Reproduce “Weather Influence and Classification with Automotive Lidar Sensors” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Robin Heinzler, Philipp Schindler, Jürgen Seekircher, Werner Ritter, Wilhelm Stork
This research quantifies the impact of heavy rain and dense fog on automotive lidar sensor performance using a comprehensive dataset from controlled climate chamber and real-world road conditions. It also presents a lidar-only weather classification method that achieved up to 97.14% mean IoU in controlled environments for the VLP16 sensor, demonstrating the ability of autonomous systems to intrinsically detect adverse weather.
Key insights
- Heavy rain and dense fog significantly reduce lidar detection range and point density, especially for non-retro-reflective objects like pedestrians, which become largely undetectable in dense fog below 40m visibility.
- Multiple-echo lidar sensors are beneficial in adverse weather as the contribution of second and third echoes to object detection increases through atmospheric scattering.
- Lidar attributes like Echo Pulse Width (EPW) for Scala (increasing in fog, smaller in rain) and the variance of Intensity for VLP16 (highest in rain) are strong indicators for weather classification, with dynamic scenes introducing higher feature variance.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.