src/control_backend/agents/communication/ri_communication_agent.py

@@ -10,7 +10,7 @@ from control_backend.agents.actuation.robot_gesture_agent import RobotGestureAge
 from control_backend.core.config import settings
 
 from ..actuation.robot_speech_agent import RobotSpeechAgent
-from ..perception import VADAgent
+from ..perception import FacePerceptionAgent, VADAgent
 
 
 class RICommunicationAgent(BaseAgent):
@@ -201,6 +201,13 @@ class RICommunicationAgent(BaseAgent):
                 case "audio":
                     vad_agent = VADAgent(audio_in_address=addr, audio_in_bind=bind)
                     await vad_agent.start()
+                case "face":
+                    face_agent = FacePerceptionAgent(
+                        settings.agent_settings.face_agent_name,
+                        address=addr,
+                        bind=bind,
+                    )
+                    await face_agent.start()
                 case _:
                     self.logger.warning("Unhandled negotiation id: %s", id)
 

src/control_backend/agents/perception/face_rec_agent

@@ -0,0 +1,53 @@
+import asyncio
+import zmq
+import zmq.asyncio as azmq
+
+from control_backend.agents import BaseAgent
+from control_backend.core.config import settings
+
+
+class FacePerceptionAgent(BaseAgent):
+    """
+    Receives and processes face detection events from Pepper.
+    """
+
+    def __init__(self, name, address, bind=False):
+        super().__init__(name)
+        self._address = address
+        self._bind = bind
+        self._socket: azmq.Socket | None = None
+
+    async def setup(self):
+        self.logger.info("Starting FacePerceptionAgent")
+
+        ctx = azmq.Context.instance()
+        self._socket = ctx.socket(zmq.SUB)
+        self._socket.setsockopt_string(zmq.SUBSCRIBE, "")
+
+        if self._bind:
+            self._socket.bind(self._address)
+        else:
+            self._socket.connect(self._address)
+
+        self.add_behavior(self._listen_loop())
+
+    async def _listen_loop(self):
+        while self._running:
+            try:
+                msg = await self._socket.recv_json()
+                await self._process_face_data(msg)
+            except Exception:
+                self.logger.exception("Error receiving face data")
+
+    async def _process_face_data(self, data: dict):
+        """
+        Central place to handle face perception.
+        """
+        face_count = data.get("face_count", 0)
+
+        if face_count > 0:
+            self.logger.debug("Detected %d face(s)", face_count)
+
+            #Post belief
+        else:
+            self.logger.debug("No faces detected")