Weekly Dev Call of June 28, 2017


  • First half of call: Roadmap (30 mins)
  • Second half of call: Driver system architecture (led by David Sidrane)

Yet another pain point is feedback during errors. Why won’t the system arm or let me enter that mode? Why did navigator position timeout if my GPS never lost a fix (EKF issue), etc. If something goes wrong we need to be extremely clear why and what can be done about it.

Longer term perhaps more of this could be exposed. Why does the GCS let me try to arm or request a mode if it’s just going to be rejected?

1 Like
  • Linux port should make use of the native kernel scheduler so it benefits of multiple cores.
  • Remove duplication of drivers by factoring out common functions but without introducing performance penalty
    ** we have a driver for NuttX which has better error checking and one in DriverFramework which is lacking (see NuttX MPU9250 vs DF MPU9250)
  • Get rid of DriverFramework? Only complicates things to have a submodule and I believe there are no users outside of PX4

Possible roadmap idea: Airframe cleanup