Multiple aerial platforms in dynamic scenarios: study of a prototype for solving the problem of multiple travelling salesman problem with a reinforcement learning approach