摘要
首先分析了国内外高精度地图车道模型的研究现状,针对车道模型在自动驾驶中的融合应用,提出了一种基于车道驾驶态势的拓扑构建和路径规划方法。车道驾驶态势是指在遵循正确交通规则的前提下驾驶车辆通过车道后将达到的空间方位和通行状态。利用车道驾驶态势可生成车道级可通行区间集合,并构建可通行区间之间的拓扑关系,再利用Dijkstra算法可进行最优路径规划。可通行区间内部通过驶入和驶出接口点列表达可任意变道的拓扑关系,快速完成车道级路径规划,为自动驾驶多源数据融合与二次规划提供基础参考。
We analyzed the global research about lane model of HAD map in this paper. Aimed at the application of lane model in autonomous driving, we proposed a topology construction and route planning method based on lane driving condition. Lane driving condition refers to space orientation and traffic situation of the vehicle when driving through the lane at the correct traffic rules. Firstly, we used lane driving condition to generate the lane-level range of passable zone, and then built the topological relationship between passable zones. In the end, we used Dijkstra algorithm to optimize route planning. Within the passable zone, the topological relationship could be achieved by entering and exiting interface points to describe that vehicle could change lane in any position of the zone. So lane-level route planning could be quickly completed, which could provide the basic reference for multi-source data fusion and second planning of autonomous driving.
出处
《地理空间信息》
2018年第5期53-56,共4页
Geospatial Information
基金
测绘地理信息公益性行业科研专项资助项目(201512007)
