Uncertainty and error in laser triangulation measurements for pipe profiling.

摘要

Underground pipeline infrastructure often receives insufficient attention and maintenance. Those responsible for ensuring the continuing functionality of this infrastructure primarily use subjective information in their decision making, and standards defining the level of damage acceptable before repair or replacement are difficult to implement. Laser pipe profiling is a relatively new technology that has emerged to take a step toward the objective assessment of buried assets. A laser profiler is a device that traverses a section of pipe, taking measurements of radius around the circumference of the inner pipe wall at multiple locations along the length of the pipe. The accuracy of the measurements obtained by a profiler is a critical piece of knowledge for the evaluation of its usefulness.;Analytical evaluation of the conical projection configuration revealed infinite measurement error for a region of the measurement space; the unbounded error was eliminated by utilizing two conical lasers. The accuracy and uncertainty of the perpendicular plane and side facing configurations were much better than for the conical configuration. Physical models of these two configurations were constructed, and measurements were collected for a pipe section to validate the measurement and uncertainty predictions of the analytical models. The difference between observed worst-case laser measurement error and predicted uncertainty was on the order of 0.1% of nominal pipe radius. This work provides pipe profiler designers the analytical detail required to understand the relationship between system geometry, camera parameters and measurement accuracy. The work provides asset managers with a reference against which to evaluate laser profiling for their infrastructure condition monitoring needs.;Analytical measurement and uncertainty models were developed for three laser profiling configurations. These configurations involved a digital camera and a laser whose relative position and orientation were fixed relative to one another. The three configurations included (1) a conically projected laser aligned with the pipe axis, (2) a planar laser placed perpendicular to the pipe axis, and (3) a side-facing laser that projected a line onto the pipe wall parallel to the axis of the pipe. The models utilized normalized system parameters to compute pipe geometry from digital images that reveal the intersection of the laser light and the pipe wall; error propagation techniques were applied to compute the variation in measurement uncertainty as a function of position in the measurement space.