A wearable system for asymmetric contactless human sensing

Area of Science:

• Wearable technology
• Human-computer interaction
• Sensor systems

Background:

• Existing human detection systems often raise privacy concerns due to audio/video capture.
• Previous methods required dedicated hardware for all individuals involved in interaction.
• There is a need for unobtrusive and accessible human presence detection.

Purpose of the Study:

• To demonstrate a wearable sensor for detecting human presence and interaction parameters.
• To develop a privacy-preserving alternative to audio-visual detection methods.
• To create a system that does not require dedicated hardware for all interacting individuals.

Main Methods:

• Utilized a combination of ultrasonic distance and passive infrared (PIR) sensors.
• Developed a K-Nearest Neighbors (KNN) classifier.
• Collected data from a controlled indoor experiment simulating real-world geometries.

Main Results:

• Achieved an overall accuracy rate of 94.2% in detecting human interactions.
• The system effectively identifies human presence within a 90-degree sector and a three-foot radius.
• Demonstrated robustness across various indoor geometries.

Conclusions:

• The developed wearable sensor offers a privacy-conscious and hardware-flexible solution for human presence detection.
• The system is suitable for applications requiring monitoring of human interaction parameters like distance and duration.
• The KNN classifier with three features proved effective for accurate human interaction detection.