PhyCritic: Multimodal Critic Models for Physical AI
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866914323384762368 |
|---|---|
| author | Xiong, Tianyi Wang, Shihao Liu, Guilin Dong, Yi Li, Ming Huang, Heng Kautz, Jan Yu, Zhiding |
| author_facet | Xiong, Tianyi Wang, Shihao Liu, Guilin Dong, Yi Li, Ming Huang, Heng Kautz, Jan Yu, Zhiding |
| contents | With the rapid development of large multimodal models, reliable judge and critic models have become essential for open-ended evaluation and preference alignment, providing pairwise preferences, numerical scores, and explanatory justifications for assessing model-generated responses. However, existing critics are primarily trained in general visual domains such as captioning or image question answering, leaving physical AI tasks involving perception, causal reasoning, and planning largely underexplored. We introduce PhyCritic, a multimodal critic model optimized for physical AI through a two-stage RLVR pipeline: a physical skill warmup stage that enhances physically oriented perception and reasoning, followed by self-referential critic finetuning, where the critic generates its own prediction as an internal reference before judging candidate responses, improving judgment stability and physical correctness. Across both physical and general-purpose multimodal judge benchmarks, PhyCritic achieves strong performance gains over open-source baselines and, when applied as a policy model, further improves perception and reasoning in physically grounded tasks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_11124 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | PhyCritic: Multimodal Critic Models for Physical AI Xiong, Tianyi Wang, Shihao Liu, Guilin Dong, Yi Li, Ming Huang, Heng Kautz, Jan Yu, Zhiding Computer Vision and Pattern Recognition With the rapid development of large multimodal models, reliable judge and critic models have become essential for open-ended evaluation and preference alignment, providing pairwise preferences, numerical scores, and explanatory justifications for assessing model-generated responses. However, existing critics are primarily trained in general visual domains such as captioning or image question answering, leaving physical AI tasks involving perception, causal reasoning, and planning largely underexplored. We introduce PhyCritic, a multimodal critic model optimized for physical AI through a two-stage RLVR pipeline: a physical skill warmup stage that enhances physically oriented perception and reasoning, followed by self-referential critic finetuning, where the critic generates its own prediction as an internal reference before judging candidate responses, improving judgment stability and physical correctness. Across both physical and general-purpose multimodal judge benchmarks, PhyCritic achieves strong performance gains over open-source baselines and, when applied as a policy model, further improves perception and reasoning in physically grounded tasks. |
| title | PhyCritic: Multimodal Critic Models for Physical AI |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2602.11124 |