A model for automation of infrastructure maintenance using representational forms

This paper proposes a model for the application of imaging to the automation of infrastructure maintenance. The model incorporates: (a) a domain structure, (b) human-machine interaction, (c) processes including machine vision, sensor data fusion, automated task planning, and control loops, and (d) graphical representation and visualization. In this paper, results are presented for automated crack and joint sealing, focusing on the development and use of a line snapping machine vision algorithm. It is concluded that the algorithm should be broadly applicable to applications in pavements, buildings, bridges, storage tanks, tunnels and utility pipes. The model can provide context and structure for development of new algorithms.