This is a small companion server you run on your PC (same machine as Ollama). It gives the robot a stable HTTP target and keeps the on-robot code modular.

Endpoints

POST /v1/chat/text JSON: { "text": "..." } → { "reply": "..." }
POST /v1/chat/audio JSON: { "wav_base64": "..." } → { "reply": "...", "text": "<transcript>" }

Requirements

Python 3.9+
Ollama running locally
- default Ollama chat URL: http://127.0.0.1:11434/api/chat

Optional (for AUDIO mode):

faster-whisper + ffmpeg

Run

From this folder:

python3 server.py --host 0.0.0.0 --port 8020

Environment variables (optional):

OLLAMA_MODEL (default phi3.5)
OLLAMA_URL (default http://127.0.0.1:11434/api/chat)
WHISPER_MODEL (default base)

Note: Ollama can stay bound to 127.0.0.1:11434 on your PC; the robot only talks to this bridge server (:8020).

Install optional STT deps:

pip install faster-whisper
install ffmpeg (platform-specific)

Robot configuration

Open the tunables UI (http://<robot-ip>:3333) and set:

Jibo AI Bridge → Server URL: http://<your-pc-ip>:8020
Jibo AI Bridge → Input:
- AUDIO (records a short WAV clip on the robot and sends it)
- TEXT (uses globalTurnResult ASR text if available)