Evolving Efficient Theme Park Tours

Abstract

In this paper, we describe the use of an evolutionary algorithm (EA) for the problem of visiting rides in a theme park. Generating a tour of the rides in a theme park is an instance of the time dependent traveling salesman problem (TDTSP) in which the cost of visiting any two rides depends not only on the distance between the rides, but also the time to wait in line and ride each ride. We describe the implementation of the evolutionary algorithm, including EA's ability to satisfy the constraint of scheduling lunch in the theme park. We present several test problems and compare both the running times and costs of tours generated by the EA to those of a previously described dynamic programming heuristic.