PX4 Vision Estimation: Questions on Integration with Pixhawk

@Andrew_at_Airlitix Okay thanks, that is much more clear.

So I guess if you have connection with the pixhawk with the FTDI, then you should stay with that becuase that means the connection part is working. → Just to be clear, you no longer need to debug hardware

There are a number of different reasons that the vision data might not reach or become read from the px4 side, but this you need to figure out.

As far as I know, there has been some changes on both the firmware and mavros side for the EV pipeline. So it might be better to try with both on the master version, or figure out which release works with which.

I am confused of this evaluation. How did you come to this conclusion?

Are you sure about this? I would love to be able to move forward with the Orbitty and assume there is no intervening hardware issue, but I’m not sure how to proceed assuming it’s all software at this point. I know there have been changes with the framework, and right now we are using the master version of the firmware (recent, up to about a week or two ago).

Maybe you’re right, and that’s a premature conclusion. What I do know is even with the setup where we are receiving IMU data, there is nothing being published on the FCU side related to the vision messages we are sending (i.e. running listener vehicle_visual_odometry results in a “never published” response). Is there a more straightforward way to check if MAVROS messages published on the CC are reaching the FCU (maybe some other channel I can test)?

@Andrew_at_Airlitix If you can get the imu_data messages through MAVROS, it means that your communication pipeline is in a good state. If you want to check the quality of your link, you can check the data rates to see if there are any losses but there is a low probability that this is unidirectional.

You also need to check the version of mavros

An easy way to check if ‘any’ messages are reaching the flight controller, you can start with sending services / topics which may trigger a mode change or arming commands to the flight controller.

The fact that you don’t have vehicle_visual_odometry says never published doesn’t confirm anything on your setup. It simply means its not reaching the firmware correctly.

Hi all,

Sorry for the relative silence, but I wanted to let you know I’ve resolved the issues. Ultimately, the hardware configuration was fine, and the problem was improper timestamp information on the vision messages sent over MavROS.

Thank you everyone for your help on this.

1 Like

Hello @Andrew_at_Airlitix thank you for reporting this info. Could you be more specific on the timestamp information on the vision messages? I am having very similar problems.