Are point clouds changing surveying — again!

Point clouds sit at the cutting edge of surveying. Ultimately, the industry still relies on many of the fundamental equations and principles that it always has. It has, however, gone through several revolutions that changed how surveyors undertake their jobs and the types of maps that they could produce. This article will discuss the impact of point clouds and laser scanners on these surveying outputs.

Why point clouds are so great

You could point to the invention of the theodolite during the 16th century or the public introduction of GPS during the 1980s as the most significant advancements in surveying history. For my money, however, the development of commercial LiDAR systems during the 1990s had the greatest transformative impact on surveys. Point clouds are critical to making sense of the data delivered by laser scanners. This makes them an indispensable aspect of the transformational impact of LiDAR.  

Paired with point clouds, laser scanners took the ‘point-and-click’ power of total stations (and other autonomous range-finding instruments) and exponentially increased the speed with which measurements could be taken and the amount of data that could be assessed. This didn’t just improve efficiencies, it changed what could be produced. All of a sudden, it was possible to make so many individual measurements that accurate (and incredibly precise) 3D maps could be constructed of any physical space. This was something surveyors had never before been capable of creating.  

As technology has advanced, additional types of scanners have been developed. But, the terrestrial laser scanner remains at the bedrock of 3D imaging and survey technology. Nothing can match its ability to gather detail within enclosed environments. Point clouds remain the only way to transform this data into something useful.  

Where point clouds have always let down surveyors

Point clouds have always been hamstrung by slow processing and an often manually intensive ‘registration’ procedure. The transformational capability of the technology has meant that point cloud surveys have still been used despite these deficiencies. However, their use has often been limited to the largest construction projects and specific circumstances where it delivers a massive return on investment through high commissioning costs.

Laser scanners make line-of-site measurements. To gain sufficient coverage of a scene, multiple scans are usually needed. To turn this dislocated data into a composite survey, each scan must be ‘fixed’ in a common context and delicately aligned with adjacent point clouds.   

Traditionally, this alignment has depended on artificial targets that are placed throughout the scan field in areas of overlap. The capability to align point clouds using natural features has existed for years. But, targets have remained the dominant means of registering scans because of inefficiency in processing software. You pay for time saved in the field with time spent in the office.  

For example, using traditional software to align 130 scans using targetless registration can take up to 28 hours. This would be fine, but conventional software also requires manual processes (setting scan parameters, cross-check data, etc.) throughout the procedure. Turning that 28 hours into more than three days of constant work.

A slow surveying revolution that is overtaking an industry

Technological revolutions that happen quickly are easy to identify. What is much harder is trying to determine when a bunch of incremental changes have transformed how an existing technology should and can be used. This is what has happened with point clouds.

There has never been one breakthrough to completely change the way processing is undertaken. But, a number of incremental advances have now coalesced to deliver the possibility of a different strategy. One big piece of this puzzle is looking at coarse registration as three distinct stages (rotational, vertical and horizontal alignments) rather than a single process.  

To do this, the positional data of the laser scanner is used to extrapolate each point into a vector. This allows entire scans to be condensed into single points, creating ‘vector spheres’. The unique identity of the scan, however, is retained in the density and angular characteristics of the vectors. This allows for easy rotational alignment by placing adjacent vector spheres within one another. Once rotational alignment is achieved, rapid 2D point density techniques can be used to achieve alignments on the horizontal and vertical axes.

This approach allows each stage to be approached faster and with more rigour — delivering a more robust outcome more quickly. Critically, it also allows for fewer manual cross-checks throughout the process. This has also allowed for frontloading what tasks remain. In total, this new approach to coarse registration has delivered processing speeds increased by 40%-80% (depending on the size of the dataset), and the ability to queue up scans for hands-off processing — greatly improving the efficiency of the entire procedure.

This new processing technique not only improves traditional registration methods, it removes the impediment of ditching target use in the field. Processing time for cloud-to-cloud registration has been made manageable, and any continuing inefficiencies have been completely mitigated by removing the need to oversee processing.

Summary: Targetless registration is here and expanding surveying applications to never before explored places  

Targetless registration finally makes sense, and software advances are drastically improving the efficiency of all types of point cloud processing. For surveyors who are embracing these developing technologies, there are a widening number of customers and applications available to them. The opportunities are nearly limitless as the technology advances.  

Smaller construction projects are able to take advantage of point clouds. But, the technology is finding wider use in manufacturing and is critical to the burgeoning 3D printing industry. This is actually feeding back into construction. Precision 3D modelling is enabling the prefabrication of critical building components off-site through the use of 3D printing and other manufacturing techniques. Critical to this is the ability to compare production output against planning in minuscule detail by using laser scanners and point cloud processing.

Buildings, themselves, are even being approached with this ‘build and check’ technique, using laser scanners to survey each stage of a construction project in order to ensure alignment with planning. These developments are poised to revolutionise construction and are being driven directly by the maturation of point cloud technology. Any design process, however, will benefit from the precision and iterative capabilities of 3D models and point clouds. Surveyors who embrace this technology will create a competitive advantage. Those who do so first will reap the largest rewards.