Magnetic Pose Estimation and Robotic Manipulation of Magnetically Actuated Capsule Endoscopes

Abstract

Magnetic actuation is currently the most promising approach for actively and wirelessly manipulating capsule endoscopes (CEs) inside the gastrointestinal (GI) tract. Actively actuated CEs have the potential to transform current medical practice and significantly improve patient outcomes. In this dissertation, two fundamental problems associated with active actuation of CEs are studied. Our approach to solving these problems were developed in the context of advancing our robotically guided soft-tethered CE system toward clinical adoption. The first problem pertains to estimating the pose of a CE while it is inside a patient. Current magnetic field based pose estimation methods face significant challenges in their path to clinical adoption. First, due to presence of regions of magnetic field singularity, the accuracy of the system can be significantly degraded while being used for actuation. Second, current pose estimation methods need accurate knowledge of the capsule's initial pose before they can successfully track the capsule. This, however, may not be possible to do in clinical settings. We propose a novel hybrid approach employing a combination of static and time-varying magnetic field sources and show that this system has no regions of singularity while eliminating the need for initialization. The proposed system was experimentally validated for accuracy, workspace size, update rate and performance in regions where magnetic singularity previously existed. The second problem has to do with closed-loop control, where computed pose estimates are used as sensory feedback to compensate for deviation of the capsule's pose from the desired or commanded pose. We present two schemes of closed-loop control expanding existing formulations of magnetic control and integrating them with our pose estimation method. The controllers were validated in autonomous path following tasks in simulation and real world experiments. Our solution to these problems are foundational to the future development of effective tele-operation systems where the motion of the CE is transparently controlled by a physician.