In this paper we analyze entropy based measures in various motivation and environmental configurations of mobile robot navigation in complex environments. These entropy based measures are used to probe and predict various environmental and robot configurations that can provide for the emergence of highly fit robotic behaviors. The robotic system uses a neural network to evaluate measurements from its sensors in order to establish its next behavior. Genetic algorithms, fuzzy based fitness and Action-based Environment Modeling (AEM) all take a part toward training the robot. The research performed shows the utility of using these entropy based measures toward providing the robot with good training conditions.
展开▼