Reproduce “OmniSpace: Efficient Geometry Awareness for Autonomous Vehicles MLLMs” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Hao Vo, Phu Loc Nguyen, Khoa Vo, Sieu Tran, Duc Minh Nguyen, Ngo Xuan Cuong, Nghi D. Q. Bui, Anh Nguyen and 2 more
OmniSpace trains a multi-view driving model to encode camera geometry through Plücker rays, epipolar attention, and training-time distillation from VGGT, which is removed at inference. On Bench2Drive, Qwen2.5-VL 7B OmniSpace achieves a 79.65 Driving Score and 60.15% Success Rate, exceeding SpaceDrive’s 78.02 and 55.11%, but not the benchmark’s unconditional highest score.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.