Adaptive Division of Labor in Multi-Robot System with Minimum Task Switching

In many multi-robot systems, various tasks are allocated dynamically to an individual robot and each robot should decide its own work that is the best commensurate with its current state. To solve complex task allocation problems, agentbased approaches based on the model of division of labor of many social insects have gained increasing attentions in recent years. In this paper, we consider the problem of adjusting the ratio of robots equally to the ratio of given tasks to handle the division of labor dynamically with less number of task switches. Inspired by several insect societies displaying an effective division of labor with the limited abilities, the response threshold model is applied. An Individual robot has a limited, constant-sized task queue and the information obtained from the observation behavior is stored within this queue. Using the ratio of tasks in queue and the predefined response threshold values for all possible tasks, an individual agent decide its task and to handle the desired division of labor dynamically and obtains the specialization for the specific tasks that induces the less number of task switching. To show the robustness and flexibility of our proposed method, various experiments are executed and the results are compared with an other method.