Hey everyone! I'm working on a cool Tamagotchi-style project using an ESP32, where I can interact with my virtual pet through a local web server. I'm curious about the possibility of incorporating a tiny AI model into it so that my pet can 'learn' and modify its behavior based on how I interact—like when I feed it or if I ignore it. I realize that the ESP32 has limitations in terms of memory and power, but I've heard about options like TinyML and TensorFlow Lite Micro. Has anyone attempted to do something similar? I'd love to get ideas, links, or examples to help me get started. Thanks a ton! Check out my project here: [TamaPetchi](https://cyberxcyborg.github.io/ESP32-TamaPetchi/)
2 Answers
Just curious, what programming language are you using for this project? It can help tailor the suggestions for integrating AI.
That's a neat idea! The level of "smartness" you need really depends on the interactions you're aiming for. Think about something like the original Furby; it didn't truly learn but reacted based on how much you interacted with it. You could keep track of user interactions and make adjustments to its behavior based on thresholds.
For sure! I’d suggest digging into branching behavior trees or maybe even considering probabilistic state changes to add some unpredictability, all while keeping it lightweight for the ESP32.
I'm actually using C++ for the Arduino/ESP32 environment. Running on the ESP32 microcontroller with the Arduino core!