Traffic congestion has severe impacts on the productivity of society due to long travel times, on the economics due to wasted gallons of fuel and electrical energy, and on the environment due to harmful CO2 emissions. In the European Union (EU), congestion costs nearly EUR 100 billion, or 1% of the EU’s GDP, annually. However, intersections are the bottlenecks of traffic networks. They constitute a significant part of this issue, and therefore, can also play a crucial role in mitigating it.
Classical adaptive solutions such as SCATS can achieve a 28% reduction in travel time, 12% reduction in fuel consumption, and 15% reduction in emissions compared to non-adaptive ones. However, they rely on stylist models with (strong) assumptions, e.g., uniform traffic flow, for mathematical tractability. Such models are prone to fail due to accidents, weather, and road conditions. Therefore, we can improve the travel quality further through an artificial intelligence (AI) based solution without making such model assumptions.
There are several challenges for AI-based solutions in intersection control. For example, traffic networks are safety-critical systems; however, robustness and reliability are weaknesses of AI-based solutions. The scale of urban traffic networks is massive and time-critical, which causes a significant challenge for computational solutions. Drivers also react and adapt dynamically by finding the shortest-time travel route. This leads to counter-factual outcomes in the traffic flow.
This project will address all these challenges comprehensively through an interdisciplinary research agenda bringing technical concepts from AI, transportation theory, control theory, and game theory. We will design strategic, scalable, and safe (S3) multi-intersection control mechanisms for urban traffic networks both for the classical traffic lights and the autonomous intersection management systems, foreseen to replace traffic lights soon.
- M. O. Sayin, C.-W. Lin, S. Shiraishi, “Managing Roadway Intersections for Vehicles,” US Patent App. 15/924,979, 2019. [Patent]
- M. O. Sayin, C.-W. Lin, E. Kang, S. Shiraishi, and T. Başar. “Reliable Smart Road Signs,” IEEE Trans. Intell. Transp. Syst., 2020.
- M. O. Sayin, C.-W. Lin, S. Shiraishi, J. Shen, and T. Başar, “Information-driven Autonomous Intersection Control via Incentive Compatible Mechanisms,” IEEE Trans. Intell. Transp. Syst., 2019.