Research map / Reinforcement learning / Visual and representation RL
OAM: An Option-Action Reinforcement Learning Framework for Universal Multi-Intersection Control
Enming Liang, Zicheng Su, Chilin Fang, Renxin Zhong
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use OAM instead of the strongest listed baseline, MPLight | Nanchang area 1, traffic volume 2,000 · Nanchang area 1, traffic volume 2,000 | average travel time (s): 410 -> 371 | Improved |
| Use OAM instead of the strongest listed baseline, MPLight | Nanchang area 2, traffic volume 2,000 · Nanchang area 2, traffic volume 2,000 | average travel time (s): 323 -> 268 | Improved |
| Use OAM instead of the strongest listed baseline, MPLight | Nanchang area 3, traffic volume 2,000 · Nanchang area 3, traffic volume 2,000 | average travel time (s): 436 -> 403 | Improved |
| Use OAM instead of the strongest listed baseline, AttendLight | Nanchang area 1, traffic volume 4,000 · Nanchang area 1, traffic volume 4,000 | average travel time (s): 407 -> 405 | Improved |
| Use OAM instead of the strongest listed baseline, MPLight | Nanchang area 2, traffic volume 4,000 · Nanchang area 2, traffic volume 4,000 | average travel time (s): 347 -> 327 | Improved |
| Use OAM instead of the strongest listed baseline, AttendLight | Nanchang area 3, traffic volume 4,000 · Nanchang area 3, traffic volume 4,000 | average travel time (s): 546 -> 489 | Improved |
| Use OAM instead of the strongest listed baseline, AttendLight | Nanchang area 1, traffic volume 6,000 · Nanchang area 1, traffic volume 6,000 | average travel time (s): 387 -> 392 | Hurt |
| Use OAM instead of the strongest listed baseline, MPLight | Nanchang area 2, traffic volume 6,000 · Nanchang area 2, traffic volume 6,000 | average travel time (s): 381 -> 364 | Improved |
| Use OAM instead of the strongest listed baseline, AttendLight | Nanchang area 3, traffic volume 6,000 · Nanchang area 3, traffic volume 6,000 | average travel time (s): 538 -> 513 | Improved |
| Add cell-based state representation (ablation) | OAM on Hangzhou · Hangzhou | average travel time | Improved |
| Replace the regularized-delay reward with pressure-based reward (ablation) | OAM on Hangzhou · Hangzhou | average travel time | Hurt |
| Add the outflow-based regularizer (ablation) | OAM on Hangzhou · Hangzhou | convergence rate | Improved |
Closest papers by meaning
- IntersectionZoo: Eco-driving for Benchmarking Multi-Agent Contextual Reinforcement Learning (ICLR 2025 · similarity 0.86)
- π-Light: Programmatic Interpretable Reinforcement Learning for Resource-Limited Traffic Signal Control (AAAI 2024 · similarity 0.86)
- DiffLight: A Partial Rewards Conditioned Diffusion Model for Traffic Signal Control with Missing Data (NeurIPS 2024 · similarity 0.86)
- CoLLMLight: Cooperative Large Language Model Agents for Network-Wide Traffic Signal Control (ICLR 2026 · similarity 0.86)
- Prompt to Transfer: Sim-to-Real Transfer for Traffic Signal Control with Prompt Learning (AAAI 2024 · similarity 0.85)
- RobustLight: Improving Robustness via Diffusion Reinforcement Learning for Traffic Signal Control (ICML 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
