PEAfowl: Perception-Enhanced Multi-View Vision-Language-Action for Bimanual Manipulation
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_ | 1866910000752885760 |
|---|---|
| author | Fan, Qingyu Li, Zhaoxiang Lu, Yi Chen, Wang Shen, Qiu Long, Xiao-xiao Cai, Yinghao Lu, Tao Wang, Shuo Cao, Xun |
| author_facet | Fan, Qingyu Li, Zhaoxiang Lu, Yi Chen, Wang Shen, Qiu Long, Xiao-xiao Cai, Yinghao Lu, Tao Wang, Shuo Cao, Xun |
| contents | Bimanual manipulation in cluttered scenes requires policies that remain stable under occlusions, viewpoint and scene variations. Existing vision-language-action models often fail to generalize because (i) multi-view features are fused via view-agnostic token concatenation, yielding weak 3D-consistent spatial understanding, and (ii) language is injected as global conditioning, resulting in coarse instruction grounding.
In this paper, we introduce PEAfowl, a perception-enhanced multi-view VLA policy for bimanual manipulation. For spatial reasoning, PEAfowl predicts per-token depth distributions, performs differentiable 3D lifting, and aggregates local cross-view neighbors to form geometrically grounded, cross-view consistent representations. For instruction grounding, we propose to replace global conditioning with a Perceiver-style text-aware readout over frozen CLIP visual features, enabling iterative evidence accumulation. To overcome noisy and incomplete commodity depth without adding inference overhead, we apply training-only depth distillation from a pretrained depth teacher to supervise the depth-distribution head, providing perception front-end with geometry-aware priors.
On RoboTwin 2.0 under domain-randomized setting, PEAfowl improves the strongest baseline by 23.0 pp in success rate, and real-robot experiments further demonstrate reliable sim-to-real transfer and consistent improvements from depth distillation.
Project website: https://peafowlvla.github.io/. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_17885 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | PEAfowl: Perception-Enhanced Multi-View Vision-Language-Action for Bimanual Manipulation Fan, Qingyu Li, Zhaoxiang Lu, Yi Chen, Wang Shen, Qiu Long, Xiao-xiao Cai, Yinghao Lu, Tao Wang, Shuo Cao, Xun Computer Vision and Pattern Recognition Artificial Intelligence Robotics Bimanual manipulation in cluttered scenes requires policies that remain stable under occlusions, viewpoint and scene variations. Existing vision-language-action models often fail to generalize because (i) multi-view features are fused via view-agnostic token concatenation, yielding weak 3D-consistent spatial understanding, and (ii) language is injected as global conditioning, resulting in coarse instruction grounding. In this paper, we introduce PEAfowl, a perception-enhanced multi-view VLA policy for bimanual manipulation. For spatial reasoning, PEAfowl predicts per-token depth distributions, performs differentiable 3D lifting, and aggregates local cross-view neighbors to form geometrically grounded, cross-view consistent representations. For instruction grounding, we propose to replace global conditioning with a Perceiver-style text-aware readout over frozen CLIP visual features, enabling iterative evidence accumulation. To overcome noisy and incomplete commodity depth without adding inference overhead, we apply training-only depth distillation from a pretrained depth teacher to supervise the depth-distribution head, providing perception front-end with geometry-aware priors. On RoboTwin 2.0 under domain-randomized setting, PEAfowl improves the strongest baseline by 23.0 pp in success rate, and real-robot experiments further demonstrate reliable sim-to-real transfer and consistent improvements from depth distillation. Project website: https://peafowlvla.github.io/. |
| title | PEAfowl: Perception-Enhanced Multi-View Vision-Language-Action for Bimanual Manipulation |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence Robotics |
| url | https://arxiv.org/abs/2601.17885 |