Evolving automatons for distributed behavior arbitration

Designing controllers for mobile robots is an advanced engineering task. We develop a method for designing the controllers automatically using artificial evolution. In contrast to similar approaches, our aim is to evolve controllers for more complex tasks involving multiple interactions and behaviors. We achieve this by providing a set of pre-designed simple low-level behaviors, which are either programmed manually or designed automatically. The output of our algorithm is a behavior arbitration mechanism consisting of a set of finite-state automata. Inspired by the incrementality of the Behavior-Based Robotics, we employ incremental evolution to decrease the size of the search space and to automatically design the robot controller in a bottom-up manner