Differential Magnetic Sensing with Dynamic Background Calibration for the Capsule Robot Localization in Dynamic Environments

Magnetic localization systems for capsule robots face significant precision degradation in clinical environments due to dynamic electromagnetic interference such as physiological peristalsis and environmental ferromagnetic noise, where traditional static calibration methods exhibit limitations in real-time adaptability. To address this challenge, this study presents a novel magnetic localization system integrating three key innovations: (1) A software-defined differential sensing topology achieving common-mode rejection ratio for effective noise cancellation; (2) A Kalman filter-based dynamic background calibration algorithm reducing environmental interference through real-time noise modeling; (3) A distance-adaptive weighted LM-GN hybrid optimization algorithm, which enhanced convergence speed by 40.6% compared to conventional methods while resolving the trade-off between near-field model errors and far-field convergence efficiency. Experimental validation using a 5x5 triaxial magnetometer array (resolution: 0.2μT, sampling rate: 200 Hz) demonstrated 2.52 mm RMS positioning and 2.97° orientation accuracy in dynamic interference scenarios such as moving metallic objects, outperforming traditional static calibration (6.28 mm) by 59.4% in error reduction. Further medical capsule tracking tests in gastrointestinal phantoms confirmed submillimeter stability under physiological peristalsis and external electromagnetic disturbances. This system established a robust solution for high-dynamic clinical environments, advancing precision magnetic navigation in minimally invasive diagnostics.