A Real-Time Self-Sensing Approach to Sensor Array Configuration Fusing Prior Knowledge for Reconfigurable Magnetic Tracking Systems

Shijian Su,Xindi Yang,Zhen Li,Hongliang Ren

DOI: https://doi.org/10.1109/tmech.2024.3435456

2024-01-01

Abstract:Wearable and reconfigurable magnetic tracking systems (MTSs) for tracking passive magnetic targets inside the body offer the benefit of closely conforming to the curvature of the human abdomen and accommodating individuals with different abdominal circumferences. However, the dynamic changes in sensor array configuration, known as deformation, caused by posture changes during long-duration examinations, impose significant challenges on MTSs that heavily rely on magnetometer poses. To address this, we propose a real-time self-sensing method based on the sensor array structural model and magnetic dipole model, which simultaneously estimates the magnet pose and the hinge angles on the sensor array. Experimental results demonstrate that our reconfigurable MTS achieves an average localization accuracy of 1.36 +/- 0.21 mm and 1.00 +/- 0.46(degrees) even under severe deformation. Furthermore, we incorporate the prior knowledge about the sensor array configurations into the self-sensing algorithm. Theoretical analysis and experimental verification demonstrate that integrating prior knowledge can enhance algorithm performance, with a 28.57% enhancement in stability and up to a 6% improvement in global convergence rate.