An efficient testing route planning method for automatic driving system on public roads based on testing requirements

With the development of automatic driving technology, it is crucial to test and certify the driving ability of automatic vehicles. Public road testing is an essential part of testing and certification. However, public road testing consumes a significant amount of human and material resources, and low efficiency has always been a major limitation of this certification scheme. Because the probability of encountering typical and valuable natural driving scenarios is not high when the tested vehicle is driving in natural traffic flow. At the same time, the lack of efficient planning for the testing route results in the vast majority of testing time being wasted in a 'boring state' without scenario challenges and minimal interaction with surrounding traffic participants. To address this issue, this paper proposes an efficient testing route planning method for automatic driving on public roads based on testing requirements, This method utilizes the collection and annotation of vehicles to large-scale annotate dangerous terrain points and typical natural driving terrain points that are valuable for automatic driving system testing. It further summarizes the characteristics of each terrain point, decomposes the terrain points into multiple feature attributes, formulates appropriate scoring rules based on the requirements, and obtains the value, emphasis, and rating of each terrain point. The results indicate that this method can achieve precise planning of testing routes based on requirements, greatly improving testing efficiency.