APF 기법을 이용한 자율주행차량의 운행 적합도 지표 개발

초록

In this paper, we design a driving suitability Indicator based on situational measurement data around autonomous vehicles. Autonomous vehicle affect surrounding vehicles due to driving speed deceleration/acceleration and lane change in process of path tracking to the target position. In order to indicator the impact on surrounding vehicles, the Artificial Potential Field(APF) is constructed using longitudinal/lateral relative velocity and relative distance to the surrounding vehicles in traffic flow. This application quantifies the suitability of autonomous vehicles. The derived driving suitability index can be seen as an indicator of how conservative autonomous vehicles are driving toward surrounding vehicles in traffic flow or whether autonomous vehicles are driving properly. To validate the designed goodness-of-fit indicators, we construct longitudinal, lateral fusion scenarios based on traffic accident case data provided by the KoROAD and AEB tests provided by Euro NCAP. Based on the configured scenario, conservative/nonconservative driving data of many drivers were acquired and used for verification using dSPACE’s Automatic Simulation Model(ASM) Driving simulator. Scenario verification shows numerical suitability for all longitudinal and lateral situations based on the suitability indicators proposed in this paper. In conservative driving situations, it was found that the driving suitability of surrounding vehicles is high compared to non-conservative driving situation.