Modeling proactive effects of connected autonomous vehicles on urban traffic in adverse weather

Adverse weather conditions significantly degrade the environmental perception capabilities of autonomous vehicles (AVs), thereby compromising both traffic safety and operational efficiency. Connected autonomous vehicles (CAVs), leveraging vehicle-to-vehicle (V2V) communication technology, have the potential to mitigate these challenges through cooperative perception mechanisms. Before large-scale deployment of CAVs, it is essential to understand the significant impacts of CAV application on urban traffic characteristics, especially in adverse weather conditions. However, building a realistic simulation for CAV traffic system in adverse weather conditions can be challenging. On the one hand, adverse weather, with chaotic atmosphere behaviors and rapid complex interactions with electromagnetic waves, imposes unpredictable effects on automotive sensors. On the other hand, the dynamic interplay between sensor physics, communication networks, and multi-agent data fusion contributes to uncertainty in CAV driving decisions. To address the challenges, this paper firstly introduces radar theories and builds a physics-based model to realistically simulate weather impacts on sensors at scale. Then, a novel simulation model is proposed for CAV traffic system in rainy conditions, which includes weather-related degraded sensor, unreliable V2V communication, and cooperative perception-based decision making module. Finally, simulations in different levels of rainy conditions are conducted based on a large-scale road network (in the City of Luxembourg) with real traffic data. Results show that CAVs are more effective in improving traffic safety and efficiency under challenging weather conditions. The limits of CAVs in adverse weather are also discussed.