Modelling and solving the position tracking problem of remote-controlled gastrointestinal drug-delivery capsules

A theoretical basis is presented to develop a real-time position tracking system for a remote-controlled drug-delivery capsule that would enable the targeted delivery of drugs to specific locations in the gastrointestinal tract. Tracking is accomplished by dual magnetic vector detection over time, using two separate sets of magnetic field sensing devices. One is used to detect the alternating magnetic field excited externally with certain frequencies, while the other is used to detect the geomagnetic field. A mathematical model of the magnetic flux density in space and time is constructed using the Biot-Savart law. Additionally, the earth's magnetic field and quaternion rotation theory are also used in the model to compensate for the constantly changing spatial orientation of the capsule as it travels through the gastrointestinal tract. Based on the model, an improved artificial bee colony algorithm is used to solve the inverse magnetic field problem. Firstly, chaotic sequencing improves the initial solution diversity, and a ranked selection strategy is applied. At later stages, the Levenberg-Marquardt algorithm is introduced in order to accelerate convergence. To verify the theoretical basis presented above, a prototype of the tracking system is developed. Calculations verification results in a convergence rate of 100% with an average of 179 iterations. The prototype testing shows that the dual magnetic vector detection method simplifies the solution of the inverse magnetic field problem, shortens the tracking time for each round of data, and increases the solution accuracy.