Edge AI and Sensor Fusion for Audio-Based Human Robot Interaction in Elderly Care

Abstract

This thesis looks at how to design and build an Edge AI system that uses sensor fusion to support audio-based communication between humans and robots in elderly care. It combines speech processing with data from several sensors so that the robot can understand spoken commands and notice emergency situations in real time. Instead of sending data to the cloud, the work focuses on running machine learning models directly on small, low-power microcontrollers. In this way, it shows that advanced AI features can still work well on simple embedded devices with limited resources. The integration of edge computing, audio recognition, and sensor fusion aims to create accessible, responsive, and assistive technologies that enhance the safety, autonomy, and well-being of older adults in their daily lives.