Two algorithms for vehicular obstacle detection in sparse pointcloud

TL;DR

This paper introduces two novel algorithms for detecting obstacles in sparse lidar pointclouds, enabling accurate 3D bounding box retrieval using low-plane lidar sensors, which are cheaper and less computationally demanding.

Contribution

The paper presents two new methods based on occupancy grid and geometric refinement for obstacle detection with low-plane lidar sensors, improving cost-effectiveness and efficiency.

Findings

Validated on a custom dataset with ground truth.

Achieved accurate 3D bounding boxes with 8 and 16 plane lidars.

Demonstrated feasibility of low-cost lidar obstacle detection.

Abstract

One of the main components of an autonomous vehicle is the obstacle detection pipeline. Most prototypes, both from research and industry, rely on lidars for this task. Pointcloud information from lidar is usually combined with data from cameras and radars, but the backbone of the architecture is mainly based on 3D bounding boxes computed from lidar data. To retrieve an accurate representation, sensors with many planes, e.g., greater than 32 planes, are usually employed. The returned pointcloud is indeed dense and well defined, but high-resolution sensors are still expensive and often require powerful GPUs to be processed. Lidars with fewer planes are cheaper, but the returned data are not dense enough to be processed with state of the art deep learning approaches to retrieve 3D bounding boxes. In this paper, we propose two solutions based on occupancy grid and geometric refinement to retrieve a list of 3D bounding boxes employing lidar with a low number of planes (i.e., 16 and 8 planes). Our solutions have been validated on a custom acquired dataset with accurate ground truth to prove its feasibility and accuracy.