Tuck Mapping Automated High Precision Lidar Boresight Alignment Calibration
In 2008, Tuck Mapping took delivery of a Riegl LMS-Q560 full wave form 164 KHz Airborne laser scanner. One of the options featured with the system was Riegl’s new technique for determination of boresight alignment. The new technique relies upon finding common planar surfaces (e.g. building roofs) in the data and processing several flight lines in the algorithms which results in a robust estimation of the systems boresight angles.
Recent years have brought dramatic improvements in lidar performance related to increased density of points from higher performing laser sources and signal processing advances such as “full wave form” Lidar scanners. However, workflows and QA/QC methods have not kept pace with the technological advances.
Tuck Mapping of Virginia, which specializes in high density/high accuracy airborne Lidar mapping, quickly found the limits of these traditional methods. Bobby Tuck, the president of Tuck Mapping, searched for an innovative and cost-effective solution to evaluate boresight alignment and a way to accurately perform the calibration necessary to match the technology advances and customer expectations.
The principle challenge was to determine the angular alignment between the inertial measurement unit (IMU) and the laser scanner. The new method needed to accommodate a variety of variables, such as altitude, speed, and weather, and still be sufficiently accurate. In addition it needed to be portable and not tied to a test range.
For many years the industry relied upon proprietary procedures such as the expertise of personnel and/or software solutions to address data anomalies caused by system misalignment. Surveyed test sites, use of retro-reflective targets, scanning objects of known size and position all lacked flexibility and required much effort. Measurement errors of reference targets decreased confidence in this approach. Currently the assessment of the boresight angles is often not automated and based on trial and error algorithms. Data strip adjustments are performed primarily at the processing level focusing on improving alignment of overlapping strips. As such these processes are often not successful in addressing the root system level cause of the misalignment.
To test the new method Tuck Mapping created a test field and ground control plan after consulting ASPRS Guidelines for vertical and horizontal accuracies, NDEP guidelines for digital elevation data, IAPRS reports on Lidar, FEMA Flood Hazard Mapping Program, NSSDA, and the practical experience of the firm. The plan called for over 36 ground control objects located on varying elevations and with varying degrees of slope change, buildings, roads, trees, and vegetation cover. The control points were surveyed in using a Topcon reflectorless total station under the command of the lead surveyor Kenneth Sorrels. The format was NAD83 - State Plane Coordinates - Virginia South.
The Riegl LMS-Q560 full wave form scanner was integrated with an Applanix 410 GPS/IMU system in an instrument pod under the Bell long range (L3) turbine helicopter. Basic system measurements such as instrument position and lever arms with GPS antennae locations were determined. A series of test flights at varying altitudes and speeds were flown by Brian Fox and Frank Kendall. The base station for the trajectory was a Leica System 300 GPS base station.
The trajectory from the Applanix GPS/IMU system was recorded along with the data from the Riegl scanner. The algorithm looks for the plane patches and then links the corresponding plane patches from different flight lines. An iterative closest point algorithm is used to identify and quantify the misalignment angles. An optimization algorithm adjusts boresight alignment and realigns data through several iterations thereby minimizing the mean square distance error between the planar surfaces. The resulting process involves thousands of observations resulting in a very robust method. The system outputs a standard deviation between the planes, a histogram of the residues and of course an orientation chart of the data to assess any imbalance. The new methodology creates a calibration for the deviation angles for roll, pitch and yaw of the sensors.
The method facilitates a fast determination of boresight misalignment, incorrect reference data from a terrestrial GPS base station, any bias or drift of IMU data and erroneous determination of any lever arm. The method can be performed anywhere suitable flat surfaces can be found. The accuracy of the estimated boresight calibration depends strongly on the quality of the position and attitude data from the GPS/IMU system such as the Applanix.
Tuck Mapping has completed two recent projects for Virginia Department of Transportation and demonstrated the accuracy of the system. One 57-mile corridor found control checks on a section of the corridor mapped resulted in an unadjusted root mean square (RMSE) of 1.2 inches. On the Tuck Mapping website www.tuckmapping.com there is a report on the vertical accuracy of a number of additional projects.
The accuracy and precision of this new automated workflow has provided Tuck Mapping with the ability to exceed customer expectations for accuracy and fast project turnaround.
» Back to our Aerial Mapping March 2009 Issue