3D point cloud modelling in the cloud
Point clouds hold highly detailed, highly precise information about the environments they map. But that precision comes at a cost: large projects often run to thousands of scans and the data stored may be terabytes in total.
After scanning and registration, the process of importing and manipulating the 3D point cloud into a model in your CAD software requires considerable computing power. As a result, the power of your hardware is an extremely significant component in the time taken to complete your project. Provisioning appropriate hardware can cut days from the length of your project; having the wrong hardware could damage your project before you’ve even started.
By moving 3D modelling to the cloud, you don’t have to worry about hardware provision, and you don’t have to try and factor in whether you have over specified your hardware enough to give it a reasonable life. It gives you the scope to scale your processing and storage to meet your project needs, and most importantly, it enables you to switch it off when you’ve finished.
The ability to scale processing capability not only increases registration speeds, it adds flexibility. Advanced processing software deploying multi-stage, vector-based registration algorithms delivers significantly robust enough results to automate cross-check. Removing manual stages enables the parallelisation of tasks.
You can do this using traditional hardware. These same software tools take advantage of multi-thread processors and hyperthreading to execute multiple coarse registrations simultaneously. With advanced i9-7980EX processors, that could deliver up to 36 scans at one time.
In the cloud, your ability to scale virtual cores is practically infinite. Theoretically, that delivers the possibility to scale up access to match the demands of any job — no matter how large — and simultaneously complete coarse registration for a project of any size. The acceleration to your workflow this creates is tremendous if appropriately captured.
When talking about scalability, it’s also important to note that cloud computing is not standing still. Cloud computing is no longer, if it ever was, a pool of lowest cost commoditised virtual machines tailored to low demand processing applications. The big three cloud providers – Amazon, Microsoft and Google — are investing billions of dollars in custom hardware to enhance the performance of their cloud platforms or tune their services for specific uses.
For example, Nvidia and other GPU makers are fulfilling the demand from cloud and systems makers for new hardware architectures that de-emphasize microprocessor performance in favour of outboard engines that are well-suited to the parallel processing needs of workloads such as 3D point cloud modelling.
Machines built to process such workloads farm out most of the work to GPUs, which process data in parallel and feed results back to the central processor, which doesn’t even need to be that fast. This approach is yielding orders-of-magnitude leaps in performance for some workloads.
Every stage of registering and modelling 3D laser scan data sets is CPU intensive — and the deliverables sizable to store. The bandwidth requirements have been the limiting stumbling block that has slowed the adoption of the cloud within this industry when compared to others. However, it also means the benefit of the cloud is even greater once those problems have been overcome.
Interest in networking technology such as 5G is surging as cloud companies wrestle with latency and data transfer issues for customers moving large amounts of data into and out of their clouds. The critical element in the future will be more about data movement than computational capability.
Latency and data volume limitations will restrict processing all this new data in the cloud. That means that on-site scanners will become the new “edge devices” acting as collection and filtration points.
With the improvement of mobile connectivity, advanced 4G connections, and 5G just around the corner, running coarse registration in the field is perfectly within reason. There are a number of registration specialists and remote applications delivered by the big-name processing software solution that allow you to get started on the time-consuming process of registration from the moment a scan is completed.
Saving time with a better process
While the processing power and flexibility of modelling in the cloud are impressive, what shouldn’t be underestimated is the saving in time delivered by better processes and collaboration during project creation, implementation, hand-over and management. Here are a few areas where time can be saved:
You won’t have to set up any registration or CAD software, something that takes both time and effort. There’s also no need for maintenance, as new features, bug fixes and updates are being automatically added to the cloud.
When working within a traditional CAD software, you have to potentially wait your turn to access the file you and your team are working on and there’s always the fear of overwriting each other’s work. With cloud-based modelling, everyone works on the same data with no confusing copies of files. The data is constantly and seamlessly updated, and any changes made by members of your team appear instantly.
Cloud basics revolve around remote access — there won’t be many people who are not familiar with Office 365, Google Docs or some form of web email. Putting applications and data in the cloud delivers access where and when it is needed. This allows you to provide flexible work schedules, engage with a remote workforce and share deliverables and data iteratively with clients. For surveyors, it also delivers the possibility to start registration in the field.
Successful project outcomes depend on consistent communication with the team throughout every project phase. Connect teams and information in a central project location in the cloud to improve productivity, reduce rework, and accelerate project delivery.
The total cost of ownership of a cloud-based solution is typically lower than traditional desktop software. There are no installation or maintenance costs involved with using a cloud software.
Most cloud-based software will also operate on more innovative pricing structures. For example, software such as Vercator Cloud operates a pay as you go usage, giving users the flexibility to use it as required each month, without costly subscriptions. The software makes use of a token system where users only pay for what they use.
Others, such as Autodesk BIM 360 Design, offer a flexible subscription model where subscribers can pay monthly, annually or for 3 years.
One of the biggest under-rated strengths of the cloud is how it can change how “we’ve always done things” for all stakeholders in a construction project. Either at offsite meetings with clients or at home, project files can be accessed and updated and shared anywhere away from the office. This is one of the fundamentals of Building Information Modelling (BIM).
Put simply, BIM requires the management of everything from a project start date, to the names of external contractors, to drawing information such as floor level, to point clouds and models and beyond. Standardisation of how this information is organised and managed is a keystone of ISO 19650 — which is the international standard underpinning BIM.
Since BIM requires owners, architects, and contractors to get more involved in the entire lifecycle of the project, it is important for the parties to be able to quickly share information. A cloud-based repository encourages more frequent information exchanges among the stakeholders, which adds more depth and value to the data.
Look to outcomes, not technology
In every industry, technology is only a tool — not a solution. To deliver meaningful outcomes, you need to make updates to processes that access the benefits delivered by the technology.
Implementing cloud lowers barriers to capture data onsite, speed of alignment and convenience of downstream analysis. By providing high-speed, robust, automatic alignment of hundreds of 3D point cloud laser scans, new working methods will emerge.
In future, all buildings and structures could be scanned many times during construction and renovation, resulting in the ability to correct construction errors on a day-to-day basis. Further, with emerging techniques for recognising and 'extracting' complex objects, the benefits of a cloud approach will also be enjoyed by more downstream users in asset or facilities management.
Tags: point clouds