Position and orientation correction for pipe profiling robots, by Andrew John Dettmer

I always say, if you discover that someone you know has written a thesis or a dissertation, you should read it. Well, I read it. The math was beyond me. I have a Bachelor's degree in math (technically, Computer Science and Mathematics) but either I never learned this math, or I forgot it all because I haven't used it since graduation.

800,000 miles of pipes
I learned that there are over 800,000 miles of sewer pipes in America and that most of them are nearing or are past their projected lifespan. I learned that in 1999 the Governmental Accounting Standards Board (GASB) published Statement No. 34 which requires municipalities to account for infrastructure assets in their financial reporting. As a result, the "false economy of deferred maintenance will be revealed as a long term mistake disguised as a short term benefit."

How this relates to pipe profiling robots comes in how Statement No. 34 specifies that sewer pipes be accounted for: "GASB 34 requires a condition assessment of sewer assets at least every three years using a consistent basis of measurement and scale that can be repeated." Sewer pipes, being buried, are not easy to inspect. Hence the need for sewer pipe profiling robots—robots that travel through the sewer pipes and inspect them from the inside.

Lasers, math, band saws and welders
And apparently, it's not as easy as sending a remote-controlled car with a camera into a pipe. Serious pipe profiling robots have fancy laser sensors to measure defects and pipe ovality. The problem is (and I never would have thought of this) if the robot is not pointed dead-center along the pipe centerline, the measurements can be skewed. Further, it's useful to know the orientation of the pipe itself with respect to the earth, to help in developing proper three-dimensional maps.

So, enter math: Dettmer shows a lot of fancy math about how to properly detect and account for the position and orientation of the pipe profiling robot within the pipe. The math was beyond me. Then the practical: with help from Swanbom he constructs a proof-of-concept pipe profiling robot and tests it in a 72-inch test pipe to demonstrate the validity of his new position and orientation technique. He writes a lot about how he had to construct the robot, which "included the use of an end mill, drill press, thread tap, welder, band saw, sheet metal bender, sander, and grinder." All that is as far beyond me as the math is.

Kill all humans
Finally, he ends with some suggestions for further research. My favorite recommendation is #5, "Sewer inspection vehicles are difficult to market if they are expensive. Developing a low-cost alternative would be well received by the industry." The recommendation most likely to lead to robots killing all humans is #7, "Develop an automated inspection robot which can be left alone in the sewers for days, weeks, or even months. Conceptually, a system could roam the sewers, periodically uploading condition information to provide constant asset monitoring."