Construction method of scenario library based on dual-pillar data source of vehicle and road

In recent years, autonomous driving technology has developed rapidly in the fast lane, but its application and popularization still face many challenges. How to test and verify the autonomous driving system is a technical bottleneck. Scenario is the basis of automatic driving test, which brings a brand-new solution to problems based on mileage test. Therefore, the construction of scenario library is essential. In this paper, a construction method of scenario library based on both of vehicle system and roadside system is proposed. A set of vehicle scenario acquisition system with powerful perception system and a set of roadside scenario acquisition system that can continuously collect data are set up, and massive vehicle data and roadside data are obtained in parallel. Furthermore, this paper designs an automated algorithm for processing massive amounts of data, which processes data from different collection sources and extracts high-value scenarios involving traffic games and dangerous interactions. The results show that the simulation test scenario construction method proposed in this paper based on the dual-pillar data sources of vehicle and road can better reflect the real driving environment and behavior, and can efficiently obtain a large number of valuable scenario data, and quickly build scenario library, which will help the development and application of autonomous driving technology.