Research On Coordination Of AUVs Based On Multi Agents

Coordination and cooperation are some of the most important topics of multi-robot. Now, the combination of agent theory and technology of robot is the focus research, and this thesis is one. The patrolling and surrounding task of military AUVs in the two dimensions environment is the application background, and the theory of MAS is the direction.To begin with, a robot model of knowable-degree-driven is built up in the patrolling task. Then some coordination algorithms are given to solve the conflicts. In order to improve the effect of patrol, a method named adaptively adjusting distribution of regression-factor is presented in this paper. According to this method, patrolling robots can more frequently search those areas where the probability of target appearance is high, as result that the probability of discovering is advanced.After discovering the target, system comes into surrounding task. The thesis makes use of the theory of social laws, which is one of coordination methods of MAS. Unified laws, with which each robot among the system must comply, are constructed in off-line way above all. In order to accelerate surrounding, the set of laws is extended. The paper also illuminates the principle and sense of extension.Finally, a simulation system is built to realize the multi-robot system. And the feasibility of all algorithms under low communication broad condition is verified further.