My focus has been mostly on navigation accuracy, which requires changing your waypoing_nav_rad parameter to less than 1 meter. Doing that causes the drone to take much longer to 'find' the waypoint, but it does work. Preliminary tests show ~15 cm RMSE (horizontal and vertical combined), when the waypoint navigation radius is set to 10 cm. Setting the radius to less than 10cm always causes the drone to spend too much time trying to hit the waypoint. For post-processing assessment of station accuracy (not for navigation) we are able to achieve a 2.6cm RMSE.
really nice!!! could you tell use the parts that we need to make the same setup as yours? Currently have a tarot x8 with everything, just need advice with the RTK.... where can I get the usb rtk hub and the antenna? because as I see here, all I need is the pixhawk 2.1 and the here+ RTK... the module that you connect to the computer, that's where I have no Idea about. Thanks and good job!!!
I don't get it... what are you measuring? You didn't even mark the home point with anything as far as I can tell. Looking at the flight logs doesn't will only tell you how the aircraft thinks it flew, not what it really did...
iforce2d Ahh, you are right that I didn't mark the home point. I used the mission planner software to survey in the point so that I could repeat waypoints based on the relative, not absolute, position, of the points. I conducted multiple flights during the same day in order to test this, with no intent of coming back on another day. For what it is worth, the s1000 and probably any drone on the market right now, is unable to repeat the waypoints at the same positional precision as the HERE+ rtk. This is most likely do to wind conditions, motor balance, and many other factors. This was my very first flight with this system, and since then I've done work using geodetic benchmarks and supplemented with rtk based survey points in imagery. Even with a 2cm absolute positional accuracy, repeatability of waypoints are typically 0.33m horizontally and vertically at best. Hopefully now you get it :)
Ah, sorry for the late response. We were not measuring absolute positional accuracy. We are interested in measuring the repeatability of navigation to specific waypoints, meaning the accuracy of the coordinates of the home point are less relevant, as long as they don't change between flights.
Would like to slightly extend arms, upgrade motor/esc/prop combo, change battery system to a quad setup with quick change and hotswap abilities -2 above and 2 below and each pair on thier own quick release bracket, Print a rain cover, to fill top board gaps and seal sides & waterproof electronics. And swap to Cube 2.1 & Here + with Connex TX and Audrino components for additional features. Just need a gantry CNC and I’m away. The expandable battery banks that the Matrice batteries clip into and the bank slides into a housing raises my interest but the price is absolutely ludicrous and I cannot find any other information on this by anyone who has tried this before, perhaps with Inspire or P3 batteries maybe...
For navigation, I've been unable to achieve better than 0.33m horizontal, and vertical, average. That is because of wind, and other physics of the drone, not the navigation system (from what I can tell)
I would love to hear if you have been able to verify geographic coordinate accuracy and precision
My focus has been mostly on navigation accuracy, which requires changing your waypoing_nav_rad parameter to less than 1 meter. Doing that causes the drone to take much longer to 'find' the waypoint, but it does work. Preliminary tests show ~15 cm RMSE (horizontal and vertical combined), when the waypoint navigation radius is set to 10 cm. Setting the radius to less than 10cm always causes the drone to spend too much time trying to hit the waypoint. For post-processing assessment of station accuracy (not for navigation) we are able to achieve a 2.6cm RMSE.
have you upgraded here+ to ublox 1.30 firmware? should be more easy to get RTK FIX mode in an open sky like this
Rtk float is not rtk. Float is equivalent to no solution. You need to transmit a rtk correction to the controller in order to get fixed
Thank you, you're absolutely correct.
really nice!!!
could you tell use the parts that we need to make the same setup as yours?
Currently have a tarot x8 with everything, just need advice with the RTK.... where can I get the usb rtk hub and the antenna? because as I see here, all I need is the pixhawk 2.1 and the here+ RTK... the module that you connect to the computer, that's where I have no Idea about.
Thanks and good job!!!
hi, really good, please what is your telemetry radio ?
I don't get it... what are you measuring? You didn't even mark the home point with anything as far as I can tell. Looking at the flight logs doesn't will only tell you how the aircraft thinks it flew, not what it really did...
iforce2d Ahh, you are right that I didn't mark the home point. I used the mission planner software to survey in the point so that I could repeat waypoints based on the relative, not absolute, position, of the points. I conducted multiple flights during the same day in order to test this, with no intent of coming back on another day. For what it is worth, the s1000 and probably any drone on the market right now, is unable to repeat the waypoints at the same positional precision as the HERE+ rtk. This is most likely do to wind conditions, motor balance, and many other factors. This was my very first flight with this system, and since then I've done work using geodetic benchmarks and supplemented with rtk based survey points in imagery. Even with a 2cm absolute positional accuracy, repeatability of waypoints are typically 0.33m horizontally and vertically at best. Hopefully now you get it :)
You actually can see what the aircraft really did. If you wanted to track its actual flight path, you can look for the .kml files and examine those.
Ah, sorry for the late response. We were not measuring absolute positional accuracy. We are interested in measuring the repeatability of navigation to specific waypoints, meaning the accuracy of the coordinates of the home point are less relevant, as long as they don't change between flights.
Which telemetry link are you using?
Would like to slightly extend arms, upgrade motor/esc/prop combo, change battery system to a quad setup with quick change and hotswap abilities -2 above and 2 below and each pair on thier own quick release bracket, Print a rain cover, to fill top board gaps and seal sides & waterproof electronics.
And swap to Cube 2.1 & Here + with Connex TX and Audrino components for additional features.
Just need a gantry CNC and I’m away.
The expandable battery banks that the Matrice batteries clip into and the bank slides into a housing raises my interest but the price is absolutely ludicrous and I cannot find any other information on this by anyone who has tried this before, perhaps with Inspire or P3 batteries maybe...
Why the connex instead of something like the Herelink?
dear sir , can you tell me the position accuracy of it ?
For navigation, I've been unable to achieve better than 0.33m horizontal, and vertical, average. That is because of wind, and other physics of the drone, not the navigation system (from what I can tell)
Followup video??