Cewu Lu

A system called TieBot helps robots learn to knot ties using simulations.

New system enables robots to rearrange objects using visual and language inputs.

DISCO enhances robot capabilities for everyday tasks using advanced scene understanding.

New methods improve robots' ability to grasp objects in messy environments.

DiPGrasp improves robotic grasp planning with speed and flexibility for various grippers.

Kalib simplifies robot calibration without physical markers, improving accuracy and efficiency.

A new framework enhances robot learning using a mix of good and bad data.

A new system improves how machines interact with objects through combined visuals and tactile data.

FoAR enhances robot performance in contact-rich tasks by combining touch and visual data.

Research highlights the challenge of verb understanding in multimodal AI models.

Scientists are decoding human movement dynamics through innovative research.

Exploring how robots learn to interact with changing objects.

Discover how robots improve their skills in delicate object manipulation.

A new method helps robots handle complex objects using superpoints.

New research benchmarks improve understanding of everyday interactions through videos.