Achieving Out-of-the-Box Dead Reckoning with PX4/ArduPilot and the ANELLO X3 IMU
Technical application note
A practical path to GNSS-resilient navigation using Pixhawk-class flight controllers.
The integration idea is simple: add the ANELLO X3 IMU to a Pixhawk-class autopilot, merge the ANELLO driver, connect over UART, and let the autopilot EKF use the X3 as its primary IMU for GNSS-denied inertial holdover.
GNSS is the primary navigation reference for most UAV operations, but it cannot be assumed available or trustworthy at all times. Terrain masking, urban canyons, interference, jamming, and spoofing can interrupt or corrupt satellite navigation. When that happens, the aircraft still needs a credible estimate of attitude, velocity, and position.
For teams already building on PX4 or ArduPilot, the ANELLO X3 IMU adds higher-performance inertial sensing without replacing the existing flight-control stack. The X3 uses three ANELLO SiPhOG optical gyroscopes, is designed for small SWaP-C, and communicates directly with a flight controller over a UART interface.
The navigation solution remains inside the autopilot EKF. ANELLO provides PX4 and ArduPilot drivers so integrators can add X3 inertial performance while retaining their existing autopilot architecture.
1. Why This Architecture Matters
The main advantage is architectural simplicity. A team can keep a familiar COTS Pixhawk-class flight controller, such as a Cube Orange-class system, and add X3 inertial performance rather than adopting a completely separate navigation computer.
| Design Goal | What It Means for the Integrator |
|---|---|
| COTS flight control | Retain a Pixhawk-class autopilot and the existing PX4/ArduPilot ecosystem. |
| Small SWaP-C | Add GNSS-resilient inertial capability without a large external navigation computer. |
| UART integration | Connect the X3 directly to the flight controller through an available serial interface. |
| No external infrastructure | The inertial solution does not depend on beacons, maps, visual landmarks, or a prepared operating area. |
| Environment-independent sensing | Inertial sensing does not require lighting, visual texture, or a visible ground scene; this is particularly valuable over water or in poor visibility. |
| Open ecosystem | PX4 and ArduPilot have large, active user and developer communities, making the solution extensible and familiar to many UAV teams. |
2. How the Integration Works
The integration can be reduced to four practical steps.
Merge the driver. Use the ANELLO-provided PX4 or ArduPilot driver in the target flight-stack build.
Connect the X3. Install the ANELLO X3 on the aircraft and connect it to power and an available UART on the flight controller. The X3 can run on 4.5-24 VDC at 5 W.
Configure the estimator. Set up the flight stack so the autopilot EKF receives ANELLO X3 inertial data as the primary IMU. Verify orientation, update rate, timestamps, lever arms, and estimator health.
Validate aided and unaided operation. First fly with good GNSS, then introduce controlled GNSS outages and measure drift and recovery behavior.
The X3 is not creating a separate black-box navigation solution. The autopilot EKF creates the navigation solution using X3 gyro and accelerometer measurements directly.
3. What Happens When GNSS Disappears
During normal flight, GNSS provides an external position and velocity reference that continually constrains errors in the navigation solution. When GNSS becomes unavailable, that external reference is removed.
The PX4 or ArduPilot EKF continues propagating the aircraft state using the IMU and any other aiding measurements that remain available. At its simplest, this is dead reckoning: estimating where the aircraft is now based on where it was, how fast it was moving, and how its motion has changed since the previous update.
The challenge is that small inertial errors accumulate. Gyroscope performance is especially important because accurate attitude estimation is required to interpret accelerometer measurements correctly.
With GNSS available, external measurements help constrain these errors. Once GNSS is lost, the position solution becomes increasingly dependent on the quality and stability of the inertial measurements and on any remaining aiding sources.
4. Where the Architecture Performs Best
The X3 can support multiple vehicle types, but fixed-wing UAVs with measured airspeed are particularly attractive for dead-reckoning applications. A pitot tube gives the EKF an additional motion constraint that can help bound parts of the navigation solution when GNSS is unavailable.
The broader principle is important: GNSS-denied does not have to mean completely unaided. The autopilot can combine X3 inertial data with complementary sensors such as airspeed, barometric altitude, magnetometers, range sensors, optical flow, vision, or other navigation systems when the mission benefits from them.
X3 provides a strong inertial core, while the integrator remains free to add complementary navigation sources for a specific CONOPS.
5. Why Inertial Holdover Is Different from Camera-Based Navigation
Vision can be a powerful navigation aid, but it is inherently dependent on the external scene. Cameras may struggle over water, low-texture terrain, darkness, fog, smoke, glare, or other poor-visibility conditions. They can also add compute, calibration, and mission-planning requirements.
The X3 inertial path is different: it measures the aircraft's own motion. It does not require a prepared environment, known landmarks, or usable imagery. That makes it a useful navigation foundation for aircraft expected to operate across varied terrain, weather, and lighting conditions.
This is not a claim of zero drift. Any unaided inertial solution accumulates error over time. The engineering objective is to extend the interval over which the dead-reckoned solution remains useful and to measure that performance for the intended mission.
6. A Practical Validation Workflow
A good integration test should progress in stages rather than jumping directly to a long GNSS-denied flight.
Simulation
- Fly as normal, with full data logging for replay enabled from system on.
- Use replay to simulate GNSS outages.
- Verify that the flight solution is stable and navigation looks good.
Real-time testing
- Bench-check sensor orientation, signs, timing, update rate, and UART reliability.
- Run stationary tests and confirm stable attitude, bias estimates, and EKF health.
- Fly with normal GNSS while logging estimator innovations and X3 data.
- Introduce short, controlled GNSS outages, then increase outage duration.
- Measure horizontal drift, heading error, and vertical error by comparing the reported navigation solution against GNSS position over telemetry.
- Restore GNSS and confirm that the EKF validates and re-aids cleanly rather than producing an unsafe state jump.
The most useful result is not simply "the aircraft kept flying." It is a quantitative curve showing how navigation error grows versus GNSS-outage duration under representative vehicle dynamics.
7. The Result: A Capable, Extensible GNSS-Resilient Platform
The combination of a COTS Pixhawk-class autopilot, PX4 or ArduPilot, and the ANELLO X3 IMU provides a route to GNSS-resilient navigation without discarding the flight-control stack teams already use.
The autopilot retains ownership of the EKF and vehicle state, while gaining direct access to ANELLO X3 optical-gyro IMU data. That architecture provides three advantages at once: a straightforward integration path, freedom from external navigation infrastructure, and an open framework for adding complementary navigation sources as mission requirements evolve.
See the ANELLO X3 IMU at Commercial UAV Expo
Las Vegas | September 1-3, 2026 | Booth #1827
Bring your PX4 or ArduPilot platform, your GNSS-denied requirements, and your integration questions. The ANELLO team can walk through the driver, UART integration, EKF configuration, and validation approach for your aircraft.
Contact ANELLO