Skip to Content
All Research

Research

ViPRA: Video Prediction for Robot Actions

Routray, Pan, Jain, Bahl & Pathak, ICLR 2026

Input
Placeholder: what goes into the system.
Supervision Required
Placeholder: how much supervision/labeling the method needs.
Representation Learned
Placeholder: the representation learned.
How It Connects to Robot Actions
Placeholder: how the representation connects to robot actions.
Real Hardware Test
Placeholder: validated on real hardware, or sim-only.
Limitations / Failure Modes
Placeholder: limitations / failure modes.
Experiment I Would Add
Placeholder: the experiment I would add.