A Study on the Driving Performance Analysis for Autonomous Vehicles Through the Real-Road Field Operational Test Platform

Tests for autonomous vehicles (AVs) and vehicle components are conducted in various ways in a laboratory, a virtual world, and a proving ground. However, since the aforementioned evaluation has limitations for reproducing scenarios of specific edge cases and unintended situations, Field Operational Test (FOT) for autonomous vehicles on the actual road is still necessary. Due to the higher automation levels of autonomous vehicles should be able to drive in more diverse environments without the intervention of the driver, FOT is particularly essential. In this study, we built D-Live, an autonomous vehicle FOT platform in Daegu metropolitan city, to evaluate self-driving in urban environments. We also developed an analysis system to evaluate the performance of autonomous vehicles and automotive sensors. The D-Live platform can evaluate autonomous vehicles by reflecting real-time driving environment changes, weather, and road conditions. In addition, the D-Live platform can quantitatively evaluate the FOT result by processing the data collected from autonomous vehicles and roadside sensors. The processed data is analyzed using the predefined Operational Design Domain (ODD), Object and Event Detection and Response (OEDR), and use cases. This paper introduces the D-Live platform to evaluate autonomous vehicles in urban areas and FOT evaluation methods.