Scene capture & tracking RealSense RGB-D + YOLO (perception/main.py)

Face recognition InsightFace (embeddings/matching) (perception/main.py, face_recognition.py)

- User tracking (closest person, 3D head position) - On-demand snapshot capture

- User identification (name, user_id, confidence)

MQTT server & command dispatcher (control/NadineServer.py)

Actuation handlers (AgentControlHandler.py)

TTS backends (AzureTTS.py)

Animation library (XML) + motor/facial control (XMLAnimations/*.xml, LipAnimationGenerator.py, SerialComm.py)

MQTT client/wrapper (common/mqtt_comm.py)

STT (Google | local) (interaction/nadine/stt/)

Dialogue Manager session state, language, emotion estimation (agents/dm.py)

Response Generator

Multi-Agent Reasoning (LangGraph) (agents/graph.py)

Search Agent

Vision Agent

Intention Classifier

Memory Retrieval Agent

Memory Update Agent

Multimodal Memory (on-disk) User profiles, faces, embeddings, episodes/scenes (interaction/nadine/db/memory/user_profiles/<user_id>/)

Confirm Identity

Request Clarification

Generate New User ID