基于ROS的多传感器融合巡检机器人系统研究
Research on the Design of ROS Inspection Robot Based on Multi-Sensor Fusion

商场、写字楼等室内场所经常存在异物堆放和发生火灾烟雾的情况，这些室内场所作为重要的活动区域，为了消除存在的安全隐患。需要经常对商场、写字楼等室内场所进行巡检，然而人工巡检效率低下，且不能全天候巡检。智能巡检机器人能够高效、全天候对巡检区域进行巡检。设计研究基于多传感器融合的ROS巡检机器人，该巡检机器人包括对激光雷达数据的预处理，基于粒子滤波的Gmapping算法实现建图，AMCL用于室内巡检机器人的重定位。结合A^*全局路径规划算法和DWA局部路径规划算法共同实现室内巡检机器人的导航避障。深度相机结合改进YOLOv7检测算法对室内是否存在火灾烟雾与行人依靠电梯等安全隐患进行巡检。实验表明：该室内巡检机器人具有良好的建图、导航避障、视觉检测的功能。可见研究成果对室内巡检机器人的研究具有一定参考价值。
Shopping malls, office buildings and other indoor places often have foreign objects piled up and fire smoke occur, these indoor places as an important activity area, in order to eliminate the potential safety hazards. It is necessary to frequently inspect indoor places such as shopping malls and office buildings, but manual inspection is inefficient and cannot be inspected around the clock. The intelligent inspection robot can efficiently and round-the-clock inspect the inspection area. The ROS inspection robot based on multi-sensor fusion is designed and studied, which includes the preprocessing of lidar data, the mapping algorithm based on particle filtering, and the AMCL is used for the relocation of the indoor inspection robot. Combined with the global path planning algorithm and the DWA local path planning algorithm, the navigation and obstacle avoidance of indoor inspection robots are realized. The depth camera combined with the improved YOLOv7 detection algorithm inspects whether there are potential safety hazards such as fire, smoke and pedestrians relying on elevators in the room. Experiments show that the indoor inspection robot has good functions of mapping, navigation and obstacle avoidance, and visual detection. The research results have a certain reference value for the research of indoor inspection robots.