Efficient feature selection using polynomial basis functions enhances reward function learning in Inverse Reinforcement Learning.