Real-Time Diffusion Relighting as a Neural Sensor: Closing the Low-Light Domain Gap for Robotic Perception
Paper · https://arxiv.org/abs/2611.17438
arXiv:2611.17438v1 Announce Type: new Abstract: Low-light conditions severely degrade robotic perception: object detectors trained on daylight imagery are prone to failure under dark and adverse weather conditions that fall outside of their training distribution. Furthermore, from an efficiency standpoint, existing Low-Light Image Enhancement (LLIE) methods are designed for offline processing and are therefore impractical to integrate into live control loops. To address these issues, we propose to embed a Generative Neural Sensor within the control loop that treats illumination as a generative prior rather than a signal processing problem, enabling task-relevant semantic recovery. We further propose an Real-Time Perceptual Architecture that decouples high-latency semantic recovery from low-latency robotic control. Rather than blocking the control loop, our system runs a large-scale image-to-image diffusion model in a background thread, and is activated only when a drop in detection confidence signals a perceptual uncertainty event, significantly reducing computational overhead. Our method sets a new state-of-the-art performance on the ExDark low-light benchmark with 87.1% mAP, and under adverse weather conditions, significantly boosts detection rates over competing methods. On a navigation task in a closed-loop autonomous driving simulator (CARLA), we achieve a 100% lap completion rate, establishing a simulation-based proof-of-concept for integrating high-latency generative models with standard robotic control autonomy stacks.
First seen by Oriel on .
AI summary
This paper presents a generative neural sensor that uses real-time diffusion relighting to recover perception in low-light and adverse weather conditions for robotic control.
- The system treats illumination as a generative prior, targeting task-relevant semantic recovery rather than conventional signal-processing enhancement.
- A large-scale image-to-image diffusion model runs in a background thread and activates only when detection confidence drops.
- The Real-Time Perceptual Architecture separates high-latency semantic recovery from low-latency robotic control, reducing computational overhead.
- The method achieves 87.1% mAP on the ExDark low-light benchmark and improves detection under adverse weather.
- In CARLA closed-loop autonomous driving simulation, the system achieves a 100% lap completion rate as a proof of concept.
#diffusion models #robotic perception #low-light vision #autonomous driving
Generated by gpt-5.6-sol. Check the original for details.
Aliases
| Source | External ID | URL |
|---|---|---|
| arXiv AI Papers | oai:arXiv.org:2611.17438v1 | https://arxiv.org/abs/2611.17438 Read discussion |