LEARNING TO COORDINATE BEHAVIORS FOR ROBOT NAVIGATION

Past approaches to navigating behavior-based robots have relied on a p riori determined arbitration schemes to control the total behavior of a robot. Such arbitrators can only resolve deadlocks between competing behaviors by introducing randomness. Without monitoring its self-beha vior and using state information a robot cannot improve its performanc e. We show that by continually monitoring its actions (the output of b ehaviors) a robot can discover the deadlocking features (local minima) in the environment that cause failure. The robot can determine the co rrect arbitration sequence between its behaviors that is needed to res olve the conflict. We present experimental results of a Nomad robot di scovering and then using environment features to escape local minima w hile navigating in unknown environments.