复杂场景下移动机器人的SLAM与路径规划问题

SLAM and Path Planning for Mobile Robots in Complex Scenes

针对室内复杂场景下移动式机器人在构建优质地图及定位方面存在较大误差问题,基于Gazebo仿真平台,构建搭载Delta3A激光雷达的室内移动机器人虚拟仿真系统,引入Cartographer、Gmapping及Hector SLAM三种主流算法的建图机理,通过优化Delta3A激光传感器的参数配置,降低三种建图算法所获二值优质地图与原图的匹配偏差。以Cartographer算法构建的优质地图为基础,利用A-star及Dynamic Windows Approach算法进行路径规划,有效地提高了导航精度,同时,大大降低了计算功耗。将本文研究成果在Spark-T实体机器人上进行实物验证,实验结果证明了本文方法的有效性。

Aiming at the problem of large errors in the construction of high-quality maps and positioning of mobile robots in complex indoor scenes,this paper builds built a virtual simulation system of indoor mobile robots equipped with Delta3A lidar based on Gazebo simulation platform.The mapping mechanisms of Cartographer,Gmapping and Hector SLAM was were introduced to reduce the matching deviation between the high-quality binary maps obtained by the three algorithms and the original maps by optimizing the parameter configuration of the Delta3A laser sensor.Based on the high quality maps built by the Cartographer algorithm,the A-star and Dynamic Windows Approach algorithms were used for path planning,which effectively improved navigation accuracy and reduced computational power consumption.Finally,the research results in this paper are were verified on a Spark-T physical robot,and the experimental results prove the effectiveness of the proposed method.