A decoupled offline-online framework uses LLMs and latent diffusion models to generate fault scenarios for testing edge-based lane-following models, revealing large robustness drops under conditions like fog.
Lanevil: Benchmarking the robustness of lane detection to environmental illusions
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
-
LLM-Generated Fault Scenarios for Evaluating Perception-Driven Lane Following in Autonomous Edge Systems
A decoupled offline-online framework uses LLMs and latent diffusion models to generate fault scenarios for testing edge-based lane-following models, revealing large robustness drops under conditions like fog.