基于改进光线追踪算法的自动驾驶模拟路试系统  

Autonomous driving simulation system based on improved ray tracing algorithm

作　　者：张博为 崔鹏 ZHANG Bo-wei;CUI Peng(School of Information,Guizhou University of Finance and Economics,Guiyang 550025,China)

机构地区：[1]贵州财经大学信息学院,贵州贵阳550025

出　　处：《计算机工程与设计》2022年第11期3279-3286,共8页Computer Engineering and Design

基　　金：贵州省科技计划基金项目(黔科合基础[2018]1022号);贵州财经大学在校学生科学研究基金项目(2020ZXSY52)。

摘　　要：现有自动驾驶模拟路试系统对车载激光雷达等传感器的模拟效果不佳、部署环境受限制较大,影响了自动驾驶仿真技术在真实应用场景中的应用。对此,提出一种基于光线追踪算法的自动驾驶模拟路试系统架构,通过使用改进光线追踪算法实现对激光雷达的高效实时仿真,通过分离模拟路试模块和算法部署模块的功能,提升整个系统的并发能力。实验结果表明,在对激光雷达的仿真上,该方法相较传统的基于实时渲染技术的自动驾驶仿真架构平均提升了1倍的帧率和最高6倍的点云输出数量,且方便进行集群部署。Currently,the simulation effect of sensors such as LiDAR is not good,and the limited deployment environment affects the simulation of real-world application scenarios.Therefore,an automatic driving simulation road test system architecture based on ray tracing algorithm was proposed.In this architecture,the efficient real-time simulation of LiDAR was realized using the improved ray tracing algorithm,and the concurrency of the whole system was improved by separating the functions of the simulation road test module and the algorithm deployment module.Experimental results show that in the simulation of LiDAR,the proposed method improves the frame rate by an average of 1 times and improves the number of point cloud output by 6 times maximum compared with the traditional automatic driving simulation architecture based on real-time rendering technology.It is more convenient for cluster deployment.

关 键 词：自动驾驶 模拟路试系统 光线追踪算法 边缘计算 实时渲染技术 激光雷达 仿真传感器 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]