GuidedVLA
Published:
An RSS 2026 VLA project that specializes action-decoder attention heads for object grounding, temporal skill logic, and spatial geometry.
Published:
An RSS 2026 VLA project that specializes action-decoder attention heads for object grounding, temporal skill logic, and spatial geometry.
Published in Robotics: Science and Systems (RSS) 2026
GuidedVLA explicitly guides VLA action decoders with specialized attention heads for object grounding, temporal skill logic, and spatial geometry.
Resources: Project | Paper | arXiv | Talk | Code | Checkpoint | Dataset | BibTeX
Published in arXiv preprint, 2026
ELAN4D improves VLA policies with embodiment-centric 4D supervision from future robot keypoint tracks, strengthening robustness under out-of-distribution perturbations.
Published:
I was invited to share our algorithm design of Robot Sentry after the first match in national finals. You can watch this interview here.
Published:
I was nominated for the RM Award (Algorithm). As an RM Award candidate, I was invited to share our technical solution. I officially received the RM Award in October 2025. You can watch this sharing here.
Published:
After I received the RM Award 2025, the RoboMaster organizing committee visited Shanghai Jiao Tong University for a personal interview about my RoboMaster journey and our team’s work. You can watch this interview here.
Undergraduate course, University 1, Department, 2014
This is a description of a teaching experience. You can use markdown like any other post.
Workshop, University 1, Department, 2015
This is a description of a teaching experience. You can use markdown like any other post.