Connect with us


Alteia and the World Bank assess and enhance road infrastructure data quality at scale using AWS | Amazon Web Services



A road in rural Peru. Rural roads are critical for connectivity, transportation, and development. (Photo credit: World Bank Transport Team 2)

Road infrastructures connect households to higher quality opportunities for employment, healthcare and education. Since 2002, the World Bank has constructed or rehabilitated more than 260,000 kilometers of roads, lending more for roads than for education, health, and social services combined. The World Bank assesses road infrastructure faster and at less cost by using Alteia data analytics powered by Amazon Web Services (AWS), geospatial imagery, and satellite imagery available on the Registry of Open Data on AWS. This helps local governments rehabilitate road infrastructures and connect underserved communities in countries like Mexico, Peru, and Tunisia.

With 189 member countries, the World Bank is the largest multilateral development bank globally with a mission to reduce poverty and increase shared prosperity in a sustainable way. It provides technical support and lending to governments in developing countries to improve their economies and the standard of living of their people. Transportation infrastructure is key for reaching these objectives, by facilitating accessibility to economic and social opportunities, including for traditionally marginalized and isolated communities.

Alteia, a leading enterprise artificial intelligence (AI) software company headquartered in Toulouse, France, worked with the World Bank Group to build an application on the Aether software platform. The application runs on AWS and extracts accurate insights from road networks at scale in developing countries based on publicly  available data. Alteia uses computer vision and geospatial imagery to efficiently extract road network data from satellite data. The Aether platform rapidly aggregates, contextualizes, and analyzes data streams from multiple data sources. This enables Alteia to support use cases like extracting road networks, validating road network quality, and enriching network data with other types of information.

Collecting data and insights about road networks in remote locations using traditional, on-the-ground road infrastructure assessments and high-resolution satellite imagery can be time consuming and costly. Thanks to the power of machine learning (ML) and geospatial data on AWS, the approach developed by Alteia allows mapping and assessing a country’s entire road network for 15 percent of the cost of visual inspection methods. As a result, only three weeks were needed to complete the mapping and assessment of 500,000 kilometers of roads, which would take 15 years using traditional methods. Using AWS and leveraging numerous sources of free data, Alteia’s approach was more than 90 percent cheaper than it would have been if relying on high-resolution satellite imagery.

The digital road network representation also helps assess risks presented by external events like flooding. By moving from a pure statistical approach to a risk-based decision-making process, governments can proactively address potential challenges, enhancing the resilience of their infrastructure. Alteia successfully facilitated access to vast amounts of satellite imagery for countries like Mexico, Peru, and Tunisia by leveraging satellite imagery data from the Registry of Open Data on AWS and data preparation and ML algorithms running on Amazon Elastic Kubernetes Service (Amazon EKS). The application generated up to 2 terabytes (TB) of refined data for each country, efficiently stored and delivered through Amazon Simple Storage Service (Amazon S3). 

Assessing road infrastructure at scale

The Alteia application leverages Sentinel imagery from the Registry of Open Data on AWS to cross-reference with vector road data including OpenStreetMap (OSM). Using AWS compute and infrastructure, Alteia creates an enhanced road network with additional features and contextual attribution such as demographic data. Alteia uses optical imagery from Sentinel 2 and Synthetic Aperture Radar (SAR) satellite imagery to determine the roughness of roads and differentiate between paved and unpaved roads. Data accuracy is validated using crowd-sourced ground truth data. 

“We always strive to get the best possible performance for our customers, and we can do that by leveraging AWS EKS to deploy our algorithms at scale,” said Yann Ameho, chief technology officer for Alteia.

Alteia developed a microservices architecture utilizing Amazon EKS, a managed Kubernetes cluster solution. This application seamlessly integrates data from third-party sources and Amazon S3 to enable storage of large geospatial datasets and faster analysis through efficient data access. Alteia provides the security and observability of their platform using Amazon GuardDuty for threat detection and Amazon CloudWatch for monitoring and logging.

architectural diagram showing the Alteia microservices architecture that securely integrates third-party data, stores large geospatial datsets, and enables faster analysis

Figure 1. The Alteia microservices architecture uses Amazon EKS, Amazon S3, Amazon Guard Duty, and Amazon CloudWatch to securely integrate third-party data, store large geospatial datasets, and enable faster analysis.

The Alteia application uses Amazon EKS running on Amazon Elastic Compute Cloud (Amazon EC2) instances to deploy multiple container pods optimized for specific tasks. Alteia created a microservice for generating tiles from satellite images and vector data. Raster tiles were cached in Amazon S3, while vector tiles were stored in a MongoDB database, to provide quick retrieval and scalability.

The application incorporated raster and vector tile servers to provide data for the analytics and ML components, used for data inference and enrichment. The web server provided real-time visualization capabilities for end-users, delivering a seamless user experience.

architectural diagram that shows how Alteia provides quick retrieval and scalability of satellite images

Figure 2. The Alteia microservice provides quick retrieval and scalability of satellite images and vector data using Amazon S3 and Mongo DB.

By effectively computing and analyzing road networks, governments can prioritize maintenance operations, optimize spending, and enhance their roads. This application helps connect underserved communities, making significant strides in improving transportation infrastructure and accessibility. Additionally, Aether provides the ability to prioritize road investments in lending operations, build resilient networks, and monitor infrastructure progress across countries. This geospatial solution empowers decision-makers to perform statistical analyses on the impact of infrastructure on socio-economic outcomes.

The combination of cloud-based data availability, powerful cloud computing, and AI has revolutionized the evaluation of large-scale road networks. The World Bank’s mission, Alteia’s expertise, and AWS cloud capabilities have led to the innovative use of geospatial analytics, empowering governments and organizations to make data-driven decisions and create more resilient and efficient road networks worldwide.

Learn more about AWS for aerospace and satellite

Organizations of all sizes across all industries are transforming and delivering on their aerospace and satellite missions every day using AWS. You can learn more about the solution highlighted in this blog post, directly from Alteia, World Bank, and AWS subject matter experts, by attending in-person or virtually watching their breakout session at AWS re:Invent 2023.

Find out more about the cloud for aerospace and satellite solutions so you can start your own AWS Cloud journey today.

Get inspired. Watch our AWS in Space story.

Read more about AWS for aerospace and satellite:

Continue Reading