Multi-agent reinforcement learning for Markov routing games: A new modeling paradigm for dynamic traffic assignment

Shou, Zhenyu; Xu, Chen; Fu, Yongjie; Di, Xuan

doi:10.1016/j.trc.2022.103560

Cited by 36 publications

(9 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…States capture situations where the formalized system is in. In contrast to previous studies that separate observations of the decision-maker from the states [24,28,4], we define the states as fully observable information of the system for the decision-maker. When the decision-maker takes an action and gets a reward, the system transits to the next state depending on transition probabilities.…”

Section: Markov Decision Processmentioning

confidence: 99%

Airline dynamic pricing with patient customers using deep exploration-based reinforcement learning

Jo,

Lee,

Moon

2024

Engineering Applications of Artificial Intelligence

View full text Add to dashboard Cite

Section: Markov Decision Processmentioning

confidence: 99%

Airline dynamic pricing with patient customers using deep exploration-based reinforcement learning

Jo,

Lee,

Moon

2024

Engineering Applications of Artificial Intelligence

View full text Add to dashboard Cite

“…Shou et al [ 55 ] considered the approach of the Markov routing game to model the interaction between adaptive agents and traffic congestion. The proposed approach allows the agent to perform repeated interactions with other agents and the traffic environment to learn optimal policies even when they have limited or incomplete information about the environment.…”

Section: Marl For Cavsmentioning

confidence: 99%

A Comprehensive Survey on Multi-Agent Reinforcement Learning for Connected and Automated Vehicles

Yadav

Mishra

Kim

2023

Sensors

View full text Add to dashboard Cite

Connected and automated vehicles (CAVs) require multiple tasks in their seamless maneuverings. Some essential tasks that require simultaneous management and actions are motion planning, traffic prediction, traffic intersection management, etc. A few of them are complex in nature. Multi-agent reinforcement learning (MARL) can solve complex problems involving simultaneous controls. Recently, many researchers applied MARL in such applications. However, there is a lack of extensive surveys on the ongoing research to identify the current problems, proposed methods, and future research directions in MARL for CAVs. This paper provides a comprehensive survey on MARL for CAVs. A classification-based paper analysis is performed to identify the current developments and highlight the various existing research directions. Finally, the challenges in current works are discussed, and some potential areas are given for exploration to overcome those challenges. Future readers will benefit from this survey and can apply the ideas and findings in their research to solve complex problems.

show abstract

“…The number of trips decreases with increasing distance from a city center. The applied Gawron algorithm [24]- [26] allows achieving close to real-world behavior of driver routing. The output of the SUMO simulation is the travel time on each edge during the peak congestion hour.…”

Section: A Experimental Setupmentioning

confidence: 99%

Charge-Arrival-Time Profiles for Long EV Routes

Šaltenis,

Barauskas,

Brilingaitė

et al. 2023

IEEE Access

View full text Add to dashboard Cite

Electric vehicles (EVs) with increasingly large batteries and the rapid development of charging infrastructure enable convenient long-distance electric mobility. Nevertheless, as charging or waiting for an available charger may still take a significant fraction of the travel time, planning long routes is challenging. We argue that, for the plans to be useful, the main challenge lies not in computing an optimal route under some specific assumptions but in managing the inherent uncertainty. In particular, varying and uncertain traffic results not only in similarly uncertain travel times but also in uncertain energy use and, thus, uncertain required charging times. To model long EV routes, we propose Charge-Arrival-Time (CAT) profiles. A CAT profile of a route captures the expected intervals and associated probabilities of arrival time and the arrival charge level of the battery as well as any interdependence of the two. A rich underlying data model is used as a starting point for a stepwise presentation of the mathematical model of CAT profiles. A heuristic algorithm to select charging stops is presented to exemplify their use. Detailed simulations on the road network of Germany show the benefits of precise modeling compared to alternative models.INDEX TERMS Electric vehicle, long-distance EV routing, Charge-Arrival-Time profiles, energy consumption, first-in-first-out, time dependency, data modeling.

show abstract

Multi-agent reinforcement learning for Markov routing games: A new modeling paradigm for dynamic traffic assignment

Cited by 36 publications

References 38 publications

Airline dynamic pricing with patient customers using deep exploration-based reinforcement learning

Airline dynamic pricing with patient customers using deep exploration-based reinforcement learning

A Comprehensive Survey on Multi-Agent Reinforcement Learning for Connected and Automated Vehicles

Charge-Arrival-Time Profiles for Long EV Routes

Contact Info

Product

Resources

About