Fast and Noise-Resilient Magnetic Field Mapping on a Low-Cost UAV Using Gaussian Process Regression

Area of Science:

• Geophysics
• Robotics
• Data Science

Background:

• Accurate local magnetic field mapping is crucial for various applications.
• Unmanned Aerial Vehicles (UAVs) offer a flexible platform for magnetic surveys.
• UAV-generated magnetic noise can significantly degrade map quality.

Purpose of the Study:

• To develop robust methods for mapping local magnetic field anomalies using UAVs.
• To mitigate magnetic noise originating from UAV electronics.
• To enhance the precision and reliability of UAV-based magnetic field maps.

Main Methods:

• Gaussian Process Regression (GPR) for magnetic field mapping.
• PID controller gain adjustment to reduce high-frequency motor noise.
• A novel 'compromise mapping' technique to handle time-varying magnetic biases.
• Development of a 'consistency metric' for GPR map prediction validation.

Main Results:

• Successfully mapped local magnetic fields with improved robustness to UAV noise.
• Demonstrated reduction of zero-mean noise via PID controller tuning.
• Validated the effectiveness of 'compromise mapping' for time-varying biases.
• Showcased the utility of the 'consistency metric' in state estimation.
• Comparative analysis provided guidelines for optimal trajectory design.

Conclusions:

• The proposed methods significantly enhance the accuracy and reliability of UAV-based magnetic field mapping.
• The 'compromise mapping' and 'consistency metric' offer effective solutions for noise and bias challenges.
• Publicly accessible data and demonstrated best practices will advance future research in this domain.