Bridging VLMs and Embodied Intelligence with Deliberate Practice Policy Optimization
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , , , , , , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866909915644166144 |
|---|---|
| author | Zhang, Yi Liu, Che Ren, Xiancong Ni, Hanchu Zhang, Yingji Zhang, Shuai Ding, Zeyuan Hu, Jiayu Shan, Haozhe Qi, Junbo Bai, Yan Li, Dengjie Luo, Jiachen Wang, Yidong Dai, Yong Xu, Zenglin Shen, Bin Wang, Qifan Tang, Jian Ju, Xiaozhu |
| author_facet | Zhang, Yi Liu, Che Ren, Xiancong Ni, Hanchu Zhang, Yingji Zhang, Shuai Ding, Zeyuan Hu, Jiayu Shan, Haozhe Qi, Junbo Bai, Yan Li, Dengjie Luo, Jiachen Wang, Yidong Dai, Yong Xu, Zenglin Shen, Bin Wang, Qifan Tang, Jian Ju, Xiaozhu |
| contents | Developing a universal and versatile embodied intelligence system presents two primary challenges: the critical embodied data bottleneck, where real-world data is scarce and expensive, and the algorithmic inefficiency of existing methods, which are resource-prohibitive. To address these limitations, we introduce Deliberate Practice Policy Optimization (DPPO), a metacognitive ``Metaloop'' training framework that dynamically alternates between supervised fine-tuning (competence expansion) and reinforcement learning (skill refinement). This enables automatic weakness identification and targeted resource allocation, specifically designed to maximize learning efficiency from sparse, finite data. Theoretically, DPPO can be formalised as a unified preference-learning framework. Empirically, training a vision-language embodied model with DPPO, referred to as Pelican-VL 1.0, yields a 20.3% performance improvement over the base model and surpasses open-source models at the 100B-parameter scale by 10.6%. We are open-sourcing both the models and code, providing the first systematic framework that alleviates the data and resource bottleneck and enables the community to build versatile embodied agents efficiently. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2511_16602 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Bridging VLMs and Embodied Intelligence with Deliberate Practice Policy Optimization Zhang, Yi Liu, Che Ren, Xiancong Ni, Hanchu Zhang, Yingji Zhang, Shuai Ding, Zeyuan Hu, Jiayu Shan, Haozhe Qi, Junbo Bai, Yan Li, Dengjie Luo, Jiachen Wang, Yidong Dai, Yong Xu, Zenglin Shen, Bin Wang, Qifan Tang, Jian Ju, Xiaozhu Artificial Intelligence Developing a universal and versatile embodied intelligence system presents two primary challenges: the critical embodied data bottleneck, where real-world data is scarce and expensive, and the algorithmic inefficiency of existing methods, which are resource-prohibitive. To address these limitations, we introduce Deliberate Practice Policy Optimization (DPPO), a metacognitive ``Metaloop'' training framework that dynamically alternates between supervised fine-tuning (competence expansion) and reinforcement learning (skill refinement). This enables automatic weakness identification and targeted resource allocation, specifically designed to maximize learning efficiency from sparse, finite data. Theoretically, DPPO can be formalised as a unified preference-learning framework. Empirically, training a vision-language embodied model with DPPO, referred to as Pelican-VL 1.0, yields a 20.3% performance improvement over the base model and surpasses open-source models at the 100B-parameter scale by 10.6%. We are open-sourcing both the models and code, providing the first systematic framework that alleviates the data and resource bottleneck and enables the community to build versatile embodied agents efficiently. |
| title | Bridging VLMs and Embodied Intelligence with Deliberate Practice Policy Optimization |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2511.16602 |