AI Use Case: Proof of Concept for Crack Surveys Using AI at Tokyu Construction Co., Ltd.

The damage detection AI service "e-Drone AI" (Service Page: https://www.nttedt.co.jp/edrone-ai) is an inspection support service that analyzes surface images of concrete structures using AI to accurately detect cracks and deterioration damage. (Ministry of Land, Infrastructure, Transport and Tourism Inspection Support Technology Performance Catalog Technology Number: BR010076-V0126)

This time, we will introduce a use case for railway viaduct inspections at Tokyu Construction Co., Ltd.

1. Background

Tokyu Construction Co., Ltd. (Headquarters: Shibuya-ku, Tokyo), which utilized "e-Drone AI" this time, faced the following challenges in inspection surveys of infrastructure structures, primarily railway viaducts:

- There are many locations where visual confirmation is difficult, such as high places. - It takes a lot of time and manpower to extract and organize damage from a massive amount of captured images. - It easily depends on the experience of the inspector, making variations in judgment likely to occur. - Efficiency and labor-saving across the entire inspection survey operation were required.

To solve these challenges and improve operational efficiency while maintaining and improving inspection quality, they utilized "e-Drone AI".

Utilization location: A railway viaduct currently undergoing repair work.

2. Utilization Details

◇ Inspection survey targets

Deck slabs, main girders, and cross girders of railway viaducts (concrete structures)

◇ Photography method

High-resolution photography using single-lens high-resolution cameras. Comprehensively photographing the target area to acquire data for analysis.

◇ Analysis details

AI analysis of the captured images using "e-Drone AI". For some components, orthophotos are generated from the captured images and analyzed by AI. (It is possible to grasp the damaged areas planimetrically) CAD output of crack damage maps.

◇ Main detection items by AI

- Cracks (width/length) - Peeling - Rebar exposure - Water leakage - Free lime

3. Introduction Effects

By utilizing "e-Drone AI", the following effects were obtained:

- Halved the work required for crack surveys (Labor-saving) - Improved inspection quality by reducing the risk of oversights (can extract crack widths from 0.05mm) - Utilized for estimating repair work (utilized for calculating information quantities such as crack width and length) - Streamlined report document creation (Since a CAD file drawn with crack lines is generated, a damage map can be created by overlapping it on a drawing)

This contributes to creating an environment where they can focus on more value-added operations by streamlining the post-inspection processes, which previously required a lot of time.

4. Inquiries regarding this matter from customers and the press

NTT e-Drone Technology Co., Ltd. Service Promotion Department omakase_edrone@nttedt.co.jp