An empirical study of crossover and mass extinction in a genetic algorithm for pathfinding in a continuous environment
H. David Mathias, Vincent R. Ragusa · 2016
Genetic algorithms are an often used tool for the problem of pathfinding. Their ability to find good solutions to multiobjective optimization problems makes them well suited to this task. Of course, genetic algorithms embody a broad range of techniques and strategies including crossover, mutation and mass extinction, each with multiple parameters and implementations. In this paper, we examine the effects of crossover and mass extinction on a genetic algorithm for planning a path through known obstacles in an unconstrained, continuous, static environment. Using a mutation-only genetic algorithm as a baseline, we study the effect of crossover with and without mass extinction events that occur at several different probabilities. We find that while both offer sometimes significant improvement in some cases, neither is universally beneficial.