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

Radar shows great potential for autonomous driving by accomplishinglong-range sensing under diverse weather conditions. But radar is also aparticularly challenging sensing modality due to the radar noises. Recent workshave made enormous progress in classifying free and occupied spaces in radarimages by leveraging lidar label supervision. However, there are still severalunsolved issues. Firstly, the sensing distance of the results is limited by thesensing range of lidar. Secondly, the performance of the results is degeneratedby lidar due to the physical sensing discrepancies between the two sensors. Forexample, some objects visible to lidar are invisible to radar, and some objectsoccluded in lidar scans are visible in radar images because of the radar'spenetrating capability. These sensing differences cause false positive andpenetrating capability degeneration, respectively.  In this paper, we propose training data preprocessing and polar slidingwindow inference to solve the issues. The data preprocessing aims to reduce theeffect caused by radar-invisible measurements in lidar scans. The polar slidingwindow inference aims to solve the limited sensing range issue by applying anear-range trained network to the long-range region. Instead of using commonCartesian representation, we propose to use polar representation to reduce theshape dissimilarity between long-range and near-range data. We find thatextending a near-range trained network to long-range region inference in thepolar space has 4.2 times better IoU than in Cartesian space. Besides, thepolar sliding window inference can preserve the radar penetrating capability bychanging the viewpoint of the inference region, which makes some occludedmeasurements seem non-occluded for a pretrained network.

Radar Occupancy Prediction with Lidar Supervision while Preserving Long-Range Sensing and Penetrating Capabilities