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.
Visionfault-350k: A large-scale fault injection dataset for robotic vision systems
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2representative citing papers
TensorRT YOLO pipelines on Jetson Nano maintain stable GPU occupancy, power draw, and thermal behavior under heavy input degradation from LLM- and LDM-synthesized faults for both object detection and lane-following tasks.
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.
-
Hardware Utilization and Inference Performance of Edge Object Detection Under Fault Injection
TensorRT YOLO pipelines on Jetson Nano maintain stable GPU occupancy, power draw, and thermal behavior under heavy input degradation from LLM- and LDM-synthesized faults for both object detection and lane-following tasks.