Millimeter-Scale Magnetic Positioning Using a Single AMR Sensor and BP Neural Network.

Area of Science:

• Robotics and Automation
• Sensor Technology
• Machine Learning Applications

Background:

• Conventional positioning systems often require multiple sensors, increasing cost and complexity.
• Miniaturization of positioning systems is crucial for applications with space constraints.
• Accurate real-time tracking is essential in various advanced technological fields.

Purpose of the Study:

• To develop a cost-effective and miniaturized positioning system using a single sensor.
• To achieve high-precision real-time tracking and localization via magnetic field measurement.
• To explore the application of BP neural networks for magnetic field data conversion.

Main Methods:

• Utilized a single anisotropic magnetoresistive (AMR) sensor to detect the magnetic field of a target permanent magnet.
• Employed a Backpropagation (BP) neural network to process three-axis magnetic field data.
• Converted magnetic field data into precise coordinate information for localization.

Main Results:

• Achieved millimeter-level accuracy in simulations (RMSEs of 0.27 mm for X, 0.26 mm for Z) and real-world tests (RMSEs of 0.83 mm for X, 1.15 mm for Y, 0.85 mm for Z).
• Demonstrated positioning error correlation with magnetic field variations due to distance-dependent nonlinearity.
• Successfully tracked and localized within a 50 × 40 × 40 mm³ volume.

Conclusions:

• The single-sensor AMR system offers a significant reduction in hardware cost and complexity compared to multi-sensor systems.
• The BP neural network effectively converts magnetic field data to achieve accurate, real-time positioning.
• This method is suitable for high-precision applications in aerospace, medical devices, and automation where space is limited.