June 08, 2026 PX4 Weekly Dev Update: Accelerating Safety Enhancements and Core Infrastructure Upgrades

PX4 Weekly Integrated Briefing

As of June 08, 2026, the PX4 Autopilot project has seen a flurry of activity across various domains over the past week, accelerating development for the upcoming v1.18.0 release. Notably, there’s a prominent trend towards focused discussion and implementation of core functionalities crucial for safe drone operation and Regulatory Compliance. The expansion of the Detect and Avoid module to comply with ASTM F3442 standards, geo-fence aware RTL (Return-To-Land) capabilities, and updates related to Remote ID are expected to enhance the stability and fulfill legal requirements of autonomous drone operating environments.

Concurrently, fundamental infrastructure improvements are actively underway, including the upgrade of NuttX OS to version 12.12.0, refactoring of the uORB middleware, and a shift towards SIH (Software-in-the-Loop) as the primary simulation environment. These changes will play a crucial role in enhancing PX4 system stability, performance, and long-term development and testing efficiency. Furthermore, the addition of support for numerous new flight control boards, along with various bug fixes and optimizations, continuously contributes to PX4’s hardware extensibility and improved flight stability. In the community forums, developers actively discussed simulation environment configurations, hardware integration issues, and architectural direction.

PX4 Releases and Key Version-Specific Changes

There were no new official PX4 Autopilot releases over the past seven days. The PX4 development team is currently focused on testing and further developing the v1.18.0-alpha1 version, which will serve as the foundation for the upcoming v1.18.0 stable release. Developers and researchers are encouraged to test this alpha version in preparation for the next stable release.

The most recent stable release is v1.17.0, announced on May 13, 2026. This version includes various improvements since v1.16.2, with the following key changes:

  • New Multicopter Flight Mode: Altitude Cruise – A mode that maintains the vehicle’s inclination and heading to cruise at a constant speed when sticks are released.
  • Fixed-wing Takeoff Mode Improvements – Maintains ascent in level flight upon navigation loss and allows defining the loiter position using the latitude and longitude of the takeoff waypoint.
  • ROS 2 Control Interface Expansion – New FwLateralLongitudinalSetpointType and RoverSetpointTypes have been added for fixed-wing and rovers, facilitating easier control through ROS 2 workflows.
  • Zenoh Middleware Maturation – Zenoh middleware has been further stabilized, including rmw_zenoh compatibility, CDRv1 serialization, ROS 2 graph activity, and auto-generated configurations.
  • MC Neural Network Control Early Test Path – Integration of TensorFlow Lite Micro adds research and bench testing capabilities, allowing externally trained neural networks to be used as multicopter controllers.
  • New INS Driver Additions – INS drivers such as MicroStrain, sbgECom, and EULER-NAV have been integrated, and Septentrio GNSS resilience reporting and automatic barometer calibration features have also been improved.
  • Overall system improvements and bug fixes have been made across MAVLink, RC, logging, failsafe, and more.

Core GitHub Updates (PX4-Autopilot)

Over the past seven days, the PX4-Autopilot GitHub repository saw a total of 70 Pull Requests (PRs) and 11 Issues. Among these, the following notable Merged PRs and currently active critical Issues are highlighted:

Key Merged PRs

In-depth Analysis of Open Critical Issues

Notable Open PRs

Several important PRs are currently under review, with the following items, in particular, indicating the future direction of PX4:

Weekly Dev Call & Community Trends

The regular PX4 Dev Call took place on June 3rd. This meeting serves as a crucial platform for synchronization among development teams and community Q&A, where major development issues discussed on GitHub are likely addressed.

In the general PX4 Autopilot section of the Discourse forum, various user problems and development discussions were active. Specifically, simulation-related issues such as JSBSim hexarotor_x SITL spawns high / FlightGear-QGC mismatch / auto-disarm are recurring difficulties for multiple users, suggesting the complexity of simulation environment setup or the need for documentation improvements. Furthermore, the bug report DShot motors only initialize after re-assigning output functions (cold boot) — CUAV X7, AM32, v1.16 & v1.17 indicates a hardware initialization problem in the latest stable versions, requiring an urgent solution for CUAV X7 users.

Interestingly, the discussion titled Multi-device vs. separate driver module - community preference/standards? encapsulates a fundamental community question regarding driver architecture. This indicates a need for a standardized approach to driver management and modularization as PX4 evolves. Hardware-specific problems like Overheating STM32H743 Processor or compatibility issues with older versions, such as GPS UBlox F9P Version 1.51 with PX4 Version 1.12, are also continuously being raised.

Subsystem Trends (MAVLink, MAVSDK, QGC)

MAVLink

Several significant activities related to MAVLink have been observed on GitHub. The fix(mavlink): avoid reinitializing command sender semaphore PR has been merged, improving MAVLink communication stability. Additionally, battery status reporting functionality is being enhanced through the feat(mavlink): Battery_Status_V2 MAVLink stream PR, and the fix(dronecan): forward MAVLink OpenDroneID Basic ID to Remote ID PR improves OpenDroneID information transmission via MAVLink, contributing to regulatory compliance. The currently open Tracking issue for the current mavlink implementation indicates ongoing continuous improvement and refactoring across the entire MAVLink system.

MAVSDK

There was no particularly noteworthy significant activity observed in the MAVSDK Discourse forum. However, the continuous improvements in MAVLink, particularly new data streams (Battery_Status_V2) and OpenDroneID integration, hold the potential to provide new features and data accessibility for developers using MAVSDK. The safety issue with offboard control (External modes: setpoints are applied...) will be a critical consideration for developers implementing offboard control via MAVSDK.

QGroundControl (QGC)

Discussions in the QGroundControl forum primarily revolved around connectivity and communication issues. The issue QGroundControl Android Parameter Download Stalls Over Wireless Link but Works via USB Network Sharing points to a parameter download delay problem over wireless links in QGC for Android. This suggests the need for optimization in wireless communication stacks or large data transfer efficiency. Furthermore, PX4 not receiving QGroundControl replies through MOOS UDP bridge indicates a potential MAVLink communication problem between PX4 and QGC in specific setups using a MOOS UDP bridge. These issues directly impact the user experience when QGC interacts with PX4 systems, emphasizing the importance of MAVLink protocol robustness and compatibility across various network environments.

Similar Posts

답글 남기기