Object motion detection based on passive UHF RFID tags using a hidden Markov model-based classifier.

Area of Science:

• Computer Science
• Electrical Engineering
• Robotics

Background:

• Activity recognition relies on understanding object movement and identity.
• Existing methods for tracking work objects can be limited in dynamic environments.
• Passive UHF RFID tags offer a low-cost sensing modality for object tracking.

Purpose of the Study:

• To develop an object motion detection system using backscattered signal strength from passive UHF RFID tags.
• To enhance activity recognition by providing reliable information on object movement and identity.
• To evaluate the system's performance in a real-world, high-stakes environment.

Main Methods:

• Utilized backscattered signal strength (RSSI) from passive UHF RFID tags as a sensor.
• Developed a novel Markov model with continuous observations, including an RSSI preprocessor, frame-based data segmentation, and a motion-transition finder.
• Employed a Hidden Markov Model (HMM)-based classifier trained for dynamic settings to detect transitions between stationary and moving states.

Main Results:

• The system reliably detects object movement by analyzing changes in backscattered signal strength due to tag relocation.
• The HMM-based classifier achieved high accuracy in distinguishing between stationary and moving states.
• The RFID motion detection system demonstrated 89.5% accuracy in a hospital trauma bay during simulated resuscitations.

Conclusions:

• Passive UHF RFID tag RSSI is a viable sensor for robust object motion detection.
• The proposed Markov model and HMM-based classifier effectively identify object movement and transitions in dynamic settings.
• This system has significant potential for improving activity recognition and workflow management in critical environments like hospital trauma bays.