Abstract: The cooperative simultaneous localization and mapping(SLAM) finished in an active way by multiple robots in unknown environment is investigated.Local submap strategy is introduced in which each robot carries out map building in the local area around itself so that it can build local submap independent of other robots,then the SLAM problem in global environment can be decomposed into multiple sub-problems.The problem of active SLAM is converted into that of multiobjective optimization when each robot builds its local submap.The robot chooses the optimal control inputs so that the accuracy of localization and mapping,information gain and the cooperative motion will be synthetically optimized.Finally, all the submaps are fused into global map by extended Kalman filter(EKF).Simulation results prove the effectiveness of the prsented approach.