Abstract
Abstract. Point cloud datasets for perception tasks in the context of autonomous driving often rely on high resolution 64-layer Light Detection and Ranging (LIDAR) scanners. They are expensive to deploy on real-world autonomous driving sensor architectures which usually employ 16/32 layer LIDARs. We evaluate the effect of subsampling image based representations of dense point clouds on the accuracy of the road segmentation task. In our experiments the low resolution 16/32 layer LIDAR point clouds are simulated by subsampling the original 64 layer data, for subsequent transformation in to a feature map in the Bird-Eye-View(BEV) and Spherical-View (SV) representations of the point cloud. We introduce the usage of the local normal vector with the LIDAR’s spherical coordinates as an input channel to existing LoDNN architectures. We demonstrate that this local normal feature in conjunction with classical features not only improves performance for binary road segmentation on full resolution point clouds, but it also reduces the negative impact on the accuracy when subsampling dense point clouds as compared to the usage of classical features alone. We assess our method with several experiments on two datasets: KITTI Road-segmentation benchmark and the recently released Semantic KITTI dataset.
Highlights
Modern day Light Detection and Ranging (LIDAR) are multi-layer 3D laser scanners that enable a 3D-surface reconstruction of large-scale environments
LIDAR scan-based point cloud datasets for automated driving (AD) such as KITTI usually were generated by high-resolution LIDAR (64 layers, 1000 azimuth angle positions (Fritsch et al, 2013)), referred to as a dense point cloud scans
The objective of this study is to examine the effect of reducing spatial resolution of LIDARs by subsampling a 64-scanning layers LIDAR on the task of road segmentation
Summary
Modern day LIDARs are multi-layer 3D laser scanners that enable a 3D-surface reconstruction of large-scale environments They provide precise range information while poorer semantic information as compared to color cameras. In recent nuScenes dataset for multi-modal object detection a 32-Layer LIDARs scanner has been used for acquisition (Caesar et al, 2019). Another source of datasets are large-scale point clouds which achieve a high spatial resolution by aggregating multiple closely-spaced point clouds, aligned using the mapping vehicle’s pose information obtained using GPS-GNSS based localization and orientation obtained using inertial moment units (IMUs) (Roynard et al, 2018). We shall focus on the scan-based point cloud datasets in our study
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
