.  Possibility Theory - An Approach to Computerized Processing of Uncertainty , 1988 .

[5]  C. Watkins.  Learning from delayed rewards , 1989 .

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

[7]  Gerald Farin,et al.  Curves and surfaces for computer aided geometric design , 1990 .

[8]  D. Dubois,et al.  Fuzzy sets in approximate reasoning. I, Inference with possibility distributions , 1991 .

[9]  Hamid R. Berenji,et al.  Learning and tuning fuzzy logic controllers through reinforcements , 1992, IEEE Trans. Neural Networks.

[10]  Martin Brown,et al.  A perspective and critique of adaptive neurofuzzy systems used for modelling and control applications , 1995, Int. J. Neural Syst..

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

[12]  Didier Dubois,et al.  Handling uncertainty with possibility theory and fuzzy sets in a satellite fault diagnosis application , 1996, IEEE Trans. Fuzzy Syst..

[13]  Jianwei Zhang,et al.  Constructing fuzzy controllers with B-spline models , 1996, Proceedings of IEEE 5th International Fuzzy Systems.

[14]  Martin Spott,et al.  First results on the application of the Fynesse control architecture , 1997, 1997 IEEE Aerospace Conference.

[15]  Harro Kiendl,et al.  Fuzzy Control methodenorientiert , 1997 .

[16]  Rudolf Kruse,et al.  Numerical stability analysis for fuzzy control , 1997, Int. J. Approx. Reason..

[17]  Martin A. Riedmiller,et al.  Reinforcement learning without an explicit terminal state , 1998, 1998 IEEE International Joint Conference on Neural Networks Proceedings. IEEE World Congress on Computational Intelligence (Cat. No.98CH36227).

[18]  M. Spott.  Using classic approximation techniques for approximate reasoning , 1998, 1998 IEEE International Conference on Fuzzy Systems Proceedings. IEEE World Congress on Computational Intelligence (Cat. No.98CH36228).

[19]  Ron Sun,et al.  A subsymbolic+symbolic model for learning sequential navigation , 1998 .

[20]  D. Dubois,et al.  Fuzzy sets in approximate reasoning, part 1: inference with possibility distributions , 1999 .

[21]  Martin A. Riedmiller,et al.  Using reinforcement learning for engine control , 1999 .

[22]  L. Zadeh.  Fuzzy sets as a basis for a theory of possibility , 1999 .

[23]  Martin Spott,et al.  A theory of possibility distributions , 1999, Fuzzy Sets Syst..