Multi-robots Cooperative Online FastSLAM

Multi-robots cooperative online FastSLAM algorithm is proposed to solve the problem of particles surviving in the resamping processes, for paying no attention to the measurement in FastSLAM 1.0. The leader robot is defined and measured as a Robot landmark (RL) by the follower robot. At each step, the RL’s posteriori estimation is used as the leader robot’s priori estimation for its next pose prediction. Then the leader robot’s pose posterior estimation is used as the RL’s following priori estimation to be put into landmarks prediction. Through this implementation, the leader robot can enjoy the measurements by the follower robot’s help and avoid the complexity of EKF updating of FastSLAM 2.0. The simulation results show that the errors of the leader robot in localization and mapping are both reduced, more effective particles can survive, and the most important encouragement is that this method is easy in implementation.