English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Automated vehicles require an accurate perception of their surroundings forsafe and efficient driving. Lidar-based object detection is a widely usedmethod for environment perception, but its performance is significantlyaffected by adverse weather conditions such as rain and fog. In this work, weinvestigate various strategies for enhancing the robustness of lidar-basedobject detection by processing sequential data samples generated by lidarsensors. Our approaches leverage temporal information to improve a lidar objectdetection model, without the need for additional filtering or pre-processingsteps. We compare $10$ different neural network architectures that processpoint cloud sequences including a novel augmentation strategy introducing atemporal offset between frames of a sequence during training and evaluate theeffectiveness of all strategies on lidar point clouds under adverse weatherconditions through experiments. Our research provides a comprehensive study ofeffective methods for mitigating the effects of adverse weather on thereliability of lidar-based object detection using sequential data that areevaluated using public datasets such as nuScenes, Dense, and the CanadianAdverse Driving Conditions Dataset. Our findings demonstrate that our novelmethod, involving temporal offset augmentation through randomized frameskipping in sequences, enhances object detection accuracy compared to both thebaseline model (Pillar-based Object Detection) and no augmentation.

Enhancing Lidar-based Object Detection in Adverse Weather using Offset Sequences in Time