I’d like to start a conversation about the accuracy of coordinates recorded by acoustic Waterlinked GPS and the error in mission planning.
I’d like to share here the testing result using UWGPS for locating the position of the BlueROV2.
The robot testing result using UWGPS and mission planning:
As you can see, the location of BlueROV2 visualized in QGround and Waterlinked is somehow different (a few meters). I wonder which one is more accurate in this. QGround or Waterlinked?
When we tested the mission planning, we let the BlueROV2 move forward but it turned out that the robot moved in a different way, which is backward. We tried many times but the mission planning still failed and got the robot in the wrong direction as we set it up at the beginning. Could you please suggest us some problems with this mission planning?
Hi!
To check for errors, use the diagnostic page of Waterlinked.
Also there, click the “stats for nerds” option to find out more clues.
One likely error might be setup of receivers. http://192.168.2.94/#/diagnostic
The positions in the Water Linked interface will be from the position estimates of the device, whereas the positions displayed by QGroundControl will be those reported by the autopilot, which are filtered based on its other sensor readings and the passage of time.
The autopilot runs an “Extended Kalman Filter” (EKF), which is used to estimate the vehicle state at high frequency, and combines the readings of several sensors and their perceived uncertainties to achieve that. As a general principle, an absolute position sensor could be more accurate for any (or even every) position estimate it provides, but the autopilot’s estimates will be higher frequency, represent more of the vehicle state (e.g. orientation, velocity, acceleration, etc), and should provide a generally more “reasonable” estimate of the vehicle’s path.
It would help to know which ArduSub and QGroundControl versions you’re trying to use.
