RLFTSim: Realistic and Controllable Multi-Agent Traffic Simulation via Reinforcement Learning Fine-Tuning
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909055091474432 |
|---|---|
| author | Ahmadi, Ehsan Schofield, Hunter Khamidehi, Behzad Arasteh, Fazel Shan, Jinjun Mou, Lili Bai, Dongfeng Rezaee, Kasra |
| author_facet | Ahmadi, Ehsan Schofield, Hunter Khamidehi, Behzad Arasteh, Fazel Shan, Jinjun Mou, Lili Bai, Dongfeng Rezaee, Kasra |
| contents | Supervised open-loop training has been widely adopted for training traffic simulation models; however, it fails to capture the inherently dynamic, multi-agent interactions common in complex driving scenarios. We introduce RLFTSim, a reinforcement-learning-based fine-tuning framework that enhances scenario realism by aligning simulator rollouts with real-world data distributions and provides a method for distilling goal-conditioned controllability in scenario generation. We instantiate RLFTSim on top of a pre-trained simulation model, design a reward that balances fidelity and controllability, and perform comprehensive experiments on the Waymo Open Motion Dataset. Our results show improvements in realism, achieving state-of-the-art performance. Compared with other heuristic search-based fine-tuning methods, RLFTSim requires significantly fewer samples due to a proposed low-variance and dense reward signal, and it directly addresses the realism alignment issue by design. We also demonstrate the effectiveness of our approach for distilling traffic simulation controllability through goal conditioning. The project page is available at https://ehsan-ami.github.io/rlftsim. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_19033 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | RLFTSim: Realistic and Controllable Multi-Agent Traffic Simulation via Reinforcement Learning Fine-Tuning Ahmadi, Ehsan Schofield, Hunter Khamidehi, Behzad Arasteh, Fazel Shan, Jinjun Mou, Lili Bai, Dongfeng Rezaee, Kasra Robotics Artificial Intelligence Computer Vision and Pattern Recognition Machine Learning Multiagent Systems 68T40, 68T45, 68T42, 68T10, 68T07, 68T05, 68T37, 93A16 I.5; I.6; I.4; I.2; G.3; J.2 Supervised open-loop training has been widely adopted for training traffic simulation models; however, it fails to capture the inherently dynamic, multi-agent interactions common in complex driving scenarios. We introduce RLFTSim, a reinforcement-learning-based fine-tuning framework that enhances scenario realism by aligning simulator rollouts with real-world data distributions and provides a method for distilling goal-conditioned controllability in scenario generation. We instantiate RLFTSim on top of a pre-trained simulation model, design a reward that balances fidelity and controllability, and perform comprehensive experiments on the Waymo Open Motion Dataset. Our results show improvements in realism, achieving state-of-the-art performance. Compared with other heuristic search-based fine-tuning methods, RLFTSim requires significantly fewer samples due to a proposed low-variance and dense reward signal, and it directly addresses the realism alignment issue by design. We also demonstrate the effectiveness of our approach for distilling traffic simulation controllability through goal conditioning. The project page is available at https://ehsan-ami.github.io/rlftsim. |
| title | RLFTSim: Realistic and Controllable Multi-Agent Traffic Simulation via Reinforcement Learning Fine-Tuning |
| topic | Robotics Artificial Intelligence Computer Vision and Pattern Recognition Machine Learning Multiagent Systems 68T40, 68T45, 68T42, 68T10, 68T07, 68T05, 68T37, 93A16 I.5; I.6; I.4; I.2; G.3; J.2 |
| url | https://arxiv.org/abs/2605.19033 |