Evolved Neurocontrollers for Pole-Balancing

An evolutionary algorithm for the development of neural networks with arbitrary connectivity is presented. The algorithm is not based on genetic algorithms, but is inspired by a biological theory of coevolving species. It sets no constraints on the number of neurons and the architecture of a network, and develops network topology and parameters like weights and bias terms simultaneously. Designed for generating neuromodules acting in embedded systems like autonomous agents, it can be used also for the evolution of neural networks solving nonlinear control problems. Here we report on a first test, where the algorithm is applied to a standard control problem: the balancing of an inverted pendulum.

[1]  Dipankar Dasgupta,et al.  Evolving Neuroeontrollers for Pole Balancing , 1993 .

[2]  A. P. Wieland,et al.  Evolving neural network controllers for unstable systems , 1991, IJCNN-91-Seattle International Joint Conference on Neural Networks.

[3]  J. Urgen Branke.  Evolutionary Algorithms for Neural Network Design and Training , 1995 .

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

[5]  Richard S. Sutton,et al.  Training and Tracking in Robotics , 1985, IJCAI.

[6]  Xin Yao,et al.  A review of evolutionary artificial neural networks , 1993, Int. J. Intell. Syst..

[7]  J. D. Schaffer,et al.  Combinations of genetic algorithms and neural networks: a survey of the state of the art , 1992, [Proceedings] COGANN-92: International Workshop on Combinations of Genetic Algorithms and Neural Networks.

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

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