ML Updates for OpenStreetMap: Analysis of Research Gaps and Future Directions

TL;DR

This paper analyzes current ML techniques for updating OpenStreetMap, identifies research gaps, and proposes DeepMapper as a novel solution to automate and improve the map updating process.

Contribution

It provides a comprehensive analysis of existing ML approaches for map updating and introduces DeepMapper as a new method to automate OpenStreetMap updates.

Findings

Current ML methods have limitations in automating map updates.

Research gaps include data quality and automation challenges.

DeepMapper shows potential for improving update accuracy.

Abstract

Maintaining accurate, up-to-date maps is important in any dynamic urban landscape, supporting various aspects of modern society, such as urban planning, navigation, and emergency response. However, traditional (i.e. largely manual) map production and crowdsourced mapping methods still struggle to keep pace with rapid changes in the built environment. Such manual mapping workflows are time-consuming and prone to human errors, leading to early obsolescence and/or the need for extensive auditing. The current map updating process in OpenStreetMap provides an example of this limitation, relying on numerous manual steps in its online map updating workflow. To address this, there is a need to explore automating the entire end-to-end map up-dating process. Tech giants such as Google and Microsoft have already started investigating Machine Learning (ML) techniques to tackle this contemporary mapping problem. This paper offers an analysis of these ML approaches, focusing on their application to updating Open-StreetMap in particular. By analysing the current state-of-the-art in this field, this study identi-fies some key research gaps and introduces DeepMapper as a practical solution for advancing the automatic online map updating process in the future.