Autonomous Rescue Boat using AI

This paper provides an overview of the development of an Artificial Intelligence (AI) driven Automated Rescue Boat (ARB), capable of performing more reliable and efficient rescue operations for urban flooding events. Each year, drowning is responsible for a significant number of accidental deaths globally, primarily in areas where there is insufficient and immediate access to an effective rescue process, standard lifebuoys and manual rescue processes are constrained by human effort which can lead to delays in rescue responses and potentially an extremely hazardous situation for rescuers undertaking the rescue efforts. To mitigate these issues, we propose the Automated Intelligent Lifesaving Buoy, an autonomous rescue boat hybrid, equipped with a camera that can detect, monitor and facilitate rescue actions to drowning victims in a real time situation. This is the opposite of a conventional rescue operational systems such as lifebuoys that are overly reliant on person visibility, and an operator who will need to remote control and act with precision under pressure . We apply computer vision, object detection and victim's face recognition using Yolov8, Tensor Flow Lite, Py Torch, OpenCV. The ARB is built using a Raspberry Pi including BLDC motors and Electronic Drive Controller (EDC) technology as part of buoy propulsion. The U-Shape design means that a victim will be safely gripped by the buoy, and face detection is based on Local Binary Pattern Histogram (LBPH) which increases certainty of the victim, even in low visibility and inclement conditions. In our tests, the ARB achieved 95% obstacle detection, a 92% identification rate of a drowning victim and successfully completed 15-meter rescues in an average of just 30 seconds using real time analysis