Barometer-aided attitude estimation offers reliable backup for GNSS-denied drones
When GPS fails, a simple barometer paired with an IMU can accurately estimate attitude.
A team from the University of Technology of Troyes and the University of Sherbrooke—Melone Nyoba Tchonkeu, Soulaimane Berkane, and Tarek Hamel—has published a paper detailing a novel barometer-aided attitude estimation framework for autonomous vehicles operating in GNSS-denied environments. The core innovation lies in leveraging barometric altitude measurements to disambiguate gravitational and inertial accelerations, a challenge that plagues pure IMU-based systems during aggressive maneuvers. The authors propose two complementary observers: first, a deterministic Riccati observer cascaded with a complementary filter that preserves the geometric structure of SO(3) while ensuring almost-global asymptotic stability under uniform observability conditions. Second, a unified nonlinear observer on SO(3)×R² that uses IMU inputs and barometer/magnetometer outputs, achieving local exponential stability under relaxed observability assumptions.
The proposed architecture is validated using both simulated datasets and real flight data from a small UAV. Results show that barometer feedback significantly improves tilt estimation accuracy compared to IMU-only configurations, especially during periods of high acceleration or when velocity measurements from GNSS, pitot tubes, or optical flow are unavailable. The barometer adds negligible weight and cost, making it a practical sensing modality for minimal-sensor setups. The paper argues this approach can serve as a reliable fallback or complement to existing velocity-based observers, enhancing robustness in GPS-denied navigation for drones, micro-aerial vehicles, and other autonomous systems.
- Two observer designs: a Riccati observer with complementary filter (AGAS under uniform observability) and a unified nonlinear observer on SO(3)×R² (LES under relaxed conditions).
- Barometric altitude resolves the tilt ambiguity from IMU alone during high-acceleration flight, eliminating the need for GNSS, Doppler radar, or visual odometry.
- Validated on both synthetic data and real flight logs; demonstration of lightweight, low-cost attitude estimation for minimal-sensing UAV configurations.
Why It Matters
This gives drones a cheap, reliable attitude backup when GPS drops, critical for package delivery and search-and-rescue.