Adaptive Kernel Kalman Filter for Magnetic Anomaly Detection-based Metallic Target Tracking

This paper proposes the use of the adaptive kernel Kalman filter (AKKF) to track metallic targets using magnetic anomaly detection (MAD). The proposed AKKF-based approach enables accurate tracking of moving metallic targets using magnetometer sensors, even in the presence of dynamic and unknown magnetic moments. The experimental results demonstrate that the proposed method exhibits favourable tracking and estimation performance with reduced computational complexity compared with the bootstrap particle filter (PF). For example, in magnetic moment strength estimation, the relative root mean square error (RRMSE) of the proposed algorithm using 50 particles can approach 2.5% with a computation time of 0.18 seconds, whereas the RRMSE of the PF using 2000 particles is 4.5% with a computation time of 1.4 seconds. This study highlights the potential of AKKF in MAD for metallic target tracking using magnetometer sensors.