An evolutionary approach to passive learning in optimal control problems