DEVELOPMENT OF A COMPUTER VISION TRACKING SYSTEM FOR AUTOMATED 3D RECONSTRUCTION OF CONCRETE BRIDGES

Concrete bridge inspection is nowadays primarily a slow, subjective, non-comprehensive and costly set of procedures. Automation of the acquisition method is especially desirable for economical and repeatability reasons. Digital data is normally derived from well established non-destructive testing techniques, high resolution cameras and, more recently, by 3D laser scanning. This latter technique has some advantageous aspects in terms of reliability, repeatability, completeness and intuitiveness of the analysis of the resulting 3D reconstruction of the concrete structure. Statical laser scanning is, though, impractical for a variety of different reasons. A possible way of overcoming such difficulties is represented by dynamical measurement, achieved by moving in a prescribed manner the laser scanner during the scanning process. This procedure, on the other hand, requires a reliable tracking system for the laser scanner position and orientation. This work focuses on the development of such system, based primarily on computer vision measurement systems. A compact and lightweight 3D laser scanner has been placed on an automated carrier able to move along a standard inspection by-bridge, and a system of cameras and transducers has been designed to measure the carrier position and orientation based on the assumption of rigid body motion of the by-bridge multi-link arm during inspection operations. Several experimental tests have been performed to assess the viability of the proposed system and to evaluate its performance.