A dynamic strategy based on road-partitioning model in Robocup Rescue Simulation

摘要

In RoboCup Rescue Simulation System (RoboCupRescue Simulation System, RCRSS), agents have to rescue civilians trapped in the burning buildings and extinguish fires after an earthquake. In order to improve the efficiency of the rescue coordination and reduce the consequential loss, we use partitioning strategy to disperse the agents into different regions, but the agents can not clear up important areas so that affect the overall efficiency of the rescue. We focus on the problem of how to allocate the task regions toward heterogeneous agents efficiently. We formally define road model and partition model by their geometrics and the surrounding environment information. Then, this paper proposes a partition fusion algorithm and road model refinement against police force. For fire brigade, we add the convex hull and cluster model in road-partitioning model and give different priorities to fire buildings so that the agents will control the fire better. Overall, the paper proposes a collaborative approach based on the classification of dynamic partitioning with road model and fire-controlling model to solve the task allocation. The approach has been evaluated in the Robocup China open 2013 in which we won the third place.