LibSignal: An Open Library for Traffic Signal Control
arXiv:2211.10649 · doi:10.1007/s10994-023-06412-y
Abstract
This paper introduces a library for cross-simulator comparison of reinforcement learning models in traffic signal control tasks. This library is developed to implement recent state-of-the-art reinforcement learning models with extensible interfaces and unified cross-simulator evaluation metrics. It supports commonly-used simulators in traffic signal control tasks, including Simulation of Urban MObility(SUMO) and CityFlow, and multiple benchmark datasets for fair comparisons. We conducted experiments to validate our implementation of the models and to calibrate the simulators so that the experiments from one simulator could be referential to the other. Based on the validated models and calibrated environments, this paper compares and reports the performance of current state-of-the-art RL algorithms across different datasets and simulators. This is the first time that these methods have been compared fairly under the same datasets with different simulators.
arXiv comment11 pages + 6 pages appendix. Accepted by Machine Learning Journal (2023). A short version is accepted by NeurIPS 2022 Workshop: Reinforcement Learning for Real Life. Website: https://darl-libsignal.github.io/
References in corpus (8)
- Sim-to-Real Transfer of Robotic Control with Dynamics Randomization
- CoLight: Learning Network-level Cooperation for Traffic Signal Control
- CityFlow: A Multi-Agent Reinforcement Learning Environment for Large Scale City Traffic Scenario
- PettingZoo: Gym for Multi-Agent Reinforcement Learning
- IG-RL: Inductive Graph Reinforcement Learning for Massive-Scale Traffic Signal Control
- GeneraLight: Improving Environment Generalization of Traffic Signal Control via Meta Reinforcement Learning
- DynSTGAT: Dynamic Spatial-Temporal Graph Attention Network for Traffic Signal Control
- AttendLight: Universal Attention-Based Reinforcement Learning Model for Traffic Signal Control