Development of Robust Validation Method through Driverless Test for Vision-based Oncoming Vehicle Collision Avoidance System

This paper presents a robust validation test method to prove and improve the oncoming vehicle collision avoidance systems. This system is developed to reduce traffic accidents while a driver’s vehicle crashes into an oncoming vehicle out of its lane. It consists of a front camera on the windshield to detect the vehicle on the other side and an electric power steering to control the host vehicle to prevent head-on collision. It requires high performance for planning and controlling accurate avoidance path at close distance at the right time. Thus, safe and accurate validation vehicle test method is essential to develop the high quality system and determine the performance. The proposed validation test method includes robust vehicle test scenarios with test specification, vehicle test equipment based on automated robots and detailed analysis method for system performance. The robust test scenarios with test specification are developed to prove robustness of the system and seek weakness points from diverse conditions on the road and test specifications. The vehicle robots based on GPS/INS are utilized to conduct validation tests safely, repeatedly and accurately as the designed the test scenarios. Overall, the suggested analysis method determines the reliability of robot tests by the error distribution of the dynamics and avoidance performance of the system through estimating the internal errors of the components which cause errors of the avoidance system. The vehicles tests using presented robust test scenarios and driverless tests based on vehicle robots are conducted repeatedly to prove the robustness of the system thoroughly. The results of vehicle tests show the proposed method is powerful to validate the system and present proper value of design parameters applied to optimal avoidance path to improve the performance under the robust environment.