Question about using RTK about mrs_uav_system HOT 10 CLOSED

Hello, so we use the RTK a bit different, in all our applications, we have the radio receiver for RTK corrections connected straight to the RTK board, and the RTK board then sends the RTK position to the main computer, avoiding the Pixhawk completely. Our odometry node then subscribes the RTK messages (topic rtk_gps/global), which are used in the estimator.

We have not tried pushing the corrections through pixhawk, or getting the RTK position back from pixhawk, but it is definetly possible.

Dan H.

from mrs_uav_system.

Hi @DanHert,
thanks for the clarification. So basically, I could have any source of a RTK solution (own processing, Pixhawk...) and just need to assure, that I publish this solution to rtk_gps/global. the mrs_uav_odometry package will accept it as valid input then?

Best regards,

Max

from mrs_uav_system.

Yes, that is correct.

Cheers, Dan H.

from mrs_uav_system.

I now successfully set up a ublox F9P with a NTRIP correction service (source). According your last reply, I am now planing to write a simple publisher like void RtkRepublisher::callbackTersus(const mrs_msgs::BestposConstPtr &msg) to publish the mrs_msgs/RtkGps.msg. As I understand it based on the RtkRepublisher::callbackTersus, I don't need to provide the twist and pose within my RTK solution (which I don't have as it is a RTK only solution...). So I am wondering, what the original purpose of these twist/pose message parts you are using in
void RtkRepublisher::timerMain(const ros::TimerEvent &event) is? Is it only when "simulating" RTK in Gazebo or does it have any further use in mrs_odometry?

Cheers,

Max

from mrs_uav_system.

Hi, you do not need to provide the twist and pose, lat and lon is enough. The twist and pose is intended to be used with UTM coordinates, so if you for some reason have the UTM instead of lat and lon, just fill out the frame id as UTM and use the pose and twist (some GPS/RTK receivers will give you position and speed, so that is the reason for the twist, also the quaternion in pose is not used at all).

And yes, the timerMain is only for simulation, it can also do random jumps in the RTK position (this can happen with the real RTK if you change from RTK Float to RTK Fix for example). So it is for testing failures of the RTK.

Dan H.

from mrs_uav_system.

ok.
Thank you very much for the help and clarification!

from mrs_uav_system.

Hello,
I integrated RTK within ROS and MRS with my custom publisher to /uav1/rtk_gps/global.
For actually using it I still have one question:
Where are the lever arms defined in MRS? (@DanHert you mentioned the Z-Offset in #43 (comment))
Or do I need to correct for the lever arms befor sending the positions to the MRS odometry fusion?

And just for clarification: I can use the RTK state estimation by just setting the lateral_estimator to RTK right?

Thanks for any help in advance!

from mrs_uav_system.

Hey, before flying, pull the uav_core. We have had a bad remapping of the RTK topic into the rtk_republisher node's launch.

from mrs_uav_system.

Hi,
I will do, but I don't understand why the rtk_republisher is relevant here for me? I am not using the Tersus RTK. I am using a ublox F9P and publishing its solution already to rtk_gps/global. My question concerns how to apply the lever arm of the RTK. Is it done inside MRS or should I correct it before publishing it to MRS.

from mrs_uav_system.

Hi, yeah you do not need the rtk_republisher. The lever arm is not handled at all by the system, so if you want to be as precise as possible, you need to correct it yourself.

Dan H.

from mrs_uav_system.