When you take an idea from the lab to the field, there always are hurdles to overcome. Even with the best of planning, the time comes for a reality check, and this fall Colin Milberg is again dealing with such an issue.
Milberg heads the ACI 117 tolerances committee's effort to establish a protocol for collecting as-built data to help in determining rational construction tolerances. He and his team are using laser scanners and geospatial imaging to establish a benchmark as a significant part of that effort. The technology yields large volumes of accurate data, but because the scanner requires a clear line of sight to each area, several images must be combined to provide a complete record of an element's as-built dimensions and location.
In theory, the solution is simple. Place three (or more) targets where they will be part of each scan and use those reference points to align and combine the data. Early on Milberg and his team discovered that spheres worked better than flat targets in this process. Using strategically placed baseball-sized plastic balls allows the scanner to see the same target from any angle, which means that in matching multiple scans, the software simply refers to those points in space.
The decision to use spherical targets was based on using a Trimble laser scanner, which has a long range with good accuracy but uses a relatively slow data collection method. The team subsequently acquired a Faro scanner that uses a data collection method that is significantly faster but over shorter distances. Both were demonstrated at the March 2008 ACI meeting in Los Angeles.
It's in trying to speed up the process that the targeting issue has emerged. Milberg says the two scanners look at targets somewhat differently. With the Trimble scanner's roots in a land surveying tradition, the scanning resolution can be set separately for targets, giving it the ability to pinpoint spherical targets with greater accuracy.
Faro scanners have been optimized more for mechanical applications, such as documenting piping systems in power plants, where distance has been less of an issue. For the Faro scanner to recognize the same sphere at the same distance as the Trimble, its resolution must be set much higher than what Milberg's team requires for its data collection. Of course, collecting that excess data slows the scanning process. Two steps forward, one step back.
Georeferencing is another target-related problem facing the team. Once a point cloud of data has been combined from separate scans, it still must be located with respect to a project's reference grid. This area of spatial reference is one where the Trimble scanner offers the better approach—you scan a flat target that has been set on a tripod exactly over a known point, which locks in the exact location. Meanwhile, the Faro scanner offers the faster data collection for a given scene.
Milberg has a two-part solution in mind. First, the team simply will include more targets in an area, which should yield large incremental benefits. The other thing he's planning is to do dual scanning of the area. One scan using the Trimble scanner will include all the target spheres and flat targets at known locations. Meanwhile, the Faro scanner will collect point data from multiple setups, as needed to have line-of-site access to all the surfaces under consideration, each including several targets. By matching the Faro scans to the Trimble scan, Milberg hopes to be able to combine the locational expertise of one system with the data collection speed of the other. It's a real luxury to have both machines available, but the results of this trial run may well advance the state of the art.
Now, if you're still wondering what all the fuss is about between flat targets and spheres, consider looking up “Flatland.” Written in 1884 by Edwin Abbott, this book illustrates how we add inference to what we actually perceive with our eyes to “see” in our minds the greater reality of an object, such as a sphere. This wonderful examination of how we perceive dimensions is available online for no charge at http://books.google.com.