Reinforcement learning encourages drivers to reduce traffic and emissions
Learning to Incentivise: Using Reinforcement Learning for Sustainable Urban Mobility
Multiagent SystemsMachine Learning
Summary
Traffic management often tries tolls to control congestion, but these can upset drivers or be banned. The authors use reinforcement learning to give drivers incentives to take less crowded routes, even if longer, reducing total travel time and emissions. Their system learns how to best offer rewards dynamically to improve overall traffic flow and air quality. They found that focusing only on time or emissions can worsen the other, but balancing both leads to benefits in traffic and pollution.
What this means in practice
- •For city traffic planners: Adopt AI-based incentive schemes to reduce congestion and pollution by dynamically encouraging drivers to use alternate routes.
- •For fleet management companies: Incorporate dynamic incentives to optimize route choices, lowering operating times and carbon footprint across vehicle fleets.
- •For ride-sharing platform operators: Use adaptive reward systems to influence drivers’ route decisions for improved urban traffic conditions and environmental impact.$Commercial implications: The paper enables incentive-driven routing features that platforms can sell to cities for smarter transportation management.
Authors
Germán Pardo-González, Shaghayegh Vosough, Katerina Papadaki, Claudio Roncoli
Abstract
Traffic management has traditionally focused on toll-based and road pricing solutions. However, road pricing often raises concerns about accessibility and public dissatisfaction, leading to its prohibition in some places, such as Finland. This study optimises the dynamic allocation of incentives to drivers, encouraging them to reroute onto alternative (potentially longer) paths to achieve greater societal benefit, reduced total travel time (TTT) and total emissions in the transportation network, contributing to sustainable urban mobility. We employ a multi-agent reinforcement learning approach to dynamically assign incentives to drivers to reduce both TTT and emissions, with travel times estimated using traffic simulation software. We demonstrate that, with an unlimited budget and an objective of minimising travel time, the incentive scheme reduces TTT by 16%, compared to the dynamic User-Equilibrium (UE) with a budget equivalent to about 11% of the UE total time. When the goal is to minimise emissions, a 9% reduction in CO2 emissions is observed under an unlimited budget. We demonstrate a critical trade-off: minimising TTT leads to an increase in emissions, while prioritising emission reductions raises TTT. However, with the right combination of weights in the multi-objective function, both TTT and total emissions are improved beyond the baseline.