Support reset_counter of VISION_POSITION_ESTIMATE message

[michael-p]

Describe problem solved by the proposed feature
Mavlink message VISION_POSITION_ESTIMATE has a field reset_counter (Mavlink documentation) which is very useful if the position estimates are generated by a SLAM system (that is, not just visual odometry) but it looks like this is field currently ignored in the PX4 firmware.

Describe your preferred solution
Pass reset_counter through to the ekf and make sure that large position jumps from the vision system do not cause any problems there.

[TSC21]

This is a quite new field - it's not that is ignored, it's just not handled yet.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[TSC21]

Still relevant.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. Thank you for your contributions.

[james-modalai]

Bringing this back from the dead. We are trying to get relocalization via loop-closure and fiducial markers working and sure enough I find PX4 ignores this field after getting everything else working. Any pointers would be appreciated as I evaluate what it would take to implement this.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. Thank you for your contributions.

[Prototyx]

Could be very usefull if it was supported for SLAM system.
Is it in the roadmap ?

[stale]

This issue has been automatically marked as stale because it has not had recent activity. Thank you for your contributions.

[michael-p]

@james-modalai Did you get to any conclusions regarding this issues and possible implementation? I just bumped into this issue in a different project again... :)

[james-modalai]

We've just been giving px4 continuous inputs and keeping track of offsets offboard. This will be a problem when swapping back and forth from GPS to VIO though.

…

On Thu, Apr 15, 2021, 07:32 Michael Partheil ***@***.***> wrote: @james-modalai <https://github.com/james-modalai> Did you get to any conclusions regarding this issues and possible implementation? I just bumped into this issue in a different project again... :) — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#11970 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AOLRZJXCUKYHB5IQZX5O7XDTI32JFANCNFSM4HLAMXWQ> .

[dagar]

If you can start populating the VISION_POSITION_ESTIMATE reset_counter we can look at an initial implementation PX4 side. We've already done the work in the controllers to handle the delta in different estimates (position, velocity, attitude), it would be great to handle this cleanly across the entire system.

[james-modalai]

Thank's Daniel! we've been populating the reset counter since we first got VIO going with PX4, I've just avoided doing any resets or relocalization jumps mid-flight. I'll give it a go with px4 master branch soon.

…

On Thu, Apr 15, 2021 at 10:18 AM Daniel Agar ***@***.***> wrote: If you can start populating the VISION_POSITION_ESTIMATE reset_counter we can look at an initial implementation PX4 side. We've already done the work in the controllers to handle the delta in different estimates (position, velocity, attitude), it would be great to handle this cleanly across the entire system. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#11970 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AOLRZJXL5U7X72ZFKF3SBCTTI4NWJANCNFSM4HLAMXWQ> .

[bresch]

The reset counter is now supported; the EKF resets its states when received:

PX4-Autopilot/src/modules/ekf2/EKF/ev_control.cpp

Line 50 in 444e5d2

bool ev_reset = (ev_sample.reset_counter != _ev_sample_prev.reset_counter);