A 2.873Mb PDF of this article as it appeared in the magazine—complete with images—is available by clicking HERE
Today, we live in a world where the trend is to have the latest or the best currently available gadget, be that a humble MP3 player, a digital camera or a cell phone. While this trend is most likely quite obvious to users of consumer level electronics, it should be noted that it is also a trend that is reflected in many professional areas, including the supply of geoinformation. The demand for increased resolution, detail, and accuracy has resulted in many technological developments for geoinformation sensors, ranging from new digital survey cameras to very high repetition rate airborne lasers.
Access to geoinformation products has increased as well. Only a few years ago access to satellite or aerial imagery was limited to a few experts, in contrast to today where everyone has access to high resolution imagery via Google Earth or Microsoft Virtual Earth for almost every corner of the globe. The trend now pushes even farther ahead, toward the visualization of a detailed 3D world, not a 2D one.
Recent airborne sensor advances have contributed to the development and creation of city scale 3D models for visualization purposes. But are they sufficient? Do they allow these 3D visualizations to become truly immersive environments? Can you, for example, stand on a street corner and really visualize what that street looks like from the vantage point of a person? The answer is "yes" in some limited cases, where very detailed modeling work has been undertaken (usually with a significant amount of labor), but more often than not the answer is "no."
Recently launched new technology, in the form of the Optech Lynx Mobile Mapper, however, has the potential to change this, as it is now possible to capture amazing street level detail rapidly and consistently from a moving vehicle. Although mobile lidar is not a new subject, systems to date have had limited capability in terms of resolution (point density), field of view and mobility. The term "mobile" has been applied to vehicle-mounted lidar systems before, but its application has not been consistent, referring often to a terrestrial lidar system that has been incorporated onto a vehicle–making deployment easier–but the system itself often has to be stationary to make measurements. Today, mobile lidar really refers to lidar sensors that can make measurements from moving vehicles.
There are many versions of mobile lidar sensors. Some are best suited primarily for 3D visualization. The generated data is inherently 3D but typically has a range resolution of no better then 2-5cm. The data generated from these systems is suitable for visualization purposes but not for surveying/engineering applications requiring precise measurement extraction.
The Lynx Mobile Mapper, however, is capable of engineering grade data accuracy while being collected from a moving vehicle.
Infoterra Ltd in the UK, a large commercial geoinformation service business, has taken delivery of Optech’s first Lynx Mobile Mapper system made up of two 100Khz rotating laser sensors combined with two high frame rate imaging cameras, allowing colorized point clouds to be captured from vehicle speeds ranging from walking pace to 60mph. As Infoterra’s Technical Director, I have worked closely with Optech during the later stages of the sensor development and I am very encouraged by the new system. Initial data collections have been outstanding, even on very narrow densely packed UK city streets. Previously, this was only possible using static scanning.
Each of the two laser sensors makes up to 100,000 point measurements per second by utilizing a new 9000rpm rotating mirror configuration. Each laser emits a disc of points behind the vehicle out to a radius of 100 meters. The fact that each laser emits a disc of points that is key to the success of this system, as this means that the laser can record detail up to 100 meters up the front of a building, which might only be offset to the side of the vehicle by a couple of meters. The field of view of the sensors are not restricted to looking either down or backwards behind the vehicle like some systems. The configuration of the two sensors is also critical, as the lasers are oriented 45 degrees to the axis of vehicle motion, and inclined forward by 10 degrees. This means that the two laser discs cross each other’s field of view. Driving past an object once, therefore, means that it is imaged twice by the two sensors. However, the object is not imaged by the two sensors at the same time, which in itself is an advantage. Take, for example, a tree at the side of the road. Naturally there is a shadow behind the tree when the laser view is obstructed, but because each laser is obstructed at a different time it means that the shadow area behind the tree is greatly reduced as one sensor is still seeing behind the tree when the other sensor is obscured. The benefit of this is really demonstrated in an urban context, when it is still possible to see detail on the front of a building, even though there is a large tree directly outside the building. Inclining the sensors also means that detail is collected on the front of objects, such as bridges when the vehicle passes underneath.
Point density is obviously a function of vehicle speed as well as distance of the object away from the sensor, and to a certain degree the orientation of the object in the sensors’ field of view. Initial data collections at 20mph have produced very good point densities across all surfaces. The highest point densities naturally fall in the overlap region of the two sensors directly behind the vehicle on the road surface. But point densities on vertical surfaces parallel to the vehicle (i.e., building fronts) are very impressive (e.g., approximately 1,500 points per meter at ground level and 600 at 2 meters vertical height, and 140 points per meter at 15m vertical height, for a wall offset 3 meters to the side of the vehicle). In areas where the vehicle has been driven at walking speed, point densities have exceeded 6,000 points per meter on the road surface in the sensor overlap region.
Turning Theory into Operational Reality
Infoterra Ltd has always been an early adopter of new technology, being one of the first companies in Europe to offer a commercial airborne lidar capability. But like all new technologies, transferring them from the R&D area of the business to mainstream operations always poses a few logistical hurdles. In the case of Infoterra’s new Rapid Surveyor product offering (which incorporates the Lynx Mobile Mapper) the operational challenges fall into two categories. The first is the construction of operational support around the service (that is, how do we pre-plan surveys, how do we estimate survey costs, how do we resource these?). These discussions bring other interesting issues to light, such as insurance coveragehow do we make sure the vehicle and sensors are secure overnight when working away from the normal office base?
The second set of challenges centers around data dissemination. The system is able to collect large volumes of data, given that data is being recorded from two lasers, two cameras and the navigation system at any one time. Data processing to point cloud is not really an issue, because ultimately the data processing methodology is the same as those already employed for airborne lidar data processing, however the use of the imagery at high acquisition frame rates for colorization of point cloud is a new addition.
Ultimately the challenge is not how the data is processed but how the information contained in the scans is extracted. This is an area that In
foterra is currently focusing on as there are very few "off the shelf" data extraction tools suitable for use on mobile lidar data. While it might be possible to modify existing tools to extract simple features automatically, like power lines or road edges, tools to extract building facades automatically do not exist and need to be developed. It is the development of these tool kits that will allow the real wealth of information recorded by the system to be fully utilized into a wide range of applications, from city center visualization through to urban redevelopment and security.
Infoterra’s primary reason for investing in this new unproven technology at such an early stage was because of the potential benefit that could be seen by collecting large amounts of precise information at a scale coincident with that at which people interact with the built environment. Such data, up until now, could only be collected using repeated setups of terrestrial laser scanners, making wide area surveys cost prohibitive.
For the first time, this new technology has the real potential to provide realistic visualizations at street level, where all objects are in their precise locations and are a true representation of the built environment, data that can be used to build a truly immersive 3D environment. This data also has great scope for "collect it once, re-use it many times," because so much information is collected in one pass. Even if it is not all used initially, it has the potential to be incorporated into new applications as these are developed over time.
Within Infoterra we have adopted the phrase "world digitalization" which is the phrase we use to describe the trend of capturing a precise digital record of the built environment at an ever increasing resolution and accuracy. With the advent of Lynx Mobile Mapper technology we have certainly taken another major step forward in terms of our own world digitalization capabilities. The ultimate question is: are our clients ready to make the same step change with respect to the products they receive and use on a daily basis? Only time will tell.
Anthony Denniss is the Technical Director for Infoterra Ltd, responsible for in-house R&D and new product development, as well as the evaluation of new mapping technologies and sensors. His academic background is in cartography and geological remote sensing.
A 2.873Mb PDF of this article as it appeared in the magazine—complete with images—is available by clicking HERE