Reflexes, Cognition, and Thought

The author is building a droid with a multi-layered brain. The Reflex Layer, prototyped with an Arduino Uno, focuses on basic movement logic. A Visual Odometer was created to demonstrate integer overflow, highlighting the importance of choosing correct data types. Simulated motion was achieved using a photoresistor and light flashes, with distance calculated via the Pythagorean theorem. Connecting motors proved more challenging than anticipated due to unfamiliarity with servo wiring and power requirements. The Cognition Layer utilizes a Raspberry Pi 5 for advanced functions. This layer incorporates computer vision and local AI. A camera was connected to the Pi, and a local Vision Language Model was set up using Ollama. A Python script captures camera frames and sends them to the AI for local analysis. The AI successfully described the room and the author, though processing took about three minutes per frame. Future plans include optimizing processing speed and exploring facial recognition.