An approach to the pursuit problem on a heterogeneous multiagent system using reinforcement learning