Benchmarks for reinforcement learning in mixed-autonomy traffic

By Eugene Vinitsky, Aboudy Kreidieh, Luc Le Flem, Nishant Kheterpal, Kathy Jang, Cathy Wu, Fangyu Wu, Richard Liaw, Eric Liang, Alexandre M. Bayen. 2018

178 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

The Effects of Reward Misspecification: Mapping and Mitigating Misaligned Models (2022, methodology): “We use the Flow trafﬁc simulator, implemented by Wu et al. (2021) and Vinitsky et al. (2018), which extends the popular SUMO trafﬁc simulator (Lopez et al., 2018).”
D4RL: Datasets for Deep Data-Driven Reinforcement Learning (2020, background): “(Non-RL policies, realism) The Flow benchmark [35] is a framework for studying traffic control using deep reinforcement learning.”
Projection-Based Constrained Policy Optimization (2020, methodology): “(d) Bottleneck task: the agent controls a set of autonomous vehicles (shown in red) in a traffic merge situation and is rewarded for achieving high throughput but constrained to ensure that human-driven vehicles (shown in white) have low speed for no more than 10 seconds (Vinitsky et al., 2018).”
Flow: A Modular Learning Framework for Mixed Autonomy Traffic (2017, methodology): “…was made available online in 2017 [93], several works have employed this framework to achieve results in the directions of the discovery of emergent behaviors in trafﬁc, transfer learning, bottleneck control, the design of trafﬁc control benchmarks, and sim2real [107], [108], [109], [110], [111].”
S4RL: Surprisingly Simple Self-Supervision for Offline Reinforcement Learning (2021, unknown): “, 2017), Flow domain which requires using RL for traffic management (Vinitsky et al., 2018), among others.”