Evolving neurocontrollers for balancing an inverted pendulum

This paper introduces an evolutionary algorithm that is tailored to generat e recurrent neural networks functioning as nonlinear controllers. Network s ize and architecture, as well as network parameters like weights and bias t erms, are developed simultaneously. There is no quantization of inputs, out puts dr internal parameters. Different kinds of evolved networks are presen ted that solve the pole-balancing problem, i.e. balancing an inverted pendu lum. In particular, controllers solving the problem for reduced phase space information (only angle and cart position) use a recurrent connectivity st ructure. Evolved controllers of 'minimal' size still have a very good bench mark performance.