PX4 Dev Call: June 10, 2020

June 10, 2020

Agenda

  • Status update by component/project
  • Roadmap, and Release discussion
  • Community Q&A
  • In-Depth discussions

Join Meeting

Meeting ID: 946 175 205
Join using your mobile/desktop

Call in using your phone, find your local number: https://zoom.us/u/aetSYKbiMF


Component update

There was a drop in public slack channel activity around May which we can’t directly explain. Lockdown, holidays, technical problem?

Flight Testing

@rroche Will update on Slack where the testing partner selection process is standing at the moment. Currently, the developer summit is first priority.

Failure injection possible in simulation but also potentially for real world by @JulianOes. It’s for now implemented for simulation but it would be awsome to be able to use the exact same mechanism on the real vehicle. There still needs to be discussion around how to do that architecturally.

System Architecture

OS / NuttX

  • For STM32H7 BDMA was added
  • There’s some H7 issues with the GPIO slew rate

Driver

New driver for Bosch BMI88 by @dagar

Some fixes for magnetometers that are present in Invensense chips also when they are used externally over I2C.

Sensor external I2C probe

Negative current on INA226 was interpreted as error code:

Commander

@LorenzMeier What’s the plan for the next step of Commander refactoring.
There’s a bigger plan for Navigator/state machine architecture that was broken down into smaller pieces and is very slowly executed.

Estimation

Terrain estimation fix by @bresch for the selection between using a range sensor or optical flow to estimate the distance to ground. Including unit tests to make sure it doesn’t break.

@dagar got the basics for Multi-EKF support working on master.

VTOL

@JulianOes found the failsafe reaction to losing GPS in fixed-wing mode is a blind loiter towards the ground with minimal airspeed.

The future approach should be to use TECS to slowly reduce the altitude instead of descending with a fixed throttle and attitude. We need an improved architecture to allow that. @sfuhrer would be up to noting down requirements for the state machine such that we have an idea of the requirements.

@sfuhrer Is proposing to clamp the displayed airspeed sensor in the UI because most airspeed sensors have a lower limit for the minimum measured speed and might display a slightly negative airspeed instead of 0. We need to make sure components get the real airspeed measurements while the end-user should only see strong negative airspeeds in the UI and not small negative noise. @LorenzMeier suggests to not crop the published airspeed but only filter it before displaying. There should be a flag indicating the health of the sensor and the

Fixed Wing

No updates

Multicopter

Issue flying altitude mode in the simulation was fixed.
@MaEtUgR Follow up action item to change the checks in the filter.

MAVSDK

MAVROS / RTPS / ROS2

It was proposed a live call to the summit regarding micro-ROS and the possible migration to it in PX4.
:point_right: More on micro-ROS:


@LorenzMeier We have to start using it and identify our needs to make sure it’s serving us well before we spend too much effort upfront.

@dagar Who is using ROS2?
Academia has some traction some bigger companies started to use it. There’s some light shows realized with it.

Avoidance

No updates

Simulation

@Jaeyoung-Lim realized the Typhoon mesh, model and its gimbal were all rotated by 180°. That was inverted to normal.

SITL tests tailsitter fails for different reasons. This was not yet looked into.

UAVCAN

DS-015 spec still under discussion. All feedback is welcomed:

Roadmap, and Release discussion

@dagar We should cut another beta.
Blocking issues:

  • Oneshot has some problems
  • Multicopter land detector discussion afterwards

Developer summit

@rroche There were a lot of registrations and content proposal submissions. Thanks to everyone participating and we’re looking forward to meeting everyone virtually. There will be more updates on the schedule.

Community Q&A

Kai Gottschalk was testing fast descend speed and the drone slowed down too late. Analyzing the log it looks like the drone slowed down to land speed but with too smoothed out setpoint generation and hence too late. Kai will create an issue and @jkflying will have a first look if there’s a simple configuration check fix.
Log: https://logs.px4.io/plot_app?log=f4db817d-90fa-419c-80e4-a32bf9067817

In-Depth discussions

  • Multicopter land detector
    @dagar would like to discuss an issue for the release:

related:


We’ll give @Roman’s suggestion to require low thrust AND the velocity setpoint comparison for land detection and see if there’s negative effects on all other vehicles.


Errata and Feedback

Let me know below if I failed to capture anything the right way, and if there are any updates, or you have feedback on the call format.