Multi-robot CSLAM in the Unknown Environment

Abstract

In order to solve the problems of slow detection speed, incomplete perception of environmental information and poor system robustness in a single-robot system, a method for autonomous simultaneous localization and mapping in an unknown environment based on the multi-robot cooperation is proposed. The Rapidly-exploring Random Tree (RRT) is applied to the rapid selection of frontier points in multi-robot systems. With the improved task allocation algorithm based on market mechanisms, the target points are allocated to each robot, which effectively coordinates the behavior among the robots and improves the efficiency. Using the optimized DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm and multi-sensor data to obtain the initial position and orientation of each robot, a multi-robot global coordinate system is established, which optimizes the map fusion process. The experimental results show that this method can successfully complete the multi-robot autonomous exploration and mapping in different environment, and the efficiency is higher than that in single-robot systems. Also, the method is implemented and tested using the Robot Operating System (ROS) framework, so it can be applied to practical scenarios and extended to the scenario of more robots.