Research On Multi-Robot Cooperative And Coordinated Collision-free Navigation And Formation

The problem of multi-robot cooperation and coordination is central to mobile robotics. Cooperation and coordination will improve the intelligent performance of robots and can complete lots of impossible missions for single robot. Since multiple cooperative robots will relieve human of dangerous or tedious tasks, the research on multi-robot cooperation and coordination is of great both academic and applied significance.The project, "key techniques in multi-objective cooperation and control architecture of AUVs", is financed by the Post-Doctor Research Fund of Heilongjiang Province. This paper deals with the multi-robot cooperation and coordination of the project, and makes scientific researches on the environment modeling, collision-free navigation and formation initialization etc. Some novel points and approaches are also presented in this paper.How to build a representation of the environment is also critical to multi-robot cooperation. Generally speaking, there are two approaches to environment modeling: geometrical modeling method and topological modeling method. Having valued the advantages and disadvantages, this paper adopts the topological modeling method, and then plans a static collision-free path for every robot by using the heuristic A star algorithm. In local area the planer coordinates the behavior set by using multi-objective decision theory and collision-free rules based on real-time sensor information that ensurethe navigation free of collision.Having studied the problem of formation of multiple mobile robots, in this paper, we bring forward that the key of the formation is formation initialization. We present an approach to this problem based an praxeic utility theory and evaluate the selectability and rejectability for each option in the decision space, and assigns a score to that option using the praxeic utility equation. In this process, we let the detection of an impending collision trigger a redecision by the agent so as to solve the collision.It is shown in the simulation experiment that the algorithm of both the collision-free navigation and the formation initialization is feasible and effective.