Hi,
I have a question about using a RTK gps system for a quadricopter project.
I hope i'm in the right category for that.
We have to precisely position the drone (3m +/- 10cm) around a wing of a
windturbine for taking high quality pictures.
We have mounted a rotating lidar (RPLidar A2) on our Erle Copter to get a
position feedback on the pitch and roll axis, that we run through a P controller
to send velocity commands to ArduCopter. The commands are sent in Python through
DroneKit using MAV_FRAME_BODY_OFFSET_NED messages.
The problem is that, while this system works well in our Gazebo simulation, in
the physical world the drone is having an erratic behaviour, with wide random
movements. Enough for it to get out of range of the sensor or crash into the
wind turbine.
Our current assumption is that the regular GPS that we are using along with
ArduCoper's EKF is not precise enough to position the drone around the wind
turbine.
We have thought so far of two possible solutions: the first one, that we like
less, would be to use the GUIDED_NoGPS mode to send angle commands, develop our
own PID in python and infer the current speed from the Lidar measurement. A lot
of additional dev, uncertainity and no way to simulate this as the GUIDED_NoGPS
mode is not implemented in the Ardupilot version that we are using in Gazebo.
The other solution that we have thought of would be to use a RTK gps to get more
precision. This is why we would like some advice before spending 600+€ on it.
So, what do you think ? Is it likely that the cause of our problem is the
precision of the GPS ? Do you think that using a RTK would solve our problem ?
Would it even work for altitude instead of using a vertical rangefinder ? Have
you already faced a similar situation, and if yes how have you solved it ? Or
could you point us to somebody / somewhere we might get help from ?
Thanks in advance for your help.
Replies
Maxime Bourgeois said:
Maxime Bourgeois said: Tallysman 3710 for base and 2710 for rover, better S/N and less multipathing. You can try first with the supplied antennas maybe they are good enough in your environment, you can upgrade any time.
Two more questions about the Emlid Reach:
- In the docs they advise to not use it as a primary GPS but only as a secondary, however I only have one serial port on my autopilot to connect a GPS. How did you solve that problem ? Is it still OK to use it as a primary GPS, in spite of their warning ?
- If you are not using the basic Emlid antennas, which ones are you using for better results ? How much better are the results, in what regards ?
Though there are two points I didn't understand in your answer:
- What do you mean by " IF you have a Fix solution" ? Is it that hard to get a fix with a RTK system ?
- What is the "protection ring" that you are mentioning ? Are you referring to a "choke ring antenna", or is it something else to put around the UAV rather than around the antenna ?
If you were referring to a choke ring antenna, do you have one that you recommend would work with an Emlid Reach or a Here+ ?
On another note, we are between a Emlid Reach or a Here+ (which uses an Ublox m8p) for our RTK system, is there one that you recommend more ?
Hi,
the CEP of a standard GNSS is 2.5m, which translates into more than 2.5m error 50% of the time ! This assumes clear sky view, which you don't have as the mast and to some extent the wing will partly block the view for the GNSS, making things worse.
A RTK like Emlid reach can give you 25mm in an ideal world IF you have a Fix solution.
Usually the UAV for this kind of operation has a protection ring around it. Helicopters would be much better for flying in windy areas, but are MUCH more demanding than multicopters.