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

We focus on the problem of LiDAR point cloud based loop detection (orFinding) and closure (LDC) in a multi-agent setting. State-of-the-art (SOTA)techniques directly generate learned embeddings of a given point cloud, requirelarge data transfers, and are not robust to wide variations in 6Degrees-of-Freedom (DOF) viewpoint. Moreover, absence of strong priors in anunstructured point cloud leads to highly inaccurate LDC. In this originalapproach, we propose independent roll and pitch canonicalization of the pointclouds using a common dominant ground plane. Discretization of thecanonicalized point cloud along the axis perpendicular to the ground planeleads to an image similar to Digital Elevation Maps (DEMs), which exposesstrong spatial priors in the scene. Our experiments show that LDC based onlearnt embeddings of such DEMs is not only data efficient but alsosignificantly more robust, and generalizable than the current SOTA. We reportsignificant performance gain in terms of Average Precision for loop detectionand absolute translation/rotation error for relative pose estimation (or loopclosure) on Kitti, GPR and Oxford Robot Car over multiple SOTA LDC methods. Ourencoder technique allows to compress the original point cloud by over 830times. To further test the robustness of our technique we create and opensourcea custom dataset called Lidar-UrbanFly Dataset (LUF) which consists of pointclouds obtained from a LiDAR mounted on a quadrotor.

FinderNet: A Data Augmentation Free Canonicalization aided Loop Detection and Closure technique for Point clouds in 6-DOF separation