Now we have defined the basic principles behind the NS method, we need to find a way to integrate it into the definition of the fitness function that will be used to guide the neuroevolution process. In other words, we need to define the novelty metric that can capture the amount of novelty that is introduced by a particular solver agent during the evolutionary process. There are several characteristics that can be used as novelty metrics for a solver agent:
- The novelty of the solver genotype structure—the structural novelty
- The stepping stones found in the search space of the solution—the behavioral novelty
Our primary interest in this chapter is to create a successful maze navigator agent. To successfully navigate through the maze, the agent must pay equal attention to most places in the maze. Such behavior can be achieved...