Code for miniature TARS from Interstellar! WIP => working full voice command functionality w/ language model running onboard! The Vosk branch currently has the best response time (~3-5 seconds for stt and request, 3-4 seconds for tts). Vosk branch also has code for camera control so TARS can see and analyze the world (running gemma3 model instead of llama3.2 for vlm capabilities).
- Credits to Charlie Diaz for original design
- As well as the TARS-AI community for general debugging help (TARS-AI)