Topographic mapping from a moving vehicle
Traditional topographic mapping, such as from fixed surveying equipment or aerial photogrammetry, is well established and dependable, but is quite expensive when detailed high-resolution infomation is desired. The cost comes from having a trained operator using expensive equipment that requires either setup and breakdown over days (land surveying) or aerial data gathering that requires much post-processing. For some customers, vehicles are already traversing the area needing to be mapped, so a strap-on topographic mapping system attached to a ground vehicle could save much money/time. Examples are agricultural drainage maps and golf course mapping.
The work below shows a complete system that determines the vehicle position and orientation while simultaneously scanning the ground with a 1D scanning lidar. The prototype logged multiple grades of inertial input, to allow association of mapping accuracy vs system cost; the filtering software will allow many model types of sensor input, for such system tailoring.
The figure below shows the simple testbed on a car, with a SICK laser mounted at the rear pointing down at 45deg, a GPS-RTK system, and 2 different grades of 6dof inertial (mems, and ring laser gyro).
The first test area was a suburban house in a desert environment. We wish to map both the street and sidewalk, as well as the yards on either side of the road.
This shows the cloud of solution points from one pass driving by the yard, visualized from roughly the same viewpoint as the photograph above. Note the plants and walls, as well as the road surface transitions.
Vertical view of the same data, color coded by height.
Exported autocad rendering of the area.
Next, we wished to use this system on a golf course. Below is a typical John Deere fairway mower outfitted with the same navigation package (we were also running a test of network RTK simultaneously, which showed only a small performance degradation while saving the trouble of setting up a local basestation).
Trajectory covered on a test segment of a fairway, superimposed on an aerial photograph.
Note that on subsequent passes there is overlap in altitude estimation, shown below. This acts as a self-contained test of accuracy. The system can also be calibrated using these overlapping altitude estimates, correcting laser mounting parameters and vehicle roll/pitch biases.
Finally, these results were exported in a variety of useful formats. Here is a simple exaggerated rendering of the fairway visualized in matlab.