Salvato in:
| Autori principali: | , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2512.13262 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866914201802375168 |
|---|---|
| author | Seong, Hyunki Lee, Jeong-Kyun Myeong, Heesoo Shin, Yongho Cho, Hyun-Mook Kim, Duck Hoon Desai, Pranav Surana, Monu |
| author_facet | Seong, Hyunki Lee, Jeong-Kyun Myeong, Heesoo Shin, Yongho Cho, Hyun-Mook Kim, Duck Hoon Desai, Pranav Surana, Monu |
| contents | Learning interactive motion behaviors among multiple agents is a core challenge in autonomous driving. While imitation learning models generate realistic trajectories, they often inherit biases from datasets dominated by safe demonstrations, limiting robustness in safety-critical cases. Moreover, most studies rely on open-loop evaluation, overlooking compounding errors in closed-loop execution. We address these limitations with two complementary strategies. First, we propose Group Relative Behavior Optimization (GRBO), a reinforcement learning post-training method that fine-tunes pretrained behavior models via group relative advantage maximization with human regularization. Using only 10% of the training dataset, GRBO improves safety performance by over 40% while preserving behavioral realism. Second, we introduce Warm-K, a warm-started Top-K sampling strategy that balances consistency and diversity in motion selection. Our Warm-K method-based test-time scaling enhances behavioral consistency and reactivity at test time without retraining, mitigating covariate shift and reducing performance discrepancies. Demo videos are available in the supplementary material. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2512_13262 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Post-Training and Test-Time Scaling of Generative Agent Behavior Models for Interactive Autonomous Driving Seong, Hyunki Lee, Jeong-Kyun Myeong, Heesoo Shin, Yongho Cho, Hyun-Mook Kim, Duck Hoon Desai, Pranav Surana, Monu Robotics Computer Vision and Pattern Recognition Learning interactive motion behaviors among multiple agents is a core challenge in autonomous driving. While imitation learning models generate realistic trajectories, they often inherit biases from datasets dominated by safe demonstrations, limiting robustness in safety-critical cases. Moreover, most studies rely on open-loop evaluation, overlooking compounding errors in closed-loop execution. We address these limitations with two complementary strategies. First, we propose Group Relative Behavior Optimization (GRBO), a reinforcement learning post-training method that fine-tunes pretrained behavior models via group relative advantage maximization with human regularization. Using only 10% of the training dataset, GRBO improves safety performance by over 40% while preserving behavioral realism. Second, we introduce Warm-K, a warm-started Top-K sampling strategy that balances consistency and diversity in motion selection. Our Warm-K method-based test-time scaling enhances behavioral consistency and reactivity at test time without retraining, mitigating covariate shift and reducing performance discrepancies. Demo videos are available in the supplementary material. |
| title | Post-Training and Test-Time Scaling of Generative Agent Behavior Models for Interactive Autonomous Driving |
| topic | Robotics Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2512.13262 |