Title
Research on the Simulation Test Method of Autonomous Driving Vehicle Based on Cloud Platform
Abstract
A study of autonomous driving simulation under car-following traffic flow was carried out in this paper based on AD Chauffeur simulation cloud platform developed by CATARC. Based on the platform interface, a perception recognition algorithm under the framework of ”YOLO v5 +Deep Sort+ triangular perspective estimation”, a decision algorithm based on finite state machine and a control algorithm based on pure tracking were designed, verified and updated by the cloud platform. The experimental results show that the updated autonomous algorithm based on monocular vision can satisfy the driving demand under car-following traffic flow. With the help of the cloud platform, the development cycle of autonomous driving algorithm can be shortened, and the adaptability among perception, decision-making and control algorithms can be also improved, which is an effective and convenient autonomous driving simulation method.
Year
DOI
Venue
2022
10.1109/ACIRS55390.2022.9845548
2022 7th Asia-Pacific Conference on Intelligent Robot Systems (ACIRS)
Keywords
DocType
ISBN
cloud platform,car following traffic flow,automatic driving,simulation verification
Conference
978-1-6654-8520-3
Citations 
PageRank 
References 
0
0.34
3
Authors
4
Name
Order
Citations
PageRank
Yongwang Shen100.34
Quanshan Hou200.34
Jianfei Wang300.34
Xianglei Zhu400.34