v1.17.0
Chainlit/chainlitv1.17.0May 13, 2026by github-actions[bot]
AI Summary
Major release for PX4 Autopilot adding new flight modes, ROS 2 control interfaces, and on-device AI capabilities.
Key Highlights
- New Altitude Cruise flight mode
- Fixed Wing Takeoff improvements on navigation loss
- ROS 2 control interfaces for fixed-wing and rover
- Zenoh middleware maturity (rmw_zenoh compatibility)
- TensorFlow Lite Micro integration for on-device AI
New Features
- New multicopter flight mode: Altitude Cruise
- Fixed Wing Takeoff mode improvements
- ROS 2 control interfaces (FwLateralLongitudinalSetpointType, RoverSetpointTypes)
- Zenoh middleware maturity
- MC Neural Network Control integration (TensorFlow Lite Micro)
Full Release Notes
## What's Changed PX4 v1.17 builds on [PX4 v1.16](https://docs.px4.io/main/en/releases/1.16), with the changes below landing since v1.16.2. This release adds [Altitude Cruise](https://docs.px4.io/main/en/flight_modes_mc/altitude_cruise) mode, improves Fixed Wing Takeoff behaviour on navigation loss, and exposes cleaner high-level fixed-wing and rover control interfaces for ROS 2 workflows. The in-tree Zenoh middleware matures to rmw_zenoh compatibility, simulation gains Gazebo Jetty support and Ackermann SIH, and three new INS drivers (MicroStrain, sbgECom, EULER-NAV) join the ecosystem alongside Septentrio GNSS resilience reporting and barometer auto-calibration against GNSS height. PX4 v1.17 also includes user-visible MAVLink, RC, logging, failsafe, and rover refinements across the stack. ## Major Changes * New multicopter flight mode: [Altitude Cruise](https://docs.px4.io/main/en/flight_modes_mc/altitude_cruise). Holds tilt and heading on stick release so the vehicle keeps cruising at a steady velocity instead of stopping like Altitude mode does. * [Fixed Wing Takeoff mode](https://docs.px4.io/main/en/flight_modes_fw/takeoff) now keeps climbing with level wings on navigation loss and can use the takeoff waypoint latitude and longitude to define the loiter position. ([PX4-Autopilot#25226](https://github.com/PX4/PX4-Autopilot/pull/25226)) * Fixed-wing vehicles (and VTOLs in fixed-wing mode) can now be controlled from ROS 2 via the new [FwLateralLongitudinalSetpointType](https://docs.px4.io/main/en/ros2/px4_ros2_control_interface#fw-lateral-longitudinal-setpoint) in the PX4 ROS 2 Control Interface, exposing direct lateral and longitudinal setpoints. * Rovers can now be controlled from ROS 2 via the new [RoverSetpointTypes](https://docs.px4.io/main/en/ros2/px4_ros2_control_interface#rover-setpoints) in the PX4 ROS 2 Control Interface, with valid combinations of position, speed, throttle, attitude, rate, and steering setpoints exposed as guaranteed-valid setpoint types. See [Rovers Apps & API](https://docs.px4.io/main/en/flight_modes_rover/api). * The in-tree [Zenoh middleware](https://docs.px4.io/main/en/middleware/zenoh) matures to rmw_zenoh compatibility (CDRv1 serialization, ROS 2 graph liveliness, auto-generated config from dds_topics.yaml, Domain ID parameter, Zenoh CLI). Zenoh is built into the default firmware on FMU-v6xRT (make px4_fmu-v6xrt_default); on FMU-v6x and SITL it ships as a zenoh build variant (make px4_fmu-v6x_zenoh, make px4_sitl_zenoh). * Initial [MC Neural Network Control](https://docs.px4.io/main/en/neural_networks/mc_neural_network_control) test path: PX4 v1.17 integrates [TensorFlow Lite Micro](https://docs.px4.io/main/en/neural_networks/tflm) on-device so an externally trained network (for example trained with reinforcement learning in [Aerial Gym](https://ntnu-arl.github.io/aerial_gym_simulator/)) can be loaded as a tflite model and substituted for the multicopter controller for research and bench testing. It is not a replacement for the production controller stack.([PX4-Autopilot#24366](https://github.com/PX4/PX4-Autopilot/pull/24366)) ## Release Notes * [Click here to view the full Release Notes](https://docs.px4.io/main/en/releases/1.17) * Full Changelog [v1.16.2...v1.17.0](https://github.com/PX4/PX4-Autopilot/compare/v1.16.2...v1.17.0) ## New Contributors * @ischollETH - first contribution: #25215 * @vololand - first contribution: #25362 * @czx-fly - first contribution: #25364 * @Sayshara - first contribution: #24991 * @accton-iot - first contribution: #25102 * @ljarvela - first contribution: #25454 * @fbaklanov - first contribution: #24534 * @renjieDLUT - first contribution: #25527 * @rmahoney-skai - first contribution: #25539 * @HTRamsey - first contribution: #25486 * @jyhminwang - first contribution: #25411 * @tolesam - first contribution: #24137 * @SolderSyntax - first contribution: #25441 * @alexespinoza28 - first contribution: #25525 * @Hs293Go - first contribution: #25444 * @tobias-auterion - first contribution: #25602 * @Luka-Filipovic - first contribution: #25618 * @Louis-max-H - first contribution: #25012 * @msberk - first contribution: #25587 * @Parkhb1106 - first contribution: #25619 * @radiolinkW - first contribution: #25562 * @Siri2K - first contribution: #25637 * @airpixel-cz - first contribution: #25651 * @JacopoPan - first contribution: #24040 * @annoybot - first contribution: #25649 * @asherikov - first contribution: #25742 * @ttechnick - first contribution: #25897 * @MDEAGEWT - first contribution: #25776 * @AkaiEurus - first contribution: #26199