Here is a humanoid robot with a custom local LLM and face tracking that can maintain natural conversations. It has a custom LLM that is mistral 7B-derived plus a Whisper derived ASR. It has a customizable personality and voice. It can speak in over 20 languages. This robot can be remote controlled for telepresence applications.
Captcha uses a front-mounted camera for face tracking. It can use a handheld microphone in noisy environments but can also use a beamforming microphone array in quieter areas. It has 3 degrees of freedom in the torso and 3 more in the head. It can be used in seated or standing position.
[HT]