Design, Field Evaluation, and Traffic Analysis of a Competitive Autonomous Driving Model in the a Congested Environment

Recently, numerous studies have investigated cooperative traffic systems using the communication between vehicle-to-everything (V2X), which includes both vehicle-to-vehicle and vehicle-to-infrastructures. Unfortunately, if cooperative driving using V2X communication is disabled, there can be a conflict of optimal conditions between various autonomous vehicles. This study assumes a rather pessimistic approach for the transportation system, that is, racing in an urban environment. In South Korea, virtual and live urban autonomous multi-vehicle races were held in March and November of 2021, respectively. In these competitions, each car drove in the congested urban environment while minimizing the transversal time and obeying traffic laws. In this study, we propose a full autonomous driving software stack to deploy a competitive driving model covering module-wise autonomous driving modules. After developing the module-level navigation, perception, and planning systems for the autonomous vehicle, we performed a traffic analysis. Finally, we validated the proposed system at the module level. In addition, we analyzed a model consisting of competitive driving models to determine the similarity of each team's driving log data.