September 07, 2026 PX4 Weekly Dev Update: NuttX 12.12.0 Upgrade Completed, Autonomous Flight and UAVCAN Integration Accelerated

PX4 Weekly Integration Briefing

Over the past week, the PX4 Autopilot project has seen vigorous development activities, focusing on strengthening core system stability and expanding next-generation autonomous flight capabilities. Notably, the successful merger of the NuttX 12.12.0 platform upgrade has laid a robust foundation for profound improvements in PX4’s low-level operations and future feature expansion. In conjunction, enhanced UAVCAN integration is increasing the reliability and flexibility of sensor and actuator communication in distributed system environments, while advanced autonomous flight features such as mission route planning libraries, precision landing and takeoff, and external avoidance setpoint fusion are actively being discussed and developed.

These advancements are not merely additive features but are fundamentally aimed at enhancing drone safety and mission execution capabilities. Numerous bug fixes and backports are underway across a wide range of areas, including control, navigation, drivers, and board-specific optimizations, demonstrating PX4’s strong focus on delivering a stable flight experience. The development community continues to engage in discussions on technical depth and practical problem-solving through weekly dev calls and forums, testifying to the project’s healthy ecosystem.

PX4 Releases and Key Changes by Version

While there have been no new stable version releases in the past 7 days, v1.18.0-beta2, released on August 09, 2026, serves as a crucial pre-stage for the upcoming 1.18 stable release, encompassing a wide range of improvements and fixes as follows. This is part of an ongoing effort to address various issues discovered during flight testing and maximize stability.

  • Version: v1.18.0-beta2
  • Release Date: 2026-08-09T04:44:58Z (Prerelease)
  • Key Changes:
    • Fixed-wing: Resolved NaN eas2tas issue due to zero-airspeed division in lat/lon control, fixed tailsitter malfunction caused by incorrect attitude frame.
    • QGC Compatibility: Re-accepted QGC compatibility regressions (multicopter guided takeoff, camera mission items, VTOL landing patterns) caused by stricter command validation.
    • STM32H7 Boards: Fixed SD card read corruption issue by improving NuttX SDMMC cache consistency.
    • Parameter Storage: FMUv6X-RT FRAM multi-page write and moved flashparams compression at boot.
    • FMUv6X-RT: Resolved load-dependent hardfaults with FlexSPI DLL calibration at boot.
    • Estimator: Force reset of external position estimate during auxiliary function activation, fixed DroneCAN rangefinder out-of-range readings from being fused into terrain data.
    • Navigation: Fixed DO_JUMP to correctly interpret the target when set as the current mission item.
    • Battery: Fixed RTL from operating when remaining time is unknown, used unclamped dt for coulomb counting.
    • Commander: Added COM_PARACHUTE warning-only mode, multicopter autotune can be re-triggered without reboot.
    • VOXL2 / Qurt Platforms: Implemented px4_task_join, fixed Hexagon NaN comparisons, startup stdout TLS race condition, fixed keepalive file directory.
    • GPS: RTCM corrections and moving baseline data separated into distinct uORB topics.
    • MAVLink: Resized receive thread stack; resized configuration for FMUv4 (Pixracer) flash headroom (UAVCAN, uXRCE-DDS, mag bias estimator, SIH excluded from default build).
    • Boards: CUAV X25-mega uses renamed heater sensor ID parameters.

The v1.18.0-beta2 release demonstrates the project’s deep commitment to stability, specific hardware platform support, and MAVLink communication optimization. In particular, resolving QGC compatibility regressions reflects PX4’s philosophy of valuing user experience and indicates ongoing multi-faceted efforts to ensure the stability of various flight modes and sensor integrations.

Core GitHub Updates (PX4-Autopilot)

Over the past week, the PX4-Autopilot GitHub repository saw a total of 57 Pull Request (PR) activities, of which 32 were merged (Closed), and 17 issues were reported. Key updates include:

Analysis of Significant Merged PRs

In-depth Analysis of Open Critical Issues and PRs

Weekly Dev Call & Community Trends

The PX4 Weekly Dev Call, held on September 2nd, consisted of team synchronization and a community Q&A session, with active development status sharing and Q&A. This highlights the PX4 team’s transparent operations and continuous efforts to communicate with the community.

The following key trends were observed in the Discourse forum:

  • Dronecode: The full-time drone technical engineer job postings in the Dronecode section indicate a continuous increase in demand for skilled professionals across the drone industry, particularly in areas leveraging the PX4 and Dronecode technology stacks.
  • Pixhawk: Discussions on an ‘Intercept Flight mode’ for fixed-wing UAVs demonstrate a demand for more sophisticated and strategic flight modes for fixed-wing aircraft. This reflects PX4’s focus on developing advanced control solutions for various vehicle types.
  • PX4 Autopilot:
    • Topics such as “What happens with this unstable drift, what root cause?” and “Heading drift in acceleration mode in gz sim” indicate deep community interest in navigation and estimator stability issues experienced by users in real flight and simulation environments. This emphasizes the importance of continuous improvements to EKF2 and other estimation algorithms.
    • “Introducing PID-Wise: interactive PID tuning from a logfile” is a positive example of the community’s effort to share and utilize self-developed tools to enhance development productivity. Logfile-based PID tuning tools are highly valuable resources for firmware developers.
    • Leading discussions like “How AI Can Improve Autonomous Drone Navigation & Decision-Making” demonstrate the PX4 community’s active exploration of future directions for integrating AI technologies into drone autonomy. This offers a glimpse into PX4’s long-term vision.

Subsystem Trends (MAVLink, MAVSDK, QGC)

Continuous efforts to enhance the core functionalities and interoperability of the PX4 Autopilot are also observed across key subsystems.

  • MAVLink:

    Insight: MAVLink is continuously evolving beyond a mere communication protocol, becoming a crucial interface for interaction with various subsystems, including gimbal control, external sensor data injection, and advanced parameter management.

  • MAVSDK:

    Insight: Although direct update activities for MAVSDK itself were sparse, the ongoing tests within PX4 Autopilot, considering MAVSDK integration, demonstrate continuous interest in ensuring the reliability of mission control through MAVSDK.

  • QGroundControl:
    • In the v1.18.0-beta2 release notes, “QGC compatibility regressions from stricter command validation” were mentioned, and fixes were made to allow multicopter guided takeoff, camera mission items, and VTOL landing patterns to be re-accepted by QGC.

    Insight: As PX4 firmware’s internal logic evolves, compatibility issues with ground control systems like QGroundControl may arise. However, the PX4 development team is committed to swiftly identifying and resolving these regressions, thereby striving to provide users with a consistent and stable QGC experience.

Similar Posts

답글 남기기