MobileVISTA perturbs a robot's base pose and then updates both parts of an imitation-learning example: what the onboard camera should see and what actions the robot should execute.
Synthetic views make robot policies more tolerant of pose errors
MobileVISTA jointly modifies egocentric images and robot actions to train manipulation policies for base poses absent from demonstrations.
Big Tech
Suzannah Wistreich · Stephen Tian · Isabella Huang · Vitor Campagnolo Guizilini · Sergey Zakharov · Katherine Liu · +1 more
Stanford University · Toyota Research Institute
Research Digest··3 min read
Wistreich and colleagues introduce an offline augmentation framework that converts demonstrations recorded at one canonical robot pose into physically consistent examples at perturbed poses.
Why this paper
From Toyota Research Institute and Stanford University
In one line
MobileVISTA generates pose-perturbed training data from single-pose demonstrations to improve mobile manipulation policy robustness.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ✓Limitations stated by the authors (3 noted)
- ·No benchmark numbers found
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§