Multi-robot Real-time Cooperative Localization Based on High-speed Feature Detection and Two-stage Filtering

Zeyu Zhou, Wei Tang, Zenghui Wang, Lijian Wang, Renyuan Zhang

Abstract: Multi-robot cooperative work is one of the important schemes in intelligent reconnaissance, transportation, combat and other scenes. One of the difficulties is how to use the cooperation between robots to achieve high-precision robot positioning in large maneuvering actual unknown scenes. Aiming at the above problem, this paper constructs a method of cooperative localization under multi-robot real-time communication, which provides a feasible scheme for successful deployment of multi-robot system. Firstly, a high-speed feature detection and transformation method based on visual information is proposed, which can transform real-time pose information of sub robot into effective filtering prior information; secondly, a multi-robot cooperative localization method based on two-stage EKF is proposed for dynamic real scene, we set up a novel vision information to enhance the accuracy, master robot modifies pose estimation of slave robot through observation information of vision module, to reduce the cumulative error of self-positioning; thirdly, according to low latency requirement for position's communication, a lightweight multi-robot information intersection mechanism is established, and the multi-robot cooperative localization system proposed in this paper is successfully deployed on actual platform to achieve real-time performance. Experimental results show that robots' positioning accuracy are improved, with real-time communication between robots, no collision and no deadlock area of the system.