Enabling 3D Object Detection with a Low-Resolution LiDAR

TL;DR

This paper presents a two-stage neural network framework that enhances 3D object detection using low-resolution LiDAR by combining depth completion and voxel-based detection, achieving results comparable to high-resolution sensors.

Contribution

The paper introduces a novel two-stage neural network approach that enables effective 3D object detection with low-resolution LiDAR, reducing costs while maintaining high accuracy.

Findings

Significantly improves detection accuracy over direct low-resolution LiDAR methods.

Achieves detection performance close to high-resolution LiDARs on KITTI dataset.

Effective in both easy and moderate detection scenarios.

Abstract

Light Detection And Ranging (LiDAR) has been widely used in autonomous vehicles for perception and localization. However, the cost of a high-resolution LiDAR is still prohibitively expensive, while its low-resolution counterpart is much more affordable. Therefore, using low-resolution LiDAR for autonomous driving is an economically viable solution, but the point cloud sparsity makes it extremely challenging. In this paper, we propose a two-stage neural network framework that enables 3D object detection using a low-resolution LiDAR. Taking input from a low-resolution LiDAR point cloud and a monocular camera image, a depth completion network is employed to produce dense point cloud that is subsequently processed by a voxel-based network for 3D object detection. Evaluated with KITTI dataset for 3D object detection in Bird-Eye View (BEV), the experimental result shows that the proposed approach performs significantly better than directly applying the 16-line LiDAR point cloud for object detection. For both easy and moderate cases, our 3D vehicle detection results are close to those using 64-line high-resolution LiDARs.