ROS Bionic Dog Face Recognition

I’m building an AI-powered Hiwonder bionic dog on a Raspberry Pi 5 running both ROS 1 and ROS 2 in Python. I now need a focused add-on that lets the robot recognise faces, decide whether the person is an authorised student or an intruder, and act immediately.

Core goals
• Face recognition module – choose the most suitable library (OpenCV, Dlib, TensorFlow or another you prefer) and train it with my student dataset.
• Decision logic – label each detection as “authorised” or “intruder”.
• Alert workflow – when an intruder is detected the robot must send an SMS text message to the proctor with the person’s status and the robot’s live location. GPS, Wi-Fi positioning, or other lightweight techniques are all acceptable as long as they return coordinates quickly.
• ROS integration – publish events on a dedicated topic so other nodes can subscribe, and provide a simple launch file for easy start-up.
• Validation – demonstrate live on the robot or through a recorded session that the system meets the above criteria.

This is a self-contained, functional prototype: clean code, clear setup instructions, and any training scripts or models must be included so I can retrain later if needed.

If you love robotics, Python and ROS, let’s make this bionic dog smarter and safer.