Safe Digging Org VA811 Launches New AI-Powered App to Identify Ticketing Errors

VA811: Protect Life and Property

VA811’s mission is to protect life and property by preventing underground utility damage. The organization provides a single point of contact for excavators to notify utility owners and coodinates the marking of underground lines. Damaging underground utilities carries potential risks, including fires, explosions, and other problems. Utility owners mark lines with flags or paint to prevent accidents and property damage.

A screenshot of the auditing web site

The Problem: Auditing Thousands of Tickets for Errors

Virginia 811 (VA811) faced the daunting task of manually auditing a vast number of excavation marking tickets each month. Errors in excavation can pose safety concerns, so detection and rapid response are crucial. Their standard random selection auditing practice discovered that 3% of the tickets audited had errors. VA811 had the idea to improve the discovery rate and auditing efficiency by using machine learning (ML).

With a budget funded by grants, they developed an ML algorithm that more than doubled the error detection rate. However, the challenge became one of accessing this powerful data source to validate and correct these errors using human discernment. A new interface was needed to harness the mathematical output of the ML algorithm for use by trained auditors.

A screenshot of the ML data

The Solution: a New Web App

VA811 brought the ML algorithm to Skapa to create this new interface. Skapa developed VArif-AI, a complete system for finding and reviewing ticket errors. Backend automation processes new tickets every 30 minutes using the ML algorithm, marking errors and flagging 10% for ongoing ML validation. Errored and flagged tickets are queued for review using a mobile-ready web site. Auditors now focus on handling errors instead of the discovery process.

A screenshot of the auditing web site

Promising Results and Impact

In just 3 weeks, VArif-AI’s ML algorithm scored nearly 43,000 excavation marking tickets, finding 1,100 errors for human review. Nearly half of these errors were reviewed using the new app during this time. The app’s automated and data-driven approach significantly enhanced error detection, reducing safety risks and enabling swift corrective actions. Although ongoing monitoring and evaluation are crucial, the initial impact of the app is both exciting and promising. Enhanced error detection and resolution represent a significant advancement in damage prevention for Virginia excavators.