Search papers, labs, and topics across Lattice.
2
0
6
2
Finding needles in the haystack of autonomous driving data is harder than we thought: even with fine-tuning, current retrieval methods struggle to identify rare, safety-critical scenarios in the new SearchAD dataset.
Seemingly beneficial architectural choices for end-to-end driving, like high-resolution perception, can actually hinder scalable closed-loop performance, highlighting the need for careful co-design.
