Adaptability by Behavior Selection and Observation for Mobile Robots

Behavior-based systems are very useful in making robots adapt to the dynamics of real-world environments. To make these systems more adaptable to various situations and goals to pursue in the world, an interesting idea is to dynamically select behaviors that control the actions of the system. One factor that can give a lot of information about the world is the observation of use of behaviors (or Behavior Exploitation). Using this factor, the system can self-evaluate the proper working of its behaviors, giving it more adaptability. The paper describes how Behavior Exploitation can be used to influence motives using a simulated environment for mobile robots, and to acquire knowledge about the world using a Pioneer 1 mobile robot. It also exposes the repercussion on programming the system, and how it can help in designing a general control architecture for intelligent systems.