With modern techniques based on artificial intelligence and machine learning, type and condition can be mapped quickly, efficiently and with high quality.

The solution is based on mass collection of images and tailor-made machine learning models for image recognition. The models automatically detect the type and fault of railings, and provide accurate positions for their location. iSi AS is the first to use machine learning to streamline this process.

iSi inSight has been developed as an innovation project in collaboration with Arvid Gjerde AS, the Norwegian Public Roads Administration, NTNU Ålesund, Møre og and Romsdal county municipality, and Innovation Norway.

Data collection

The collection of image data takes place using a customized rig with three high-tech cameras and GPS. The rig is mounted behind the vehicle, and data capture can be carried out at normal driving speeds, up to 80 km/h. This makes the collection quick and efficient, while not obstructing normal traffic on the road. The collected data is then stored in a secure database, and then our trained machine learning models can automatically recognize railing types and highlight errors and defects.

Reporting and analysis

The inspection results are presented in a user-friendly way in our tailor-made platform, so that our customers can easily get an overview of which guardrail faults and types exist, and where they are located.

This provides a good management basis for prioritising and initiating measures. The images are stored in our database, so that they can be used for further analysis as needed. This solution also offers access to data via API.