基于图像拼接的室外环境地面特征地图创建

Image Alignment Based Outdoor Ground Feature Mapping

为获取基于地图匹配的定位算法所要求的环境地图,提出了一种室外环境下基于图像拼接的地面特征地图创建方法.创建过程分为环境探索以及离线地图生成.在环境探索中,手动控制机器人在环境中漫游,实时保存有关的传感器数据,如GPS、里程计以及视频数据.在离线地图生成中,针对传感器误差导致地图不精确的问题,采用了地图分块存储方式以及基于图像拼接的优化算法来提高地图精度.仿真及实验证明了该地图创建方法的有效性.

An image alignment based outdoor ground feature mapping approach was proposed to obtain a map for map matching based localization algorithm. The mapping approach includes two parts： environment exploration and offline map building. In environment exploration, robot is manually guided to explore the environment and capture the data of sensors like GPS, odometry and camera. In offline map building, with the captured sensors data, a map optimization algorithm basing on map block and image alignment was presented to address the sensors error problem in mapping and improve the map accuracy. The results with both synthetic and real experiments prove the effectiveness of the proposed approach.