Thoughts From The Spatial Sidelines
A 1.674Mb PDF of this article as it appeared in the magazine complete with images is available by clicking HERE
It is hard to imagine that over a decade has passed since the first launch of purpose designed and dedicated mobile mapping systems. The realization of the pace of passage of time can send shudders down one’s spine to think of how much has transpired. This article is an overview of Figure 1: DARPA Urban Challenge 2007 winner, team Tartan how mobile mapping, and Racing from Carnegie Melon University, Pittsburgh specifically mobile LiDAR systems have evolved over the past decade.
Back in 2005, the first DARPA Grand Challenge spawned all kinds of new technologies, most importantly in the mobile LiDAR arena. There were many mobile mapping systems prior to this, but these were oriented to video roadway asset management, pavement condition and general maintenance. The best known of these was the Roadware company (now Fugro Roadware since being acquired in 2008), then based in Paris, Ontario, however the systems built for pavement maintenance were highly task specific, to deliver asset inventory information and pavement or road surface conditions.
An exploratory meeting with Roadware executives about developing a LiDAR sensor specifically for these tasks elicited the following response, Sure, our clients would love the idea, but they dont even realize how quickly they will choke on the data. They dont even have the infrastructure to handle what we deliver in imagery and pavement profiles today. They always want more but dont understand the ramifications of all that data. This is why we also do so much data processing work for them. This all sounds so familiar to industry veterans, choking on data, removable disk drive networks (previously known as floppy networks).
The most interesting thing that came out of the first two DARPA Challenges was the realization of how LiDAR sensors were a huge benefit to the autonomous negotiation techniques being employed. One very important outcome was the development of the original Velodyne sensors, the same sensors we still see today on Google’s self-driving cars.
Team Mojavaton approached Optech, now a Teledyne company, to explore if they could collaborate on the LiDAR side of track recognition initiatives in the original DARPA Grand Challenge of 2005, using a different approach than most of the other entrants. The other teams were using the LiDAR sensors for dynamic real-time sensing, but Mojavaton was opting for an intelligent "brute force" navigation model where the vehicle would rely solely on stereo vision and GPS/INS devices for normal navigation. When obstacles and roadway impediments could not be immediately reconciled, the vehicle would stop and make a scan of the area, then evaluate the scene’s 3D layout and plot a safe course around the area in question. Their strategy worked very well, resulting in the fastest running time of the semifinals until about 7 miles into the course. An accelerator servo motor failed and the vehicle was unfortunately sidelined. The 2007 Urban Challenge winner from Carnegie Mellon (using Velodyne and Sick/Ibeo sensors) and a runner up sponsored by Red Bull and Applanix (using Riegl and Velodyne sensors) are shown in Figure 1 .
The DARPA challenge was, it is thought by many, to be the single most important factor in the adoption of LiDAR for mobile mapping through wide visibility brought to primarily the Velodyne and SICK/IBEO sensors. While there were custom built systems such as those developed by Terrapoint in Ottawa (the TITAN used in Afghanistan during the early 2000’s,) and 3D Laser Mapping UK (the Street Mapper, their 2005 prototype shown in Figure 2 ) among others, these were all using terrestrial types of single-profile scanners most often from Riegl, Austria.
A main part of the discussion for new product development that the author was involved in revolved around the viability of developing a task specific sensor and scanning system. A simple comparison of the cost of deployment and infrastructure overhead of airborne versus mobile systems was developed (Table 1).
While this is a somewhat simplified comparison, it is very clear that the cost barrier for a surveying company to entry for mobile systems business is at very least $1,500,000 less than airborne. Mobile will never be able to replace airborne systems entirely, the same can be said about airborne never replacing mobile, especially when it comes to point density and location accessibility in urban areas. Now, those dynamics will eventually change with the dramatic uptake of UAS/UAV technology and the quickly emerging software which takes imagery to create high quality pointclouds from commercial cameras.
LiDAR Sensor History Overview
The years 2006 and 2007 brought a major industry change with the first purpose designed sensors and complete scanning system emerging. The first mobile deployment using a FARO scanner (then the model 880) was from Siteco, Italy (Figure 3) in 2007. Starting with test development systems, Optech used its long range terrestrial scanner, the ILRIS with special roof mount on a minivan as the system development proof-of-concept (Figure 4). With this, they worked out the control and INS systems while engineering was carried out on building the spinning mirror sensors later to be named the Lynx. The proof-of-concept was publicly presented at conferences for over a year before the radically different Lynx sensors were launched. Here is an interview from a SparPoint Research 2007 report on Optech’s mobile initiatives;
"Gelhar took care to point out that Lynx can be mounted on any kind of vehicle surveying from an automobile, a rail vehicle, even a boat is within its reach. Gelhar sees demand for this capability emerging not only from traditional survey requirements, but also from new requirements for populating applications such as Google Earth and Microsoft Virtual Earth tools that are inherently 3D and thus require more than just pretty pictures."
It was not until about full year later that competing sensor heads with similar performance being launched by companies like Riegl, who have since gone on to follow through with a variety of very impressive designs for mobile and airborne applications. Figure 5 shows a 3D Laser Mapping built system employing Riegl’s response to the Optech Lynx sensor. A very innovative lower cost system with the DynaScan, was launched in 2011 by MDL (part of Renishaw since 2013) which opened many doors for the users who did not need the extreme performance of existing systems and were satisfied with limited performance. This system is now marketed by Trimble under their MX2 model name. MDL followed through in 2012 with the prototype launch of their Dynascan-FARO2 model, leveraging the cost and performance of the ever popular FARO Focus tripod scanner with GPS/INS technology, however this never made it completely to market for whatever reasons. Velodyne has in the past 3 years launched it’s "Puck" compact sensor, Quantisys has recently released their M8 sensor and Tyto LiDAR has "OWL" compact high performance sensor. What is interesting to see is that only in the past 3-4 years have there really emerged dramatically new sensor designs, driving performance up and price down. The mainstay of mobile mapping systems, whether built by sensor manufacturers themselves (like Optech and Riegl) or by systems integrators (3D Laser Mapping, Topcon, Trimble, SITECO, Mandli, etc.)
Acquisitions, acquisitions, acquisitions…
It is always interesting to watch as markets develop through the different business cycles. In an article written by Gelhar about a year back for LiDAR news3, he spoke of in-house product development for growing market segments versus market entry by acquisition. Mobile mapping has been the perfect arena to observe how this works. Optech, Velodyne and Riegl to the approach of developing completely new sensors. Leica Geosystems entered the mobile mapping market by acquiring the Italian GeoSoft S.r.l company in June 2014. Trimble Navigation acquired the Montreal based Geo-3D Inc. company which had developed video based systems with software for mobile infrastructure mapping, very similar to the Fugro Roadware types of systems. Trimble and Topcon have extensively added to their product portfolios by further developing products and market areas which they have entered through the acquisition of new technologies, and then adding value through OEM arrangements and house branding devices which give them a broader market reach.
Most disruptive technologies do not come from the larger multi-nationals, as we can see with mobile mapping system development. Tom Greaves, the founder of SparPoint research, during the Spar 2006 conference in Houston when the Optech Lynx system was first released commented that this was a truly disruptive technology. It was only Optech and Velodyne at that point with purpose built systems. A year later Riegl entered with new sensor launches, and it was about 2 years after that the market really took off with the larger multinationals (Leica, Trimble, Topcon, etc.) offering their own house labeled systems, most typically built around the Riegl and Velodyne systems. Since that time, there have been so many company acquisitions for both hardware and software, speaking volumes about market growth following the success of the first mobile projects.
Software Advances
Companies like Terrasolid, Gexcel, Cardinal Systems, Certanty3D, Virtual Geomatics, GeoCue, among numerous others, have been at the forefront of value added processing features. Most of the sensor and systems manufacturers have for many years relied heavily on these companies to provide tools for specific functions like sensor boresighting, calibrations, ground truth integration, data or project management and feature extraction. It has not been uncommon to see a mobile sensor operator needing to buy 3-4 of these individual packages, not to mention the Microstation, Autodesk, ESRI, ArcGIS platforms that some of them have been built on top of.
What has really stood out is the relatively slow pace of really automatic feature functionality development. For many years, the expectation of system operators has been fully automatic functions (and many unrealized sales promises). Imagine simply driving down the INS trajectory and automatically look for signage, identifying the position, type and relative age of the sign. So many of these functions are still performed manually by humans, much the same way it has been done for decades. This is due to a combination of the difficulty in creating feature extraction algorithms in the 3D space, the cost of reviewing and correcting automated errors and reluctance to change from the traditional workflow. This does confirm that the human brain is a really amazing tool, still being able to solve very complex cognitive problems (i.e. 3D object discrimination and identification) quickly with only limited amounts of training.
What’s Up Next?
The last few years have seen some really interesting offerings in mobile packages from companies like LiDARUSA and Mandli, offering a "cookbook" solution to matching sensors, INS systems and software. The one newer approach that has really stood out recently is what an Italian company, SITECO S.r.l., is offering. Based in Bologna, they have actually been performing thousands of miles roadway infrastructure surveys and building their own systems. Their inhouse developed software, including the core calibration, boresighting and project planning software has been in action for more than 10 years. Their most recent developments have been to move away from the data collection service business, to focus on being a system manufacturer, building and delivering extremely flexible configurations. A survey operator can configure different INS components, and even have different LiDAR sensors attached. Systems have been delivered to the likes of Amberg, the tunnel and rail experts, allowing the use of up to 3 FARO Focus sensors (Figure 6), 2 Zoller & Froelich Profilers (Figure 7) scanners, or even a pair of Riegl scanners (Figure 8), all on the same compact system base. For a survey company which owns a number of existing tripod mounted sensors it is now possible to expand into the mobile market. This will become more and more of the trend, not having to purchase additional sensors, software and provide more operator training but rather making greater use of the instrument inventory that has already been invested in.
The days of the integrated system manufacturers are by no means ending any time soon. We are watching market forces at work here, customer demands of easily mixing and matching of hardware tools already existing in their inventory, extracting the maximum return on investment. This is no different than the market consolidation that has been happening in the airborne sensor business over the past decade, being able to more flexibly configure your flight payload with the sensors that are more to an operator’s requirements not those offered as fixed packages from a single manufacturer.
The author would like to extend special thanks to Dave Adams of Teledyne Optech Inc., Ing. Augusto Burchi of Siteco S.r.l. and Dr. Graham Hunter of 3D Laser Mapping Inc. for the quick help in providing background information and photographs from their archives.
Brent Gelhar is a technology commercialization consultant and executive business coach. He is currently involved with a variety of diverse technology start-ups. Brent works at his Toronto-based consultancy Spatial Initiatives Consulting www.spatialinitiatives.org and can be reached at gelhar@spatialinitiatives.org.
A 1.674Mb PDF of this article as it appeared in the magazine complete with images is available by clicking HERE