Function approximation for learning control door B.J. de Kruif