PX4 Sync / Q&A: Feb 19, 2025

farhang · February 18, 2025, 9:16pm

PX4 Sync / Q&A: Feb 19, 2025
Dronecode Calendar

Agenda

Announcements
Future Events
Flight Testing Update
Release Discussion
Bug Report / Q&A

Announcements

The first Dronecode meetup in South California
RSVP: South California Meetup #1 - February 2025 - San Diego, Wed, Feb 19, 2025, 3:00 PM | Meetup

Future Events

OSS NA 2025 - Denver CO - June 23-25
- CFP Closes February 17th
ICUAS May 14-17, Charlotte, USA

Flight Testing Update

We are running a beta program with Ascend Engineering (Chicago). They will be running flight testing for the community as part of an arrangement with Dronecode

How to reach out to the team

GitHub: Tag user @PX4/testflights
Discord: #flight-testing
Weekly on the PX4 Dev Call

How to Request Testing (GitHub)

Write down the steps to test your issue/pull request
- Make sure to note the risk involved in flying
- Write down things to look out for / anticipate - eg: “we are looking for no yaw jumps”
Add issue/pull request to the Flight Testing project board
Make sure to specify any hardware/software requirements as much as possible

Flight Testing board:

GitHub

Flight Testing • PX4

Flight Testing

Discussion

Comments:

Broader range of boards to be tested
More specific features

Release Discussion

Project Board

For release coordination, we will be using the project board.

Project Board Link

v1.16 Release

Discussion

Patch release 1.15.4 : Release v1.15.4 · PX4/PX4-Autopilot · GitHub

Bug report / Q&A

1- CI testing PX4 Internal and External (ROS2) code · Issue #24363 · PX4/PX4-Autopilot · GitHub

A CI is needed, @farhang @rroche will work on that.

2- @stevejohn163 I’m trying to initiate a takeoff using the MAV_CMD_NAV_TAKEOFF command via COMMAND_INT. The problem is that when I set the altitude with different MAV_FRAME values, the takeoff altitude doesn’t change as expected. No matter which MAV_FRAME I use and set the takeoff altitude to 2.5m, it always tries to take off to 2.5m. However, since I’m already at an altitude of 200m above mean sea level, it throws the error: “Already higher than takeoff altitude.”

github.com/PX4/PX4-Autopilot

src/modules/navigator/takeoff.cpp

04a3c4af2


      
          				 "Using default takeoff altitude: %.1f m\t", (double)_navigator->get_param_mis_takeoff_alt());
          
          		events::send<float>(events::ID("navigator_takeoff_default_alt"), {events::Log::Info, events::LogInternal::Info},
          				    "Using default takeoff altitude: {1:.2m}",
          				    _navigator->get_param_mis_takeoff_alt());
          	}
          
          	if (takeoff_altitude_amsl < _navigator->get_global_position()->alt) {
          		// If the suggestion is lower than our current alt, let's not go down.
          		takeoff_altitude_amsl = _navigator->get_global_position()->alt;
          		mavlink_log_critical(_navigator->get_mavlink_log_pub(), "Already higher than takeoff altitude\t");
          		events::send(events::ID("navigator_takeoff_already_higher"), {events::Log::Error, events::LogInternal::Info},
          			     "Already higher than takeoff altitude (not descending)");
          	}
          
          	// set current mission item to takeoff
          	set_takeoff_item(&_mission_item, takeoff_altitude_amsl);
          	_navigator->get_mission_result()->finished = false;
          	_navigator->set_mission_result_updated();
          	reset_mission_item_reached();

Currently, I have to manually retrieve the MSL (mean sea level) altitude, add 2.5m to it, and send that value via COMMAND_INT for it to work properly.

3- PX4 autopilot ROS2 user guide problem

Maybe we can make targets more specific?

github.com/PX4/PX4-Autopilot

[Bug] No rule to make target

opened 04:32PM - 19 Feb 25 UTC

mrpollo

bug-report

### Describe the bug Users trying to build SITL are facing issues, I personally… ran into this when I upgraded my development environment. I'm creating this issue to track a resolution to this. **Problem:** Building PX4 SITL fails if you don't have the required Gazebo requirements installed in your system ``` make px4_sitl gz_x500 gmake[1]: *** No rule to make target 'gz_x500'. Stop. make: *** [Makefile:232: px4_sitl] Error 2 ``` I brought this up on the weekly [PX4 Dev Call for February 19th](https://discuss.px4.io/t/px4-sync-q-a-feb-19-2025/43827), and we came to an agreement on how to fix this for users **Proposed Solution:** The build is intended to fail by design if you don't have requirements installed, but we can try to fallback to another target within cmake that outputs a helpful message indicating which requirements are missing.

github.com/PX4/PX4-Autopilot

(new) GZ Feature tracker

opened 07:24PM - 22 Aug 24 UTC

Jaeyoung-Lim

feature-request Sim: gazebo

### Describe problem solved by the proposed feature This is a feature tracker t…o achieve feature parity with gazebo classic. Feed free to modify the issue to append the list or mark it as completed. If there is an issue/PR, please reference it. ### Describe your preferred solution The list of features that are missing in the new GZ is as follows: **Models** - [x] Fixedwing-advanced lift drag (https://github.com/PX4/PX4-Autopilot/pull/22167) - [x] Tiltrotor VTOL (https://github.com/PX4/PX4-Autopilot/pull/24028, https://github.com/PX4/PX4-gazebo-models/pull/66) - [x] Differential rover (https://github.com/PX4/PX4-Autopilot/pull/22402) - [x] Ackermann Rover (https://github.com/PX4/PX4-Autopilot/pull/23383) - [x] Quadtailsitter (https://github.com/PX4/PX4-gazebo-models/pull/65, https://github.com/PX4/PX4-Autopilot/pull/23943) - [x] Omnicopter (https://github.com/PX4/PX4-Autopilot/pull/21818) - [ ] USVs (Boats) (ongoing in https://github.com/PX4/PX4-Autopilot/pull/24310) **Worlds** - [ ] Ocean - [x] Baylands **PX4 related Features** - [x] Gimbal Control (https://github.com/PX4/PX4-Autopilot/pull/23382) - [ ] Camera Manager / Video Streaming - [ ] Camera triggering/geotagging - [ ] Rotor failures - [x] Collision prevention (https://github.com/PX4/PX4-Autopilot/pull/23507) - [ ] Catapult launch - [ ] Parachute/Flight termination - [ ] Retractable landing gear - [x] Precision Landing (https://github.com/PX4/PX4-Autopilot/pull/23450) **Sensors** - [x] Airspeed sensors (https://github.com/PX4/PX4-gazebo-models/pull/65) - [ ] Optical Flow sensor - [ ] GPS yaw heading simulation - [ ] Multi airspeed sensor - [x] Distance sensor (https://github.com/PX4/PX4-gazebo-models/pull/62) (https://github.com/PX4/PX4-Autopilot/pull/23879) **Actuators** - [ ] Wind influence for motor thrust - [ ] Wind influence for lift and drag plugin **Simulation Features** - [ ] Faster than real-time simulation https://github.com/PX4/PX4-Autopilot/pull/23783 - [x] Vehicle view tracking (https://github.com/PX4/PX4-Autopilot/pull/22808) - [ ] MAVSDK SITL tests - [ ] Multivehicle simulation - [ ] Wind/Gust simulation ### Describe possible alternatives NA ### Additional context

PRs needing review

github.com/PX4/PX4-Autopilot

Commander: fix parachute trigger

main ← pr-fix_parachute

opened 10:54AM - 17 Feb 25 UTC

bresch

+1 -2

### Solved Problem Setting `lockdown` disables the actuators. In this mode, `fo…rce_failsafe` has no effect as the actuators are disabled, so the parachute is not getting released as it requires the output to change to its failsafe value. ### Solution Only set the `force_failsafe` (not `lockdown`) flag when in flight termination ### Alternatives We could change the priority in https://github.com/PX4/PX4-Autopilot/blob/01549a58326c47cf562dc190f7fa6eaf966516cc/src/lib/mixer_module/mixer_module.cpp#L477-L489 but I feel that this could be dangerous in HIL mode where switching to flight termination could trigger a physical parachute. ### Test coverage Tested on a Pixhawk4 with parachute configured on channel 4 with a failsafe value of 1234. The autopilot is then armed and tilted by 90 degrees to trigger the attitude failure detection. main: ``` actuator_armed timestamp: 32799326 (0.023821 seconds ago) armed: True prearmed: False ready_to_arm: True lockdown: True manual_lockdown: False force_failsafe: True in_esc_calibration_mode: False nsh> pwm_out status pwm_out: cycle: 25464 events, 692035us elapsed, 27.18us avg, min 24us max 54us 2.561us rms pwm_out: interval: 25464 events, 1472.62us avg, min 1131us max 50061us 3298.229us rms INFO [mixer_module] Param prefix: PWM_AUX control latency: 25464 events, 353165400us elapsed, 13869.20us avg, min 291us max 5852269us 230088.969us rms INFO [mixer_module] Switched to rate_ctrl work queue Channel Configuration: Channel 0: func: 101, value: 1000, failsafe: 1000, disarmed: 1000, min: 1100, max: 1900 Channel 1: func: 102, value: 1000, failsafe: 1000, disarmed: 1000, min: 1100, max: 1900 Channel 2: func: 103, value: 1000, failsafe: 1000, disarmed: 1000, min: 1100, max: 1900 Channel 3: func: 104, value: 1000, failsafe: 1000, disarmed: 1000, min: 1100, max: 1900 Channel 4: func: 401, value: **1000**, failsafe: **1234**, disarmed: 1000, min: 1000, max: 2000 <==== ``` This PR: ``` nsh> pwm_out status pwm_out: cycle: 25464 events, 692035us elapsed, 27.18us avg, min 24us max 54us 2.561us rms pwm_out: interval: 25464 events, 1472.62us avg, min 1131us max 50061us 3298.229us rms INFO [mixer_module] Param prefix: PWM_AUX control latency: 25464 events, 353165400us elapsed, 13869.20us avg, min 291us max 5852269us 230088.969us rms INFO [mixer_module] Switched to rate_ctrl work queue Channel Configuration: Channel 0: func: 101, value: 1000, failsafe: 1000, disarmed: 1000, min: 1100, max: 1900 Channel 1: func: 102, value: 1000, failsafe: 1000, disarmed: 1000, min: 1100, max: 1900 Channel 2: func: 103, value: 1000, failsafe: 1000, disarmed: 1000, min: 1100, max: 1900 Channel 3: func: 104, value: 1000, failsafe: 1000, disarmed: 1000, min: 1100, max: 1900 Channel 4: func: 401, value: **1234**, failsafe: **1234**, disarmed: 1000, min: 1000, max: 2000 <==== ```

github.com/PX4/PX4-Autopilot

Introduce boat vehicle type

main ← pr_boat_decoupling

opened 10:17AM - 10 Feb 25 UTC

TomasTwardzik

+77 -7

# Motivation Decoupling boat and rover to exists as separate entities within Na…vigator and Commander modules. Despite similarities, this should lead to more legible, self documented code. In the future, this also allows to make changes separately, without risking introducing unwanted behavior to other platforms. ## Discovered Problem Rover can be armed with high throttle, which is a significant risk. ## Solution Both Rover and Boat now require throttle stick near center to be armed. ## Changelog Entry For release notes: `VehickeStatus.msg` -> new vehicle type VEHICLE_TYPE_BOAT `Navigator module` -> boat explicitely mimics rover behavior, instead of relying on vehicle_type being VEHICLE_TYPE_ROVER Control Allocator's `module.yaml` -> Adding new airframe types for boats (Singe, Twin engine and Differential) `commander_helper` -> adding distinct separation for boat and rover instead of generic ground vehicle `Commander` -> adding method for throttle near center detection, adding, adding mission override for boats, adding dedicated arm checks for rover and boats

github.com/PX4/PX4-Autopilot

Neural Control Mode

main ← SindreMHegre:neural_control_with_tflm

opened 12:23PM - 19 Feb 25 UTC

SindreMHegre

+1059 -0

### Solved Problem Using neural networks for various tasks on UAVs in research …is becoming more and more commonplace. My aim is to provide a module that can be used to replace the MC controller with NN to lower the barrier for using RL and NN on UAVs in general, and give researchers and hobbyist a place to start for NN use. ### Solution - Add TensorFlow Lite Micro as an inference lib into PX4 - Make custom board configs with neural controllers - Make a module for NN control ### Changelog Entry For release notes: ``` Feature/Bugfix XYZ New parameter: XYZ_Z Documentation: Need to clarify page ... / done, read docs.px4.io/... ``` ### Alternatives We could also use Eigen or executorch ### Test coverage - Built and tested SITL - Passed all PR cheks when commited on v1.15.3 ### Context I have a lot of problems when building for embedded FCUs, like pixracer pro and pixhawk 6C. This is because TFLM needs the standard c++ 17 libraries. Tried a lot of different hacky solutions, and run into problems like: - Double definitions of math functions like acos: src/lib/tflite_micro/tflite_micro/tensorflow/lite/micro/tools/make/downloads/gcc_embedded/arm-none-eabi/include/c++/13.2.1/cmath:85:9: error: 'double acos(double)' conflicts with a previous declaration 85 | using ::acos; - Missing C++ std header files like <algorithm> /src/lib/tflm/tflite_micro/tensorflow/lite/micro/tools/make/downloads/flatbuffers/include/flatbuffers/flatbuffers.h:20:10: fatal error: algorithm: No such file or directory 20 | #include <algorithm> - Missing definitions: src/lib/tflm/tflite_micro/tensorflow/lite/micro/tools/make/downloads/gcc_embedded/arm-none-eabi/include/c++/13.2.1/arm-none-eabi/bits/c++config.h:545:46: error: missing binary operator before token "(" 545 | #elif __cplusplus >= 201103L && __has_builtin(__builtin_is_constant_evaluated) | ^ compilation terminated due to -Wfatal-errors.

CMake needs to be fixed first to have it tested. (preferrably Nuttx and PX4 build separated to avoid build errors)
Eigen is not used on PX4 side much (has a similar lib however) so maybe using Eigen is not going to be an idea.

github.com/PX4/PX4-Autopilot

Faster than Real -Time support in GZ

main ← dirksavage88:set_rtf

opened 02:25PM - 07 Oct 24 UTC

dirksavage88

+41 -0

### Solved Problem The new GZ simulation real time factor is configurable, howe…ver the service request has not been implemented in the GZ bridge. partially fixes Fixes https://github.com/PX4/PX4-Autopilot/issues/23602 ### Solution - Configure the GZ bridge to send a service request with the simulation speedup value environment variable "PX4_SIM_SPEED_FACTOR" ### Changelog Entry For release notes: ```Implement configurable real time factor in GZ sim (PX4_SIM_SPEED_FACTOR) Feature: Configurable real time factor in new GZ sim New parameter: None Documentation: Configurable via environment variable "PX4_SIM_SPEED_FACTOR" ``` ### Alternatives Change the real time facto in gazebo world sdf manually, or make a world file specific to an increased RTF ### Test coverage - Tested in GZ sim [rtf.webm](https://github.com/user-attachments/assets/7010c3d6-5d51-437b-bace-0a8eb4626040) ### Context Related links, screenshot before/after, video

Sindre · February 19, 2025, 12:28pm

Hi, I’m trying to add neural network support for px4 to lower the barrier for using RL on UAVs in research. I have some make problems, and was wondering if someone could help me. I will be attending the sync call.

github.com/PX4/PX4-Autopilot

Neural Control Mode

main ← SindreMHegre:neural_control_with_tflm

opened 12:23PM - 19 Feb 25 UTC

SindreMHegre

+1060 -0

### Solved Problem Using neural networks for various tasks on UAVs in research …is becoming more and more commonplace. My aim is to provide a module that can be used to replace the MC controller with NN to lower the barrier for using RL and NN on UAVs in general, and give researchers and hobbyist a place to start for NN use. ### Solution - Add TensorFlow Lite Micro as an inference lib into PX4 - Make custom board configs with neural controllers - Make a module for NN control ### Changelog Entry For release notes: ``` Feature/Bugfix XYZ New parameter: XYZ_Z Documentation: Need to clarify page ... / done, read docs.px4.io/... ``` ### Alternatives We could also use Eigen or executorch ### Test coverage - Built and tested SITL - Passed all PR cheks when commited on v1.15.3 ### Context Related links, screenshot before/after, video

dirksavage88 · February 19, 2025, 2:07pm

Need a second look on this PR: Faster than Real -Time support in GZ by dirksavage88 · Pull Request #23783 · PX4/PX4-Autopilot · GitHub

stevejohn163 · February 19, 2025, 2:09pm

I’m trying to initiate a takeoff using the MAV_CMD_NAV_TAKEOFF command via COMMAND_INT. The problem is that when I set the altitude with different MAV_FRAME values, the takeoff altitude doesn’t change as expected. No matter which MAV_FRAME I use and set the takeoff altitude to 2.5m, it always tries to take off to 2.5m. However, since I’m already at an altitude of 200m above mean sea level, it throws the error: “Already higher than takeoff altitude.”
https://github.com/PX4/PX4-Autopilot/blob/04a3c4af20df9e851e281776bfa51d02759ddee6/src/modules/navigator/takeoff.cpp#L117

Currently, I have to manually retrieve the MSL (mean sea level) altitude, add 2.5m to it, and send that value via COMMAND_INT for it to work properly.