Theoretical Data-Driven MobilePosenet: Lightweight Neural Network for Accurate Calibration-Free 5-DOF Magnet Localization

2025-01-06

AI Summary

Real-time, high-precision localization of permanent magnets in wireless capsule endoscopy robots remains challenging due to limitations of existing methods—including reliance on hardware pre-calibration, dependence on empirical measurements, or excessive computational overhead. Method: We propose MobilePosenet, a lightweight neural network that eliminates both hardware calibration and empirical data requirements. It introduces a purely theory-driven training paradigm: synthetic data generated from the analytical magnetic dipole model, augmented with coordinate embedding, noise injection, and realistic bias compensation; further enhanced by depthwise separable convolutions and SE channel-wise attention for joint efficiency–accuracy optimization. Results: In a 90×90×80 mm workspace, MobilePosenet achieves mean translational and rotational errors of 1.54±1.03 mm and 2.24±1.84°, respectively, with only 0.9 ms inference latency per sample. To our knowledge, this is the first method enabling five-degree-of-freedom magnetic localization without initial pose estimation or empirical data.

Technology Category

• Intelligent Robots: Localization, Mapping, and Navigation
• Computer Vision: Motion & Tracking
• Machine Learning: Calibration & Uncertainty Quantification

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Permanent magnet tracking using the external sensor array is crucial for the accurate localization of wireless capsule endoscope robots. Traditional tracking algorithms, based on the magnetic dipole model and Levenberg-Marquardt (LM) algorithm, face challenges related to computational delays and the need for initial position estimation. More recently proposed neural network-based approaches often require extensive hardware calibration and real-world data collection, which are time-consuming and labor-intensive. To address these challenges, we propose MobilePosenet, a lightweight neural network architecture that leverages depthwise separable convolutions to minimize computational cost and a channel attention mechanism to enhance localization accuracy. Besides, the inputs to the network integrate the sensors' coordinate information and random noise, compensating for the discrepancies between the theoretical model and the actual magnetic fields and thus allowing MobilePosenet to be trained entirely on theoretical data. Experimental evaluations conducted in a (90 imes 90 imes 80) mm workspace demonstrate that MobilePosenet exhibits excellent 5-DOF localization accuracy ($1.54 pm 1.03$ mm and $2.24 pm 1.84^{circ}$) and inference speed (0.9 ms) against state-of-the-art methods trained on real-world data. Since network training relies solely on theoretical data, MobilePosenet can eliminate the hardware calibration and real-world data collection process, improving the generalizability of this permanent magnet localization method and the potential for rapid adoption in different clinical settings.

Problem

• Magnetic Localization
• Wireless Capsule Endoscopy
• Real-time Computation

Innovation

• MobilePosenet
• lightweight neural network
• theoretical data training