Enabling Robotic Capsule Endoscopy Through Effective in-Body Capsule Localization.

This article focuses on the challenge of accurately localizing and orienting a wireless capsule endoscope within the human gastrointestinal (GI) tract, a critical requirement for robotic capsule endoscopy (RCE) to enable precise diagnosis and potential surgical interventions. The authors present a novel cyber-physical system (CPS) that models the GI tract using 3D-printed intestinal segments and employs magnetic localization techniques based on measurements from multiple Hall-effect sensors to estimate the capsule’s position and orientation. They evaluate three metaheuristic optimization algorithms—Artificial Bee Colony (ABC), Particle Swarm Optimization (PSO), and Teaching-Learning-Based Optimization (TLBO)—both individually and in hybrid combinations with the Levenberg-Marquardt (LM) algorithm, finding that the hybrid PSO + LM approach achieves the best balance of localization accuracy (approximately 3.1 mm error) and orientation error (about 2.2°) with reasonable computation time. The study demonstrates that magnetic localization, supported by this CPS, can provide highly accurate, repeatable, and real-time localization performance necessary for advancing RCE technologies.