January 13, 2021
- Community Q&A
- Component Discussions
- Roadmap, and Release discussion
META: Changes to the PX4 Dev Call in 2021
Community, we are trying a new format for the developer call, I hope the new agenda works for everyone, please let us know in the comments below (or slack!)
@Dino_Mani: Secure bootloader. All related people are in the call. We need to get this rolling again. Jukka: Continuing the work from Dino to sign PX4 images and make sure the bootloader will only accept correctly signed software. If you make a product based on an STM core you can burn the key to be checked such that there’s technically no way to updated with unsigned software anymore.
What has to be done to get the bootloader change in and what are the ramifications. If the bootloader is built as standard (unsigned) the memory map is unchanged and the application address stays the same. So it’s a matter of testing but otherwise, nothing is blocking.
- Documentation with a visual/table of the memory map would be awesome. On the dev guide https://docs.px4.io/master/en/
- The build needs to be added to CI such that we’re sure it doesn’t break.
- Note by Igor: It does not fit on F4 boards but works on F7 and H7 boards. There should be a warning to the user that F4 boards are not supported.
- Plan a follow-up such that people have time to review and the next discussion is planned by @Dino_Mani tomorrow.
wr: Questions about ROS2 and FastRTPS. Trying to connect a Raspberry Pi with a pixhawk and has problems. Will be taken offline with @TSC21.
Benjamin: Question about ROS2 FastRTPS use on a specific setup/version that uses a different DDS implementation. It should be working according to the standard and not be an issue.
Secure bootloader was discussed.
Automatic vibration filtering
There is a minimal viable uORB to UAVCAN pr that requires further testing but should otherwise be ready to merge and not break anything.
High Priority Queue
OS / NuttX
David has a pr to fix a broken STM32F412 configuration on upstream NuttX. Also, there was quite some churn upstream which will create a testing effort
MAVROS / DDS / ROS2
- When the agent and the client have different topics defined to be parsed, this results in one or both sides reporting unknown topic IDs - the source of the problem is now reported on the console warning when this happens:
- A bug on the uORB to ROS msg converter was generating capitalized field names on the ROS messages, which in turn resulted on
px4_msgsfailing to build. This is now solved:
- PR to restructure the offboard control example using ROS 2, since the initial example wasn’t self-contained and was using velocity control (which means we had to close the control loop on the node = increased complexity for the example):
px4_ros_commoved from Travis to Github Actions:
MAVROS patch release 1.5.1 happened on January 4 - changelog
- Minor bug fixes on MAVLink message struct initializations and on frame conversions.
Errata and Feedback
Let me know below if we failed to capture anything the right way, and if there are any updates (not present here), or you have feedback you would like to share with the dev-team.