It is a complex control problem for multi-robot coordination in unknown environment.The solution requires an appropriate balance between the task object inside robot and task object between robots

while the collaboration in the multi-robots systems is the most important problem for its efficient work.A robust

decentralized approach for structural adaptation in MAS is demonstrated.Based on self-organization principles

the method enables robots to modify their structural relations.And the robots adopt their next-step relationship by tasks allocation which has finished.It is shown that

the performance of self-organization is close to that of an upper bound centralized allocation methods and better than a static method and a random adaptation method.