If you haven’t noticed there are a lot more people jumping into the laser scanning business lately. Perhaps this is due to scanner prices dropping, or maybe it’s because the software is making it easier to process the data. Maybe the growth of the market is just creating a lot of opportunity. Many of the claims of scan data accuracy remind me of the stories of the accuracy of CAD I heard thirty years ago when I was transitioning from board drafting to CAD. Whatever the reason, it is becoming ever more apparent of the importance and need for automating the quality control process.
As an architect who witnessed the transition from board drafting to CAD, I recall the many conversations I had with, and statements made by, people who talked about how accurate CAD is. The funny thing is while CAD certainly can be drafted in a more precise fashion, it is still susceptible to many of the same issues that affect the accuracy of board drafting.
Take as-builts for example. Back in the day, I used to create building as-builts using a tape measure and a clip board, as was a common practice used by architects in the era of board drafting. With that process it was normal to have trouble getting my building shell to close properly. The introduction of CAD didn’t really change this. The only thing CAD did was give the impression that things were more accurate because, well, everything looked so precise. In reality, I still faced the same issues I always did when working with the tape measure. The real problem was the lack of adequate control of the measurements and an easy way to perform a quality control check.
Now fast forward to today where I often find myself with a strong case of Deja vu. Many people talk about how accurate scanning is. We’ll it certainly can be, but there is also a lot the can go wrong. When viewing a point cloud it isn’t usually apparent whether the process used to create it resulted in accurate measurements. Much like CAD looked precise and accurate, so do point clouds.
What is needed is a way to validate the accuracy of the measurements as well as the accuracy of the deliverable created from the point cloud, whether that be CAD or BIM. The U.S. Institute of Building Documentation (USIBD) has begun to help solve this need with the introduction of its Level of Accuracy (LOA) Specification. This spec provides five ranges of accuracy. While having a way to specify accuracy is great, what is needed now is a way to validate the data in an easy, automated fashion.
I know it is possible to get registration error reports that tell us how accurate the scan registration is, but what about using software to automate a comparison of the represented data (CAD line work or modeled data) to the point cloud. The software could tell us and show us where the represented data is off from the point cloud. Now imagine that this software also can tell us which LOA our represented data falls within.
All of these things should be possible to program into today’s sophisticated software packages. The need is clearly there. Without a way to easily prove the accuracy of the scan data, people are going to continue to be fooled by the perception that if the data looks accurate, it must be accurate. New inexperienced providers will certainly make mistakes often times not realizing it. When the mistakes are discovered clients may begin to doubt the technology. Software developers take note of the opportunity to create an automated quality control process that can provide confidence to both service providers and those procuring the services.
