Abstract: Autonomous driving generally consists of perception, decision-making, planning, and control modules. With the deepening of research, the role of decision-making and planning in the system becomes more and more obvious. The first thing to be solved in the planning process is to deal with various traffic scenarios. Come up with a reasonable decision-making plan. This paper proposes a design idea that separates path planning and velocity planning for real-time path planning for autonomous driving. Firstly, based on the high-precision map of a closed scene, a number of alternative paths are generated using sampling methods, and the best path is selected in combination with collision detection. Secondly, the generated path is planned based on the comfort level to provide a speed plan that satisfies the comfort, and finally passed Simulation verifies the effectiveness of this method.

Key words: Intelligence; Local route planning; Speed plan

摘要： 自动驾驶系统一般由感知，决策规划和控制等模块组成，随着研究的深入，决策规划 在系统中的作用越来越明显，在规划过程中首先需要解决的是针对各种交通场景做出合理的 决策方案。文章提出将路径规划和速度规划分开的设计思路用于自动驾驶的实时路径规划。 首先基于封闭场景下采集的高精地图，利用采样方法生成若干可供选择的路径，并结合碰撞 检测选择最佳路径；其次基于舒适度对生成的路径进行速度规划；最后通过实车测试验证该 方法的有效性。

关键词: 自动驾驶；局部路径规划；速度规划