Using llama 3.2 as an interface to a robot. If you can get the latency down, it works wonderfully