Using voting technique in mobile robot behavior coordination for goal-directed navigation

Abstract

In behavior-based mobile robot, the control strategy is distributed among a set of specialized behaviors. Each behavior with particular objective runs completely independently to send commands to control the mobile robot. However, behavior with different objectives may generate conflicting command. Therefore, behavior coordination is an important issue. An intelligent voting technique is implemented to solve this problem. Each behavior votes for a set of possible actions, with vote zero is the least desired action and vote one is the most desired action. The behaviors send votes as a possibility for each action set to achieve the objectives of the behaviors. An arbiter then performs command fusion and selects the most favored action that is pareto-optimal. This will solve the action selection problem and improve the probability to succeed. This technique has been implemented on UTM AIBOT mobile robot. The experimental results are presented and the reliability of the technique is shown.