Machine Vision for Asset Management Status: CompleteReport Date: 06/25/2025 Summary: This project investigated the use of evolving machine vision technologies to improve asset management procedures. Machine vision is technology that enables machines to “see” and interpret images, similar to how humans see and understand the world. It includes capturing an image, analyzing the image, and then completing and action or decision based on the analysis. The team successfully tested three different lane detection systems (Vaisala RoadAI, Blyncsy Payver, and Enlite AI Detekt) using a random sample of MnDOT VideoLog data. The algorithms tested were designed for lane detection in support of driving automation systems. However, since the visibility of the marking is central to their function, it also serves as a key indicator of their condition. This demonstrated that a system could be developed to automatically score pavement marking condition automatically using VideoLog data, effectively flagging locations needing additional inspection. System-wide screening is not currently feasible with the existing pavement marking inspection systems and staffing levels. Results from this study indicate the three software solutions are viable options for automating and improving MnDOT’s road asset management practices. Reports: Machine Vision for Asset Management (Fact Sheet #2025-42)
