Enhancing Untethered Robot Localization : A Fusion Approach Integrating Magnetic Coupling Localization and Ultrasound Imaging

This study introduces a noninvasive localization approach aimed at fusing the data streams from slow-rate pulse-echo sensors and fast-rate magnetic coupling localization. (MCL) sensors, to enhance the positioning precision of externally actuated untethered magnetic robots (UMRs). These systems work by robotically moving rotating permanent magnets (RPMs) enabling control of UMRs in bodily fluids for targeted therapy. By combining the position information of UMRs derived from MCL with the environment’s reconstructed imagery acquired through ultrasound imaging, essential feedback details can be swiftly ascertained with reduced uncertainty. First, we demonstrate the capabilities and limitations of MCL on its own. Investigating down-scaling of the UMR shows us that the localization accuracy of a 12-mm UMR, an 8-mm UMR, a 5-mm UMR, and a 3-mm UMR at a distance of 180 mm from the RPM, in terms of the mean absolute position error, is 0.7 ± 0.7 mm, 0.9 ± 0.4 mm, 1.3 ± 0.2 mm, and 0.3 ± 0.1 mm, respectively. This error tends to decrease when lowering the actuation frequency of the UMR, particularly when operating above the zero dB crossing of the RPM actuator. Second, we formulate a model for directly fusing data from two sensors—specifically, MCL and ultrasound localization. Then we use this fused data to showcase the effective and complementary roles provided by the sensors in enhancing localization accuracy during the wireless actuation. The fusion mechanism empowers the localization of UMRs within blood vessel phantoms, enabling continuous feedback provision even in scenarios where ultrasound signals are obstructed, temporarily or permanently, by wave reflectors. Our findings emphasize that the mean absolute position error for an 8-mm UMR, positioned 140 mm away from the RPM actuator, during pulse-echo localization interruptions, averages 1.4 ± 0.3 mm (n = 18), considering 4 different simulated interruption patterns with 3 distinct measurements each.