首页|面向模拟环境的自动驾驶道路场景生成方法研究

面向模拟环境的自动驾驶道路场景生成方法研究

扫码查看
针对生成自动驾驶系统道路场景的问题,提出面向模拟环境的道路场景生成方法。该方法定义并实现了八类道路组件,使用基于指导的道路组件拼接方法将道路组件拼接为道路场景,根据道路场景的拓扑结构计算道路场景之间的相似度完成去重。去重后的道路场景在自动驾驶系统与模拟器中开展联合仿真。实验表明,相较于基于随机拼接的道路场景生成方法,该方法能够生成可以直接应用于自动驾驶系统测试的多样化的道路场景,弥补了几何与拓扑结构多样性不足的局限性问题。
AUTONOMOUS DRIVING ROAD SCENE GENERATION METHOD FOR SIMULATED ENVIRONMENT
Aiming at the problem of generating road scenes for autonomous driving systems,this paper proposes a method to generate road scenes for simulation environment.This method defined and implemented eight types of road components,and used a guided road component composition method to compose the road components into road scenes.At the same time,the similarity between the road scenes was calculated according to the topological structure of the road scenes to complete the deduplication.The deduplicated road scene was co-simulated in the autonomous driving system and the simulator.Experiments show that compared with the road scene generation method based on random composition,the proposed method can generate road scenes that can be directly applied to the testing of autonomous driving systems,and make up for the limitation of insufficient diversity of geometric and topological structures of road scenes.

Autonomous drivingRoad scene generationSimulated environment

杨范、陈碧欢、赵文耘

展开 >

复旦大学计算机科学技术学院 上海 200433

自动驾驶 道路场景生成 模拟环境

国家自然科学基金项目

62372114

2024

计算机应用与软件
上海市计算技术研究所 上海计算机软件技术开发中心

计算机应用与软件

CSTPCD北大核心
影响因子:0.615
ISSN:1000-386X
年,卷(期):2024.41(8)