Collaboration Between the City and Machine Learning Community is Crucial to Efficient Autonomous Vehicles Routing
This addresses the critical issue of ensuring efficient and fair urban traffic systems as autonomous vehicles are deployed, highlighting the need for collaboration between city authorities and the ML community.
The paper tackles the problem of autonomous vehicles using multi-agent reinforcement learning for route optimization potentially destabilizing traffic networks and increasing travel times for human drivers, finding that standard algorithms often fail to converge or require long training periods in simulations.
Autonomous vehicles (AVs), possibly using Multi-Agent Reinforcement Learning (MARL) for simultaneous route optimization, may destabilize traffic networks, with human drivers potentially experiencing longer travel times. We study this interaction by simulating human drivers and AVs. Our experiments with standard MARL algorithms reveal that, both in simplified and complex networks, policies often fail to converge to an optimal solution or require long training periods. This problem is amplified by the fact that we cannot rely entirely on simulated training, as there are no accurate models of human routing behavior. At the same time, real-world training in cities risks destabilizing urban traffic systems, increasing externalities, such as $CO_2$ emissions, and introducing non-stationarity as human drivers will adapt unpredictably to AV behaviors. In this position paper, we argue that city authorities must collaborate with the ML community to monitor and critically evaluate the routing algorithms proposed by car companies toward fair and system-efficient routing algorithms and regulatory standards.