/AI9h ago

Robometer Zero-Shot Reward Model Joins LeRobot For Robot Learning

--0--
Quote posts
Remi Cadene#1841
Reposts
Remi Cadene#1841
Original post
LeRobot@LeRobotHF

🤖 Another zero-shot reward model is now in LeRobot: ROBOMETER.

A general-purpose, zero-shot video-language reward model from @UofSC, @UT_Dallas, @MIT, @UW, @allen_ai, and @nvidia that predicts frame-level task progress.

Trained on 1M+ trajectories from 21 robot embodiments, generalizes zero-shot to unseen tasks, scenes, and robots. 2.4–4.5x better downstream success rates across online RL, offline RL, data filtering, failure detection, and data retrieval for IL.

Project: https://robometer.github.io/ Paper: https://arxiv.org/abs/2603.02115

8:39 AM · Jun 1, 2026 · 23.8K Views
Sentiment
Sentiment building, check back later.
Cluster Engagement
-
Views
-
Comments
-
Reposts
-
Bookmarks
Expand data
Posts from X
Most Activity
Most ActivityTimeline
VIEWS5.5KBOOKMARKS17LIKES61RETWEETS7REPLIES4
Remi Cadene@RemiCadene

The one who solves universal reward model, will solve general purpose robotics

LeRobot@LeRobotHF

🤖 Another zero-shot reward model is now in LeRobot: ROBOMETER.

A general-purpose, zero-shot video-language reward model from @UofSC, @UT_Dallas, @MIT, @UW, @allen_ai, and @nvidia that predicts frame-level task progress.

Trained on 1M+ trajectories from 21 robot embodiments, generalizes zero-shot to unseen tasks, scenes, and robots. 2.4–4.5x better downstream success rates across online RL, offline RL, data filtering, failure detection, and data retrieval for IL.

Project: https://robometer.github.io/ Paper: https://arxiv.org/abs/2603.02115

9hViews 5.5KLikes 61Bookmarks 17