Use of Evolutionary Robots as an Auxiliary Tool for Developing Behavioral Models of Rats in an Elevated Plus-Maze

This work proposes the use of evolutionary robots as an auxiliary tool for the development of models to describe rats' behavior. Evolutionary robots can be useful to test hypothesis and models of behavior, what is demonstrated here in experiments where a mobile robot is trained to reproduce the behavior of the rat in an elevated plus-maze (EPM). A recurrent Elman multilayer perceptron is used to control the robot. Due to the impossibility of knowing each desired output of the neural network, the weight optimization is made by a genetic algorithm. In this way, each individual of the genetic algorithm represents a set of weights of the neural network. Experiments where the fitness of each individual was computed based on the difference between the trajectories of the evolutionary robot on a EPM replica and of the rats behavior on a real EPM are presented.