The authors continued pretraining a video foundation model on a large, diverse collection of manipulation videos.
Better video predictions improve robots’ performance on unfamiliar manipulation tasks
VPP2 combines manipulation-focused video pretraining with action learning while preserving the video model’s ability to generalize.
Top University
Yanjiang Guo · Haodong Yan · Zhide Zhong · Zhongru Zhang · Qingyuan Yang · Qingzhou Lu · +12 more
Robotera · Tsinghua University · HKUST (GZ) · University of California, Berkeley · Shanghai Jiaotong University
Research Digest··2 min read
Guo et al.
Why this paper
From Tsinghua University and 4 others
In one line
Video Prediction Policy 2 achieves strong zero-shot generalization in video prediction and action generation for robot manipulation.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ·No stated limitations found
- ✓Reports numbers on named benchmarks (5 benchmarks)
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§