K-Risk curates 31,398 high-risk driving events from 20 trajectory datasets with multi-layered semantic and LLM-generated annotations validated via closed-loop simulation.
The structured scenario description of ego vehicle 822 and its neighbors, together with the domain-specific system prompt and its five-action schema, form the LLM input (left)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
A knowledge-augmented dataset of high-risk driving scenarios with LLM annotations for autonomous driving
K-Risk curates 31,398 high-risk driving events from 20 trajectory datasets with multi-layered semantic and LLM-generated annotations validated via closed-loop simulation.