Parents are the most qualified persons to look after their toddlers. However, there has been instances where the parents were briefly distracted in doing another task and unfortunately, it resulted into the toddlers falling in swimming pools and dying from drowning. This study aims at the design and manufacture of a toddler surveillance robot to reduce the death by drowning in the country.  The robot reads the RSSI strength of a Bluetooth signal to identify when a toddler is alone outside and makes use of Computer Vision to locate and track the toddler from a safe distance. Using Deep Neural Network, the robot was trained to recognize if the toddler is playing near a swimming pool or has fallen in it. Thus, it will automatically activate its alarm buzzers to notify the parents about the dangerous situation. The Buzzer can be turned off using an acknowledgement button on the robot. The robot uses the Rocker Bogie mechanism to easily move over rough terrain and steps found in a backyard. Moreover, the robot is equipped with a Ublox NEO M8N GPS chip to locate its position and before tracking the toddler, it will save its initial GPS location. This initial GPS location is used when the parent comes close to the robot or when the acknowledgement button is pressed, which instructs the robot to return to its initial position. All the above components are controlled using the Raspberry Pi 3 model B with the help of the RaspiCam and OpenCV.

Keywords— Autonomous robot, RSSI, Computer Vison, Deep Neural Network, Rocker Bogie, Ublox NEO M8N, Raspberry Pi 3, OpenCV