Abstract

This paper presents an approach of dynamic task allocation that enables the robot to automatically perform task in a distributed manner within unknown environments. The proposed cooperative system is based on swarming behaviors inspired by biological colonies. We assume that the lightweight robot is incompetent and needs cooperation form other robots to perform a task. Communication is limited within a certain range, so each robot only communicates task information with its neighbors locally using probabilistic flooding algorithm. The robot associates each task that it is aware of with a utility value rather than pheromone in ACO-based task allocation algorithms. A new task utility function is developed which considers not only the respond threshold but also the task priority, travel cost and updated time. We also propose a concept of task social utility corresponding to task utility which is renamed as personal utility. The relative value is computed through social utility and personal utility which acts as the basis for the robot to select the most favorable task. Our experimental results indicate that our task allocation system performs better compared to other systems.

Paper Details