Multi-Sensor Data Fusion with a Reconfigurable Module and Its Application to Unmanned Storage Boxes

Area of Science:

• Engineering
• Computer Science
• Sensor Technology

Background:

• Multi-sensor data fusion is crucial for enhancing system reliability and accuracy.
• Existing fusion models may lack adaptability for diverse applications.
• The COVID-19 pandemic highlighted the need for reliable monitoring systems, such as for unmanned storage.

Purpose of the Study:

• To propose a novel reconfigurable module (RM) for adaptable multi-sensor data fusion.
• To demonstrate the RM's effectiveness in improving reliability using low-cost sensors.
• To validate the RM's application in a real-world scenario, such as monitoring unmanned storage boxes.

Main Methods:

• Developed a three-layer fusion model: data, feature, and decision layers.
• Implemented data refinement based on sensor characteristics and conversion to logical values.
• Configured a fusion tree with adjustable logical operations for feature extraction and decision making.

Main Results:

• Successfully reconstructed a reconfigurable module (RM) adaptable for specific applications.
• Implemented a prototype system for monitoring unmanned storage boxes using the RM.
• Confirmed enhanced reliability in state identification using low-cost sensors through multi-sensor data fusion.

Conclusions:

• The proposed reconfigurable module (RM) offers an adaptable approach to multi-sensor data fusion.
• The RM enables reliable state identification even with low-cost sensor systems.
• This method provides a flexible framework for various sensor fusion applications.