Case Study

Oil & Gas Service Provider Cuts GPS Processing from Hours to Minutes with AWS SageMaker

At a glance

The customers were losing time and money routing service vehicles to remote well sites due to incomplete road data—private lease roads and access routes didn’t appear in standard mapping systems, causing navigation delays and inefficient routing.

NMD deployed an AI-powered road detection system using AWS SageMaker and U-Net deep learning models that automatically identifies roads from satellite imagery and validates them through human-in-the-loop processing, reducing GPS data processing from hours to minutes and discovering previously unmapped access routes.

The MVP was delivered in a 2-week sprint and successfully moved to the development environment, with the extensible platform ready to accommodate additional data sources like pipeline maps and SAR imagery for expanded routing intelligence.

Industry

Use Case

Solution implemented

The value equation

Company Snapshot

The customer is a service routing and navigation platform for the oil and gas industry that enables efficient field operations and emergency response by mapping complete road networks—including private lease roads and access routes—to reduce travel time, fuel costs, and operational delays for service vehicles reaching remote well sites.

Location

United States

Customer Situation

The customer, a service routing platform for oil and gas field operations, was experiencing critical routing inefficiencies across remote well sites. Service vehicles needed efficient routes from public roads to remote locations, but existing mapping systems missed private lease roads and temporary access routes that could save significant travel time and fuel costs.

The core challenge: while GPS tracks indicated where roads existed, the absence of GPS data didn’t reliably indicate no road was present, and manual validation processes couldn’t scale. GPS data processing took hours instead of minutes, creating operational bottlenecks that slowed field service dispatch and emergency response.

In the oil and gas industry, logistics optimization is critical—companies operate thousands of miles of pipeline infrastructure and dispatch heavy equipment and specialized crews to remote well sites daily. Incomplete road network data leads to navigation delays, suboptimal routing, increased fuel costs, and potential safety risks during emergency response situations.

NMD Solution

NMD reviewed and found an opportunity to leverage satellite imagery and machine learning to automatically detect and validate roads that don’t appear in standard mapping systems, creating a comprehensive view of all accessible routes—public and private—across remote oil and gas operations areas.

Enabling automated road detection from satellite imagery required implementing a U-Net deep learning model using AWS SageMaker for image segmentation, Positive Unlabeled Learning to handle incomplete GPS labeling, Amazon A2I for human validation workflows, and AWS Batch for scalable processing of high-resolution Sentinel-2 satellite imagery.

Within 2 weeks, NMD’s Computer Vision & Asset Management Solution for Oil & Gas delivered an MVP road detection system that reduced GPS data processing from hours to minutes and successfully identified previously unmapped road segments from satellite imagery.

The delivered solution includes a complete ML processing pipeline deployed to the development environment with GeoJSON output compatible with existing systems, creating an extensible platform foundation ready to accommodate additional data sources like pipeline maps and SAR imagery for expanded routing intelligence across oil and gas operations.

Ready to Transform Your Oil & Gas Business?

See how New Math Data can transform your Oil & Gas business with AWS-powered innovation.