At the Spar 3D Expo and Conference in Houston I had the opportunity to chat with Nicolas Arnold, VP of Product Development at SKUR. SKUR is a young software company that has just released its first product, which automates comparing a 3D digital model to reality as represented by a point cloud from terrestrial scanning technologies (typically LiDAR). The scan can be of an existing structure or a new structure in construction. The variances are prioritized and can be inspected in an industry standard 3D visualization application.
Nicolas, SKUR provides technology that finds the differences between digital models and point clouds from scans. What are some of the applications of this technology ?
We are in the business of comparing 3D building models with point clouds that are registered to the building model. We do this for two very specific use cases.
The first use case is what I refer to as construction QC. Let's say you have a design model and during construction you capture a point cloud by laser scanning or some other technology. What you are interested in is comparing the design model to what was actually built out in the field. To do this you simply upload the design file and the point cloud, specify three tolerances that are used to prioritize the differences between the model and point cloud, and then run our cloud-based application. The application outputs colour-coded deviations, which you can visualize in Navisworks.
The second use case is what I refer to as model QC where you have the reverse situation. Let's say there is an existing structure, a building or almost anything. But there is no digital model and you need a model for a renovation or retrofit. You can use a scan-to-BIM process to scan the building and generate a point cloud from which a model can be created. The model creation could be done by your own people, by a third party provider, or by an automated process such as Clearedge. In this use case what you are interested in is how well the model fits with reality. You follow the same process, uploading the model and the point cloud, to compare the model to the point cloud to see where there might be modeling errors and other types of deviations.
These are the two use cases that SKUR serves. We target the sophisticated construction contractor (GC). One application of the technology for a GC is to perform construction QC on their projects to look for problems. Another application is pre-construction, when the GC is considering bidding on a renovation or retrofit project and wants to provide a digital model to the architect. The GC wants to be confident that the model represents the structure accurately so he can tell the architect that the model is a reliable representation of reality. These are the folks that we are targeting for these two use cases.
Does SKUR provide data capture services ?
No, we are a software company. If you are interested in this technology, you can create an account, it's up to you to capture the point cloud data however you want to do it, upload it onto our web-based system together with a digital model and then run our application. You then download the variances and visualize them using an industry standard tool. We don't do any consulting.
I'm interested in how SKUR would be used during construction. What if the construction is only 10 % or 20 % complete, do I have to specify which part of the incomplete structure I have scanned ?
We have very smart algorithms that figure out which part of the structure to look at. For example, if you bring us a point cloud of just the corner of a building and the model of the entire structure, we only focus on that corner of the building. Another example is a building that is 12 stories high, but I only have a scan of the basement. Our software is smart enough to figure out the part of the model that relates to the scan.
What about the opposite case where during a scan-to-BIM process for an existing building, I have not finished the model so I only have a model for part of the building ?
It's basically the same algorithm. The software is smart enough to figure out which part of the scan relates to the partial model. When you run our software, you simply upload the model and the point cloud and specify the tolerances. The process is the same for construction QC or for model QC. From the perspective of the software, it's really the same thing.
Where does your software run, on a desktop or in the cloud ?
We are completely web-based, that is to say, cloud-based. Our software require heavy-duty processing that would not have been possible 5 years ago. It is all in the cloud and it's high performance multi-core computing. We bring in the largest computers that are available on the Amazon cloud for these problems and they can generate the variances in a few hours.
What types of digital models do you support, 3D CAD, Revit, Bentley BIM, or IFC models ?
We are a young company, and we've just started. Our first release supports 3D DWG models. But I think Revit models are the next logical step. And then we also plan to support the Bentley world. The 3D DWG model really can be anything, with any level of complexity. I have had people asking me about things other than construction, like planes, such as an Airbus. I talked to a gentleman yesterday, who wants to use our software for a golf course. As long as you have a DWG model and a point cloud that's registered to the DWG, we can find the variances between the two. The registration between the two is critical. It has to be done accurately.
How difficult is it to choose the tolerances ?
It's very easy. You choose three values corresponding to different degrees of variance which are colour coded, green, yellow, and red in the output.
How do you visualize the variances, I imagine for a large project there could be thousands of these ?
SKUR output is a visual representation and report of the detected locations where there are discrepancies showing the precise variance from design and, very importantly, the direction of the variance. We colour code the discrepancies so you can spot the problem areas very rapidly. For example, in the graphic red and yellow coloring indicates degree of variance. Arrows show the direction of issue. In our experience most of the variances are green, so you can filter these out and just look at the ones colour-coded red, which are the big discrepancies. This allows you to focus just on those important areas. We're like an insurance company. Most of the time it's fine, but when it's not you need to know.
Where did the algorithms come from and where was the implementation done ?
We've invented everything from scratch in Oakland, California
I can imagine this technology being of tremendous value in many different sectors of the construction industry as well as in non-construction sectors. Which sectors and geographies are you targeting as the market for your technology ?
The sector we have targeted so far are construction contractors (GCs). But going toward we will be increasingly focussing on owners. We think that our software provides a view into the construction process that enables them to see during construction how well the designs by the architects and engineers compare to what is actually being built.
We are focussing on North America, but we also have clients in Europe. Europe is definitely high on our list, but as a young company we are focusing on close to home first.
If people want to try your software, how do they engage with you?
It's very simple. They sign up and create an account. Then they can upload a model and a point clouds and hit the magic button. We do handholding in the beginning to help new customers. But we don't do any data capture and or any data preparation. We simply provide a web-based software service.