Road inspections are very labor-intensive and time-consuming tasks.
In our new ebook "3 practical business cases in asset and maintenance management" we explore a case study with Unihorn (part of Strukton) that illustrates how artificial intelligence can ensure that a typical Dutch highway is inspected with consistent results in roughly a fifth of the time.
Road inspections produce subjective results because they’re based on the individual judgement of multiple inspectors. The high degree of manual work drives costs up, but in the past it was considered too difficult to automate these inspections due to the variety and complexity of road defects.
Visual road inspection
Unihorn chose to collaborate with BrainCreators to build an automated solution for visual road inspection called Inspech. Inspech provides immediate insight into road segments that are due for maintenance and for what reasons.
The BrainMatter powered solution detects damage on road surfaces using a designated camera car. It uses intelligent automation to inspect every part of a road and locate a wide variety of defects, regardless of orientation and size.
Detections are exported as GIS files and displayed on an interactive map. For the road as a whole, an extensive report is generated to assess quality and degradation at a glance.
Inspech is trained by turning the knowledge of experienced road inspectors into a formal set of examples. BrainMatter ingested a large collection of manually executed inspections, which were quickly cleaned and validated by the inspectors. Next, the system was fine-tuned to detect variants of defect classes and output results in a map viewer.
As a result, a typical Dutch highway can now be inspected with consistent results in roughly a fifth of the time, enabling Unihorn to speed up its services by a factor of five.
Are you interested to have more in-depth information about these cases? Download our free ebook!