Meeting notes devcall 7 August 2019
Component owners should flag what they want to discuss before the call.
Next calls should be quicker on updates on Issues/PRs, and should spend more time on high level discussions.
Roadmap needs regular updating.
Daniel is finalizing the list.
Project boards need to be on PX4 top level to be visible.
- Improve magnetometer consistency check
High priority and should be done for 1.10
- Block auto-disarm for catapult hand launched Fixed Wing
Roman should look at this.
- Add generic VTOL tailsitter airframe
Daniel has done work on this.
Roman will do a final review and add it.
- Implement Module params multicopter landdetector class + #12619
Roman does an assessment and gets it in.
- GTest functional test
Julian is requesting for last feedback before bringing it in. Lorenz will look at it.
- Command / safety
- Disable datalink loss reactions for SITL. Reject PR. Bring Daniel’s solution in.
- Communication architecture
- Computer vision
-Collision prevention has been merged today.
-Link to ground station: distance data is send. Gus is needed: https://github.com/PX4/avoidance/issues/460
-VIO needs rework. Julian will discuss with Paul. Needs to be added to project board.
- Fixed wing
No progress. Needs input from fixed wing stakeholders to set priorities.
1khz rate controller. Is planned for 1.10. Mathieu will discuss offline with Daniel.
Mathieu asks for input on priorities
12505: Mathieu needs tests to check last changes. Tony will do them.
acceleration stick input: important.
Discuss when Matthias on the call
- User experience
Issues/PRs discussed above.
12630 + 12455 + 12503 + 12464 have priority and are in the right order on the project board.
Silvan is working on those.
All the work required to achieve two main objectives has already been done: 1) Move more tasks to work queues and 2) move the main sensors to DMA. Some final touches and design decisions are required to bring this in and enable this. David, Daniel and Lorenz will discuss this offline and create a shortlist of PRs that generate the main value without introducing massive churn in the codebase. This will likely moving IMUs to SPI DMA (to free CPU load) and moving 1-2 controllers to work queues (which will free RAM) and getting rid of file descriptors in uORB (freeing more RAM). All of this combined will bring CPU load and RAM down to a level where 1.10, with all the additional features and improvements, will run fine on an F4.