Development of medication assistance robot with face recognition/ Henry T. Martinez, Claire Angel M. Ruan, Patricia Anne C. Subala, Ma. Niecel Ann R. Tesiorna, and Alyanna R. Topacio.--

Bachelor's thesis

College of Industrial Education.-- Bachelor of engineering technology major in computer engineering technology: Technological University of the Philippines, 2025.

Includes bibliographic references and index.

Medication adherence among elderly individuals continues to be a challenge in healthcare,
especially for caregivers responsible not only for assisting with medication intake but also
for maintaining accurate records. To address this, the researchers developed a medication
assistance robot with facial recognition and a water dispenser, ensuring both secure access
and comfort during medication intake. A web-based application was also to provide
caregivers with real-time updates, patient records, and live video monitoring. The system
was tested for its accuracy based on face recognition using HuskyLens AI camera, and
timeliness of the line tracking system, and real-time live monitoring. The HuskyLens AI
camera accuracy was tested under various conditions such as lighting, distance, and angle.
The test results showed that the HuskyLens AI camera has 89.36% accuracy wherein it can
recognize faces with normal to medium lighting condition, can accurately recognize at a
distance of 50 to 100 cm, and can detect faces at 0° and 15-degrees viewing angles. Testing
of Line Tracking and Real Time Live Monitoring systems across 20 trials showed that Real
Time Live Monitoring achieved precision with an average of 1.10 seconds compared to
Line Tracking’s 2.5 seconds. Real Time Live Monitoring results in approximately twice
the accuracy, with 90% of trials within ±1 second time response compared to Line Tracking
75% within ±2 seconds. The test result showed that Real Time Live Monitoring is more
suitable for high-precision timing applications, though both systems have consistent and
predictable performance. To evaluate the system’s performance, the researchers used the
ISO 25010 quality model, focusing on Functionality, Suitability, Performance Efficiency,
Compatibility, Usability, Reliability, and Maintainability. A total of 30 respondents,
including IT professionals, caregivers, and students, were purposively selected to assess
the prototype. Results showed that the robot performed satisfactorily across all features,
including face recognition, medicine dispensing, line tracking, and real-time monitoring,
although some limitations were observed, particularly in the line tracking system. The
system received a grand weighted mean of 4.23, with a descriptive rating of “Very
Acceptable”.