ros-drivers / microstrain_mips

Other
26 stars 111 forks source link

GX5-25 KF IMU publishing #23

Closed samkys closed 4 years ago

samkys commented 5 years ago

This PR allows the user to set and publish the filtered IMU data from the GX5-25. Currently the only filtered data the original driver output was the CF data for the -25 and the Nav data from the GPS enabled devices. I wanted access to the KF data from the -25 since we can get the accelerations with gravity removed as well as the KF orientation covariance values. This allows dynamic covariance publishing rather than the static ones that were previously set. You can still use the static values if you publish just the standard IMU data.

samkys commented 5 years ago

Don't merge this quite yet. There is something funny that I don't like about how the filtered data implementation works on the -25 model. Just realized that if you want to publish filtered data, it still relies on the AHRS imu setup rate. So, if you want really high data rates you will need to choose between either KF IMU data, CF data, or crank up the baud rate.

samkys commented 5 years ago

Few fixes should fix my last comment. Will be testing with it today and report back.

samkys commented 5 years ago

Got a chance to try it out. It works better now but there still is one more thing I want to try and fix.

samkys commented 5 years ago

@bsb808 @tonybaltovski Should be good to go now. I tested it with a -25 and we are able to get the filtered IMU data out. It is setup so you can put it into the old setup, i.e. no filtered IMU output or put it into the new mode with filtered IMU output, or both on or off.

The benefit to this PR is that with the filtered IMU output the IMU removes the gravity vector for us as well as compensates for angular and acceleration bias. With the filtered data you are not required to remove gravity as it's just a parameter in the launch file but it is helpful to prior to feeding it into an EKF, say robot_localization.

samkys commented 5 years ago

Any feedback?

tonybaltovski commented 4 years ago

Just one last nit before we can merge this

samkys commented 4 years ago

@tonybaltovski How does that look?

tonybaltovski commented 4 years ago

Thanks @samkys