A biomimicry approach to automating visual road surveys

Abstract

Road surveys that quickly and efficiently identify features and assess their condition are keystones of an effective road asset management system. Manual visual surveys are subjective and expensive, but it appears that no software package can be flexible enough to cater to all survey needs. However, nature suggests that a generic system design is possible, a parallel being the way in which animals based on the quadruped design fill a wide range of ecological niches. This paper presents a generic design whose common design components are image acquisition, image processing, feature recognition by artificial neural networks, and condition assessment by expert systems. The system can accept either real time camera feed, or video/DVD recordings made by survey vehicles. Biomimicry principles are outlined to guide the designs application to produce a survey system for a given road feature, such as line-markings and road edges. A road guide post survey is presented as a case study.