Asynchronous Update Mechanism in Collaborative Driving Simulation: Design and Verification of AACD Framework

This paper introduces AACD, an agent-based asynchronous cooperative driving framework designed to enhance the realism of closed-loop simulations for cooperative driving applications. Traditional frameworks often employ a synchronized blocking approach, where the world remains static while agents act. In contrast, AACD employs a fixed time blocker to allow traffic agents and the simulator to update their states asynchronously yet within a uniform time flow, introducing realistic inconsistencies in cooperative driving scenarios. To validate the framework, we implemented a platooning system as an application example and compared the performance of a baseline longitudinal controller under both synchronous and asynchronous conditions. The experiments showed that asynchronous updates led to irregular oscillations, while synchronous updates produced unrealistically smooth acceleration profiles. The preview of our project code is at: https://github.com/Gaochengzhi/AACD.