CNN for Very Fast Ground Segmentation in Velodyne LiDAR Data

TL;DR

This paper introduces a CNN-based method for rapid ground segmentation in Velodyne LiDAR data by encoding sparse 3D point clouds into a multi-channel 2D signal, achieving faster processing with improved accuracy.

Contribution

The paper presents a novel encoding scheme for LiDAR data and shallow CNN architectures for efficient ground segmentation, outperforming previous methods in speed and accuracy.

Findings

Significant speed improvements over previous methods

Minor accuracy enhancements in segmentation quality

Effective encoding of sparse 3D data for CNN input

Abstract

This paper presents a novel method for ground segmentation in Velodyne point clouds. We propose an encoding of sparse 3D data from the Velodyne sensor suitable for training a convolutional neural network (CNN). This general purpose approach is used for segmentation of the sparse point cloud into ground and non-ground points. The LiDAR data are represented as a multi-channel 2D signal where the horizontal axis corresponds to the rotation angle and the vertical axis the indexes channels (i.e. laser beams). Multiple topologies of relatively shallow CNNs (i.e. 3-5 convolutional layers) are trained and evaluated using a manually annotated dataset we prepared. The results show significant improvement of performance over the state-of-the-art method by Zhang et al. in terms of speed and also minor improvements in terms of accuracy.