Point Cloud Change Detection With Stereo V-SLAM:Dataset, Metrics and Baseline

TL;DR

This paper introduces a low-cost stereo camera-based framework for point cloud change detection, along with a new dataset and metrics, demonstrating effective change detection in simulated environments.

Contribution

It presents a novel stereo V-SLAM framework for change detection, creates a new dataset with metrics, and validates the approach using high-fidelity simulation.

Findings

Effective change detection in simulated environments

Low-cost sensors achieve comparable performance to expensive methods

New dataset enables standardized evaluation of point cloud change detection

Abstract

Localization and navigation are basic robotic tasks requiring an accurate and up-to-date map to finish these tasks, with crowdsourced data to detect map changes posing an appealing solution. Collecting and processing crowdsourced data requires low-cost sensors and algorithms, but existing methods rely on expensive sensors or computationally expensive algorithms. Additionally, there is no existing dataset to evaluate point cloud change detection. Thus, this paper proposes a novel framework using low-cost sensors like stereo cameras and IMU to detect changes in a point cloud map. Moreover, we create a dataset and the corresponding metrics to evaluate point cloud change detection with the help of the high-fidelity simulator Unreal Engine 4. Experiments show that our visualbased framework can effectively detect the changes in our dataset.