Guardian Bot
A ROS-based assistive robot for elderly care, combining autonomous person-following, wearable IoT health monitoring, and automated emergency response.
The problem
Elderly people living alone are at risk from falls and medical emergencies going unnoticed. We wanted to explore whether a mobile robot could act as a companion that watches out for its owner without being intrusive — following them around the home and reacting quickly if something goes wrong.
Building the robot
Guardian Bot is built on ROS and simulated in Gazebo, using OpenCV for real-time person detection and tracking so the robot can autonomously follow its owner through a space. A wearable IoT device streams health data back to the robot, which can trigger an automated emergency response — alerting a caregiver — if it detects a fall or an abnormal reading.
Co-developed with a teammate, the project combines robotics, computer vision, and IoT into a single pipeline: track the person, monitor their vitals, and respond fast when something looks wrong.
Results
In simulation, Guardian Bot achieved 95.2% tracking accuracy and a 92.1% true-positive rate for fall detection, with emergency response triggered in under 5 seconds. The work was later extended into a co-authored research paper on real-time anomaly detection, presented at IEEE CINS 2025 in Dubai.