Multi-robot Cooperative Localization In Indoor Dynamic Environment

Cooperative localization is a key issue in mobile robotics and the basis of multi-robot navigation and exploration. Compared with localization using a single robot, multi-robot cooperative localization makes full use of the sensor information fusion between robot groups, improves the precision and comprehensiveness of surrounding environment perception, thereby improving the overall positioning accuracy of robot groups. This paper mainly studies the problem about multi-robot cooperative localization based on environment information interaction in indoor dynamic complex environment.In order to carry out multi-robot cooperative localization experiments, a multi-robot experiment system based on wireless network communication was designed and realized. This system can not only do real-time robot motion control, but also can obtain the robots' real-time motion parameters, as well as real-time sensor data, thereby ensuring the experimental verification of cooperative localization algorithms.In order to avoid equipping a heavy localization marker on the robot which is needed for direct observation between robots, and to make full use of the interactive information between robots in the same scene, a method for obtaining multi-robot observations based on MbICP algorithm is proposed, which makes multi-robot cooperative localization using EKF and REHF feasible and reliable in cluttered environment. Considering the effect of robot motion state on the cooperative localization, in order to avoid over-frequent localization when robot is moving stably and to avoid invalid localization when robot is moving unstably, we present an algorithm for adaptive selecting periods of the cooperative localization based on autonomous motion state estimation. A series of experiment results based on Pioneer 3-DX multi-robot experiment system and data analysis show the method's validity, robustness and practicability.