Feature: The Power of Mobile Scanning
Professional Surveyor Magazine - April 2012
Automated feature extraction turbocharges efficiency for a major Caltran bridge project.
By Cyn René Whitfield
Automation has never lost its “wow” factor. It’s human nature to search for ways to make life easier with less effort, and it’s a natural progression that the surveyors and engineers who build and maintain infrastructure look to automated technology to do so.
In 2010, when the California Department of Transportation (Caltrans
) hired the Terrametrix survey team to use terrestrial mobile lidar scanning (TMLS) to document hundreds of bridge clearances, they used the power of automation. “This massive project needed a solution to be just as efficient in the office as it is in the field,” said Michael Frecks, PLS, a surveyor of 35 years and president of Terrametrix, LLC
. This need soon expanded nationwide.
Revised Federal Standards
Driving the need were the revised standards of the Federal Highway Administration bridge inspection program. The regulations were developed as a result of the Federal-Aid Highway Act of 1968, but 2005 revisions catapulted an effort to unify a nationwide inventory database. The act required the Secretary of Transportation to establish national bridge inspection standards with a primary purpose to locate and evaluate existing bridge deficiencies.
California, more specifically Caltrans, had a huge undertaking. A federal standard requires the longitude and latitude location of each bridge plus minimum clearance, both horizontal and vertically. According to the latest figures published by the U.S. Bureau of Statistics 2010, the national transportation network consist of 601,392 highway bridges and overpasses intermittently placed within 46,934 miles of interstate highway and 116,837 miles of national highway system roads.
In California, there are 0.644 bridges per every 1,000 people; that is a lot of safety responsibility. Therefore, the use of TMLS was specifically requested in the solicitation for obtaining horizontal and vertical clearance measurements. Caltrans wanted documentation at highway speeds, transparent to the traveling public, without impeding traffic. This meant traditional methods were not an option. The data also had to be in compliance with national bridge inspection standards, which also meant accuracies of 1” vertically and 3” horizontally.
From a system-wide perspective, traditionally collected clearance information in many cases is incomplete, inaccurate, difficult to maintain, and difficult to access, not to mention notoriously dangerous. Obtaining survey-grade accuracy at traffic speeds is the most proven and efficient way to accurately measure various structures and roadway features.
“The TMLS technology provided by StreetMapper 360 is a powerhouse that can do a large engineering company’s or DOT’s field work for an entire year in a matter of days,” said Frecks. “As the technology moves the surveyor safely ‘out of the red zone’ and into the office for feature extraction and line work, the amount of data can be overwhelming.”
Overwhelming is an understatement. In addition to the required clearances, data was being obtained along the route that totaled 11,300 miles. When all of Caltrans’ 7,250 bridges are completed Terrametrix will have collected 531 terabytes of total lidar data and 28 terabytes of imagery. (California highways may then be the most documented state in America.) That kind of comprehensive data at those speeds equates to a massive amount of man-hours to extract the data at the desktop and obtain useful information. For all 7,250 bridges, it is estimated that more than 100,000 measurements saved 1.2 million manual mouse clicks!
TMLS Answers the Challenge
TMLS was the only safe and accurate solution for obtaining the massive amount of data needed to extract bridge clearances. “Out of the red zone” benefits include the ability to drive with the traveling public while collecting data (this makes for a safer profession and a safer environment for the public). The success of what began as an introduction of TMLS on 568 bridges in California—coupled with state-wide guidelines for TMLS usage developed prior to this project—has made Caltrans very forward thinking. More requests for TMLS from local agencies, consultants, and internal Caltrans professionals validate the power of the technology.
Terrametrix was gathering vertical and horizontal dimensions at the rate of 100 bridges a day for Caltrans. A similar project managed by Frecks for Nebraska Department of Roads in 2002 using static scanners took three months, and the surveyor worked alongside the highway in the red zone. “Using the static technology today in places without shoulders and [with] heavy traffic conditions like Los Angeles and New York would not even be an option,” said Frecks. “That same project today using TMLS technology and our automated extraction routines would take less than a week.”
TMLS, like its static predecessor that collected up to 150,000 measureable 3D points (x,y,z) per second, now has the capability of collecting a million points per second. Traditionally, surveyors collected a shot every 25 feet, which equated to one entry in the field book and three minutes of a three-man field crew. Drafting in the office was additional man-hours, taking the data from the field book to drawing. Then, with total stations it became a few bytes per shot and 30 seconds of a two-man field crew. After the digital data was downloaded the resulting CAD drawing had line work and features that were coded by the field crew to an 80% to 85% completed drawing.
As the volume and speed of data increased to 600,000 points per second, at traffic speed, collecting 80 to 100 data points per square foot with TMLS, the process of sifting through it all was a natural progression for automation. The richness of data results in a 3D raster image that requires data extraction by office technicians for drawing completion.
“We needed a way for the office technician to keep up with the speed of the field data acquisition,” says Frecks. Static scan ratio of field to office is 1 to 10 (one day of field documentation took 10 technician days in the office). “With TMLS that ratio can be farther apart given the fact that the vehicle is collecting data at 45 to 65 mph. We also needed to be more accurate rather than the traditional surveyor in traffic guessing where to take a minimum clearance overhead shot.” The solution was a marriage of TMLS and automated extraction.
From Robots to Lidar
, based in Pittsburgh, Pennsylvania, provides the solution to large-scale projects. Founder Dr. Aaron Morris derived Allpoint’s Perception Engine from an autonomous robot program that mapped spaces with 3D lidar. It automatically segments, filters, classifies, extracts, and analyzes large and complicated sets of static and mobile terrestrial lidar data using algorithms.
“We target the unnecessary cost and complexities associated with processing large-scale lidar data, which allows companies to increase their efficiency, expand business or project capacity, and ultimately enable greater use of 3D laser data,” said Morris. “Workflows are compressed in a fraction of the time required for existing methods [hours vs. months].
“Due to the sheer volume and density of the data, processing and manipulating TMLS 3D scans is the most complex—and therefore hardest to automate,” said Morris. The platform provides tools and applications to surveyors and engineers who are using TMLS techniques for survey documentation, measurement, analysis, and asset management.
For the Caltrans project, “We talked about their processes and the estimated time it would take to process the first 600 bridges,” explained Morris. “Once we had a common understanding of the inputs and desired outputs, we knew we could streamline the fundamental workflows and reduce time associated with processing by an order of magnitude over manual methods.”
The work process reduces all the data to 300 megabytes per bridge, equating to a total of two terabytes for all 7,250 specific bridge areas of interest. This process involves importing trajectories, matching flight lines, and synchronizing imagery for automated extraction. The reports can be generated at a rate 10 to 20 times faster than traditional extraction and can be viewed on a mobile device, making for easy field verification.
“There are basic surveyor practices and scanning techniques we apply through our experience,” said project manager Todd Gnuse. “Mission planning is crucial to the documentation process. When you consider the scanners are running for the entire route we will have actually documented almost the entire California interstate system. Then, in the office, the software techniques we use to process the data are a large part of what we do to assure accuracy. There is a lot of tedious work.”
Extracting data is a monotonous process of looking through numerous amounts of data and files, then isolating the parts you need and leaving out the unnecessary ones. Every technician is familiar with this type of data extraction. It is a simple task—but it is a time-consuming one.
According to Frecks, the value of the automaton is in removing the mundane tasks while pertinent data is still in the control of the decision makers. He wanted human interaction, not just a complete run of programming software. “We need the surveyor’s ability and training in spatial recognition to interact with the automation to assure the identification of the data. The Perception Engine, while it automates repetitive tasks in the background, provides human interaction at key points of the extraction and reporting workflows. This gives us the ability, as surveyors, to verify completeness and accuracy.”
Frecks is often asked about automated extraction when he travels the country. Automated feature extraction allows software to recognize certain specific objects or low points that are represented in digitized imagery or point clouds from TMLS data. Processing 3D laser-scanned measurements using complex statistical algorithms can identify road surfaces, isolate bridge data, identify lane markings, and remove noise (such as traffic on the highway). The application can then extract cross sections from the data and calculate clearances and other user-defined measurements.
Frecks cautions that it is vitally important that any automated extraction be 100% accurate. “Imagine if you had 10 features and 1% of them were prone to error. Now imagine you have one million features. What does that 1% mean to your confidence level? How do you identify that 1%? You would have to sift through all the data to find the missing/erroneous features. It is possible now to let the software perform all mundane key strokes with the human oversight—the capability of spatial recognition that surveyors are trained in—as added quality assurance.”
Cyn René Whitfield is a journalist involved in marketing for land surveying and laser scanning companies for the past 27 years. She is currently the marketing coordinator for Terrametrix LLC in Omaha, NE.
StreetMapper is a 3D mapping system that uses laser scanning technology combined with a precision navigation system, advanced data processing software, and an innovative system design to scan highways, infrastructure, buildings, and vegetation from a moving vehicle. StreetMapper is a product of 3D Laser Mapping Ltd
. based in the United Kingdom, with offices in South Africa, Australia, and North America as well as a global network of distributors.
» Back to our April 2012 Issue