An Improved Frontier Point Exploration System for Mobile Robots Based on LiDAR SLAM

Robots rely heavily on environmental maps that are usually built by manually operated simultaneous localization and mapping (SLAM) systems. In scenarios such as disaster rescue and enemy probing where operators face accessibility difficulties, mobile robots are expected to use active SLAM to autonomously explore unknown environments. In this work, we establish a three-part framework consisting of environment modeling, goal determination, and motion planning for an autonomous mobile robot exploration system. We thoroughly analyze the key technologies of each part and refine the framework by analyzing existing systems. Then, we investigate the working principle of the Cartographer and implement a series of improvements to enable the built maps to meet the application requirements of the frontier point search method, which significantly enhances the scale and edge accuracy of environment modeling. The resulted scale deviation under the Cartographer system is only 0.39%, much lower than that with the pre-improvement scheme.