According to German law, federal trunk roads must be checked every four years for their condition. For this purpose, a measuring vehicle drives along all roads and photographs the surface.
These photos are then manually evaluated by people - a very time-consuming and costly procedure. We want to change that.
Together with a team of international experts, we are working on an artificial intelligence that scans images and automatically detects cracks and damage. We feed the software with photos and teach it what damage looks like. What's special: The software also predicts how reliable its prognosis is. This
- saves an incredible amount of time
- is cheaper and above all
- is more accurate than a manual evaluation procedure.
The project is financed by Germany, Austria and Switzerland.
The AI software is shown images of roads with and without defects. It learns to distinguish between damage and damage-free surfaces. This allows the AI to describe, for example, what is a normal road condition and what is a poor road condition.
The AI software provides information on how accurate the result is: road conditions can be assessed and evaluated in fractions of a second - faster than the human eye could ever do.
VIA IMC trains artificial intelligence
Project managers can see through the AI program what kind of damage there is. They can evaluate the damage and draw the necessary conclusions. Roads can be maintained for longer.
Die Projektpartner:
-
Illmenau University of Technology
-
Austrian Institute of Technology (AIT)
-
Engineering firm LEHMANN+PARTNER
-
Federal Highway Research Institute (BASt)
The project is financed with federal funds from Germany, Austria and Switzerland.
VIA IMC supports the project partners above all by training artificial intelligence. VIA IMC creates the data set that the AI needs to learn.