Title

AWS re:Invent 2023 - Alteia, World Bank, and AWS assess road infrastructure at scale (AES202)

Summary

• The session discusses a collaboration between Alteia, World Bank, and AWS to improve road infrastructure using geospatial data and machine learning.
• World Bank lends over $70 billion annually, with infrastructure being a significant part, and requires high-quality data for effective resource allocation.
• Alteia, with a background in commercial drones, developed a software platform called ether to manage visual data and AI at scale.
• The platform ingests data from various sources, offers a user interface for data interaction, and uses AI to extract insights.
• The application built for the World Bank creates a digital twin of road networks, assesses road conditions, and is accessible from multiple devices.
• The project used open-source data, Sentinel optical data, and SAR radar data from AWS Open Data Registry.
• AWS services like S3, EKS, CloudWatch, and GuardDuty were used to manage the data and run the machine learning models.
• The machine learning process involved boosted tree models and EC2 Spot Instances to keep costs low.
• The project's success was validated using ground truth data and high-resolution imagery.
• Future advancements include time series analytics, adding precipitation and topographic data for climate event predictions, and embedding machine learning with SageMaker Geospatial.
• The World Bank aims to use the data for road safety analysis, remote project monitoring, and assessing the impact on development outcomes.

Insights

• The collaboration showcases the power of combining geospatial data with machine learning to address large-scale infrastructure challenges.
• The use of open-source data and AWS's Open Data Registry highlights the importance of accessible data for global development projects.
• The project's approach to data processing and machine learning model training can serve as a blueprint for similar large-scale infrastructure assessments.
• The World Bank's focus on climate resilience and decarbonization in over 90% of its projects aligns with its vision of improving quality of life on a livable planet.
• The integration of time series analytics and climate data into the project indicates a forward-thinking approach to infrastructure planning and climate change adaptation.
• The session demonstrates the potential for technology to significantly enhance the efficiency and effectiveness of development work conducted by organizations like the World Bank.