Scanning in Motion
September 1, 2007
Because I am an end user of laser scanning technology, I get excited whenever a manufacturer comes out with something new. Some advances are more exciting than others, but the most thrilling development I have seen in the last five years occurred the other day. At my office, I opened a point cloud that was sent to me from a company that performs dynamic scanning by putting a laser scanner on a moving vehicle--it looked just like a point cloud that was collected from static locations. Knowing firsthand what it is like to spend many days setting up on highways, I have a great appreciation for this new method of data collection. In this article I will explore the technology of dynamic scanning, including the necessary equipment, the accuracy, the current manufacturers and the advantages of using a scanner on the move.
Making it Work on the MoveThe idea of putting a laser scanner on a moving vehicle is not new. Airborne LiDAR has been around since the 1980s when one of the key components of the system, GPS, was developed. This made it possible to precisely calculate the position of airplanes. While airborne LiDAR can produce similar data, airborne operations are expensive and the data is not as detailed as terrestrial data because of the long distance to the area being surveyed. But the concepts are similar. “The only difference between airborne LiDAR systems and terrestrial dynamic LiDAR systems is the vehicle the system is deployed on,” says James P. Peterson II, PE, PLS, vice president of Sanborn Map Company Inc., an aerial and terrestrial LiDAR service provider based in St. Louis. Currently the company combines the two technologies of airborne and terrestrial laser scanning on projects, and hopes to soon be able to use its airborne experience to facilitate scanning from a ground-based vehicle. “It’s the exact same concept,” Peterson says. “We are learning how to best perform static terrestrial LiDAR so we can transfer the technology to a moving vehicle.”
To get a laser scanner to work from a moving platform, four additional components have to be added to a standard static scanner setup: a vehicle, a high-resolution video camera, a navigation system and software that will integrate all of the components. Many different types of vehicles have been used for dynamic scanning, including boats, trains, vans and even unmanned vehicles in the DARPA Grand Challenge across the desert in Nevada. Typically, you need to use a laser scanner that functions in a 2D line mode. The third dimension comes as the vehicle moves. The faster the laser scanner, the closer the data points will be; for this reason, many operators choose to use a phase-based scanner such as a Profiler 5002 from Zoller+Frohlich (Wangen, Germany), which collects up to 625,000 points per second, or a fast time-of-flight scanner like the LMS-Z420i from Riegl USA (Orlando) that collects 8,000 points per second. To help interpret the point cloud data, one or more video cameras are often used. The data from the camera can be time-tagged and synchronized with the laser scanner.
One of the challenges in collecting data on the move is knowing where the scanner is located so the point cloud will be on a known coordinate system. For static scanning, this is accomplished by scanning known targets or setting the scanner up over a known point. For dynamic scanning, the coordinates are derived from a navigation system that typically consists of a high accuracy GPS unit, a gyro-based inertial measurement unit (IMU) and a distance measuring indicator. The GPS not only gives the initial location of the vehicle and continuous locations as it moves, but also helps eliminate drift in the IMU. An IMU detects the rate of acceleration and rotational changes in three axes of the vehicle. The data is fed into a computer and the current position is derived from the previous location. This type of dead reckoning system is prone to errors that are accumulated as the vehicle travels. This drift can be reduced by the solutions from the GPS receiver. The distance measuring indicator also aids in locating the vehicle, especially when there is poor GPS coverage such as in tunnels, under trees or in urban environments. All the navigation information and laser scanned data is merged and synchronized into one system. Some systems take the time stamp from the GPS and apply it to all the collected data so it can be synchronized.
The accuracy of the final point cloud and subsequent drawings and models is dependent on many factors. One factor is the error from a single laser scanner measurement. This relative accuracy can be provided by the laser scanner manufacturer; for example, this can be as small as 10 millimeters for a Riegl scanner. The ability to accurately measure objects in the point cloud is also dependent on the point density, which is affected by vehicle and scanner speed. The slower the vehicle goes, the denser the point cloud will be. The absolute accuracy of the data in relation to a local coordinate system is dominated by the navigation system. This can be as low as 3 centimeters under favorable conditions and might be as high as 0.5 meters under poor conditions.
A Variety of SolutionsSeveral manufacturers, companies and agencies offer dynamic scanning solutions. Optech Inc. (Vaughan, Ontario, Canada) recently announced a dynamic scanner called the ILRIS-3Dmc. The Canadian manufacturer is promoting the use of its motion-compensated scanner for three common applications: stop-and-scan, mobile platform vertical scan (i.e., oil rig from a boat) and mobile platform horizontal scan (i.e., road surface survey from a vehicle). Riegl offers several scanners that can be deployed as dynamic scanners, and that have successfully been used on boats to inventory waterway assets and to geo-reference obstructions (such as semi-submerged rocks) that cannot be mapped directly from a boat. The Nottinghamshire, UK company 3D Laser Mapping has released a system called the StreetMapper Mobile LiDAR mapping system based on Riegl scanners. It offers a turnkey survey vehicle with all the necessary components mounted on it or as a combination of the sensor platform and the electronic rack. The overall weight of its system is about 330 pounds, which is small enough to operate from normal passenger cars.
The Federal Highway Administration’s Turner-Fairbank Research Center is also using dynamic LiDAR. The agency has developed a system called the Digital Highway Measurement vehicle. This multi-sensor system uses laser scanners and Macrotexture lasers (lasers with a submillimeter beam diameter) to profile the texture of highway surfaces. It is also being used to explore the use of new three-dimensional ground penetrating radar for subsurface evaluations down to 20 or 30 feet for locating utilities and pavement thickness. Data collection like this would represent complete roadway cross-section and would be very valuable to highway designers.
All of these systems are fairly expensive. Even though terrestrial LiDAR is less expensive than airborne LiDAR because there are no flight expenses, the components that make up the dynamic system are quite high in cost. Laser scanners typically cost close to $100K, and a good IMU can start around $200K. Despite the cost, laser scanning service providers are anxious to add this to their line of services. “Dynamic scanning is the wave of the future and the not-too-distant future,” says Michael Frecks, vice president of Optira, a provider of high definition as-built documentation services based in Omaha, Neb. “We are waiting to see systems that perform at the subcentimeter level consistently with minimal problems.”
Dynamic AdvantagesThere are many advantages to using dynamic scanning. The first is probably safety. It is a surveyor’s dream to be able to perform a highway survey safely from inside a van with no traffic management required. The next advantage is speed. When we scan highways with static scanning, we cover about 1,000 feet per hour and think we’re pretty fast (plus we spend additional time and effort setting up and locating targets). When I opened a point cloud sample from the StreetMapper system and learned that it took about 15 minutes to scan almost four miles, I was blown away. This ability to safely acquire so much data so quickly will be useful in many different applications including mining, city modeling and shoreline mapping.
There have been many days that I wished our scanner could scan while it was moving. The closest we’ve come to it is stop-and-go scanning in plants when we keep the scanner tripod on a wheeled dolly. We only dreamed about being able to scan while we moved. The way technology continues to develop, I think it’s just a matter of time before the days of static scanning look like the days of stadia and chains. And dynamic scanning will be possible not only for outside projects, but hopefully inside facilities as well.