CVPR 2026 Workshop Competition
The GigaBrain Challenge 2026 aims to push the boundaries of embodied intelligence. We invite researchers and practitioners to explore cutting-edge topics including Vision-Language-Action (VLA) models, Embodied Foundation Models, World Models, and Reinforcement Learning.
This workshop will feature invited talks, paper presentations, and a rigorous competition across three tracks. We are accepting paper submissions on relevant topics and will host approximately 20 poster boards for selected works.
Note:
The competition will conclude by May 15, 2026. No onsite competition will be held at CVPR; the workshop session will focus on awards ceremonies and presentations by winning teams.
We welcome submissions related to VLA, embodied foundation models, world models, and reinforcement learning. Accepted papers will be presented in the workshop and selected works will be assigned poster boards. We will select a Best Paper Award.
OpenReview SubmissionEvaluate VLA performance in simulation environments.
Timeline:
Mid-March to Mid-April
Evaluate the performance of World Models acting as VLA evaluators.
Timeline:
Mid-March to Mid-April
Evaluate VLA performance on real-world robots.
Timeline:
Mid-April to Mid-May
If the WeChat QR code expires, please email jeffwang@mail.tsinghua.edu to request access to the WeChat group.
Track 1 (Simulation) and Track 2 (World Model) begin.
OpenReview submission portal opens ( OpenReview ).
Deadline for paper submissions (timezone: AoE, UTC-12).
Track 1 & 2 conclude. Track 3 (Real Robot) begins with invited teams.
All tracks conclude. Final rankings determined.
Awards ceremony and presentations by winning teams in Denver, USA.