/Tech4h ago

HUG Pipeline Predicts Human Grasps and Retargets Them to Robot Hands

41200925

Original post

In inference time, you can retarget a humanoid hand to match the predicted human hand.

We have tested out the WUJI and Inspire hands and it works quite well for those. Our industry partners have reported successes on their hardware as well.

Lerrel Pinto@LerrelPinto

To model these grasps, we use both RGB and Depth as input and predict the MANO hand pose of the robot.

We needed multiple iterations to nail down the predictions in high-DOF hand space, and hope this architecture is helpful to others trying to predict full humanoid hand pose.

9:54 AM · Jun 16, 2026 · 150 Views

Sentiment

Users are excited about extending the new humanoid hand pose model to long-horizon behavior in the wild because it enables realtime object pick-place tasks.

Pos

100.0%

Neg

0.0%

1 comments with sentiment.

Cluster Engagement

Digg Deeper

No Digg Deeper questions have been answered for this story yet.

Related links

HUG: Human Universal Grasping

GRASPING.IOVia

#749

Posts from X

Most Activity

VIEWS266LIKES3

Lerrel Pinto@LerrelPinto

This project is @kevin_y_wu 's brainchild, with support from @Will_Zhou2025 , Isaac Tu, @BillyYYan , @irmakkguzey , David Fouhey, @DandanShan_ .

More videos, full papers, and code is here: https://grasping.io/

Lerrel Pinto@LerrelPinto

Something we are pretty excited about next is to go for more long-horizon behavior in the wild. Here is HUG running in realtime for object pick-place.

4h26630

REPLIES1

Lerrel Pinto@LerrelPinto

We ran HUG on several AirBnB scenes, without any tuning. It just works out of the box. ~60-70% success rate.

Lerrel Pinto@LerrelPinto

In inference time, you can retarget a humanoid hand to match the predicted human hand.

We have tested out the WUJI and Inspire hands and it works quite well for those. Our industry partners have reported successes on their hardware as well.

4h13520

Lerrel Pinto@LerrelPinto

Something we are pretty excited about next is to go for more long-horizon behavior in the wild. Here is HUG running in realtime for object pick-place.

Lerrel Pinto@LerrelPinto

To probe a bit deeper into humanoid manipulation, we built a simulation benchmark as well with real object counter parts.

This should hopefully help the humanoid hand community benchmark results for apples to apples comparisons.

4h24730

Lerrel Pinto@LerrelPinto

To probe a bit deeper into humanoid manipulation, we built a simulation benchmark as well with real object counter parts.

This should hopefully help the humanoid hand community benchmark results for apples to apples comparisons.

Lerrel Pinto@LerrelPinto

We ran HUG on several AirBnB scenes, without any tuning. It just works out of the box. ~60-70% success rate.

4h12720