Autonomous robot for medication delivery in hospitals using YOLOv5 object detection

This study presents the development of an autonomous robot for delivering medicine boxes in hospital environments by recognising visual numeric signs and navigating through predefined indoor waypoints. The robot uses YOLOv5 to detect digit labels (e.g., “1”, “2”, “3”) representing patient room numbers. Upon detecting a digit, the robot follows a sequence: navigating to the medicine pickup point, simulating the collection, and then proceeding to the corresponding patient room. The system is built using ROS2, a camera for visual input, and mecanum wheels to enable omnidirectional movement in narrow hospital corridors. Testing was conducted in a structured environment resembling a hospital layout. The robot successfully detected all target objects—room digits and the medicine box—with 100% accuracy as shown in the YOLOv5 confusion matrix. It was able to navigate from its initial position to the pickup location and then to Room 1, 2, or 3 based on the detected digit. The integration of object detection and waypoint-based navigation worked effectively, ensuring accurate delivery according to predefined routines.