Enhanced UAV Navigation Systems through Sensor Fusion with Trident Quaternions

TL;DR

This work addresses robust UAV navigation with low-cost sensors by introducing trident quaternions, an extension of dual quaternions, to encode attitude, velocity, and position within a single algebraic object. It derives strapdown navigation equations in the trident-quaternion framework and fuses inertial measurements with GPS using an Extended Kalman Filter, exploiting a left-handed error formulation for state correction. Experimental validation on a symmetric X-config multi-rotor shows general agreement with a reference autopilot and coherent velocity/position estimates, while yaw observability depends on trajectory and magnetometer usage. The proposed framework offers a compact, numerically stable, and implementation-friendly approach with potential for barometer/vision integration and future multi-vehicle control applications.

Abstract

This paper presents an integrated navigation algorithm based on trident quaternions, an extension of dual quaternions. The proposed methodology provides an efficient approach for achieving precise and robust navigation by leveraging the advantages of trident quaternions. The performance of the navigation system was validated through experimental tests using a multi-rotor UAV equipped with two navigation computers: one executing the proposed algorithm and the other running a commercial autopilot, which was used as a reference.

Figures (12)

• Figure 1: Multi-rotor UAV used for experimental testing.
• Figure 2: Estimation of the vehicle orientation (Blue: PixHawk, Red: Trident quaternions).
• Figure 3: Estimation of the attitude covariance.
• Figure 4: Estimation of the vehicle velocity in the ECEF frame (Blue: PixHawk, Red: Trident quaternions).
• Figure 5: Estimation of the vehicle position (Blue: PixHawk, Red: Trident quaternions).