Expectations

General

No DVL product is perfect.

The Tracker 650 is designed to be excellent at holding an ROV in position in a single spot and to provide odometry to an ROV. The Tracker 650 only does part of that task, the ROV and its control system need to do the rest.

You can expect in position hold for the ROV to "wander" slowly about a mean location with a short-term circular error of less than a 1 meter radius. Longer term, you can expect a drift of less than 1 meter per minute. Both of these values depend heavily on how you have tuned the ROV control system, the nature of the seabed (target plane), vegetation cover, standoff height, and other factors.

Some graphs from an on-water test are below. The sensor was flying about 0.5 meters below the surface, attached to a BlueROV2 which was in orientation-hold mode to keep it level. The ROV was dragged over the test course by a boat moving at relatively constant speed. The ROV was dragged backwards with a 20-degree or so angular offset due to the tether being attached to the rear port side of the ROV.

The total course length was 435 meters. The test started and ended in shallower areas with vegetation. Generally, areas with altitudes of less than 6.5m are over vegetation, and the areas of altitude over 6.5m are over muck bottom.

All the graphs below are from the same test.

Note that on this test day we were able to continue out to water of about 22 meters deep with similar results. That data not shown here.

Confidence represents the Tracker 650's estimate of how accurate the instantaneous measurement is. Possible values range from 0 (don't use) to 100 (as confident as possible). The confidence in this test varies from about 50 to 99. The areas of lower confidence generally correspond to areas with heavy vegetation.
Expectations
Standoff represents the altitude of the sensor above the seabed (when the sensor is mounted in the default orientation of pointing down). The shallower areas generally have lower confidence in this test due to vegetation on the seabed.
Delta X is the distance the sensor moved along the sensor X axis in the previous sample period. Deltas are also available for the Y and Z axes. In this case all samples are 0.2 seconds long. The Delta X position started around -0.1 meters per sample and decreased to -0.04 meters per sample at the end. Although this looks noisy (and it is), see the next chart.
The ROV essentially integrates delta position into position. The total displacement in X is 398 meters over the 435 test course. The reason it is not the length of the test course is due to the rotation of the ROV which causes some of the displacement to show up in delta Y. The ROV's dead-reckoning function agreed exactly with the GPS ground-truth for this test.

Last updated