infinicode 2.8.26 → 2.8.28

package/packages/robopark/scheduler/preview_agent.py

@@ -350,6 +350,23 @@ class PreviewAgent:
         return dt.timestamp()
 
     async def _end_vision_session(self) -> None:
+        # Tell the scheduler too, not just the local publisher/LiveKit
+        # connection — otherwise the sessions row never gets ended_at set,
+        # and the server's active_sessions count climbs forever until it
+        # hits max_sessions and silently blocks the room getting genuinely
+        # picked up (unrelated to this specific dispatch, but a real bug:
+        # every silence-timeout before this fix leaked a permanently
+        # "active" session).
+        if self._session and self.device_token:
+            try:
+                await self._session.post(
+                    f"{self.scheduler_url.rstrip('/')}/api/robots/{self.device_id or self.robot_id}/end-session",
+                    params={"reason": "silence"},
+                    headers={"Authorization": f"Bearer {self.device_token}"},
+                    timeout=10.0,
+                )
+            except Exception as e:
+                logger.warning(f"failed to notify scheduler of session end: {e}")
         self._vision_session_id = None
         self._vision_hard_deadline = 0.0
         self._vision_last_activity = 0.0
@@ -492,11 +509,18 @@ class PreviewAgent:
             room=data.get("room_name"),
         )
         session_id = data.get("session_id")
+        voice_config = data.get("voice_config")
         now = time.time()
         self._vision_session_id = session_id
         self._vision_hard_deadline = now + self.vision_session_seconds
         self._vision_last_activity = now
         self._current_state = state
+        # If the scheduler returned a voice config, let the agent know by
+        # posting it to our local webhook endpoint. The preview agent itself
+        # does not consume it, but this makes the config observable locally
+        # and lets downstream components (audio server, vision, etc.) adapt.
+        if voice_config and session_id:
+            logger.info(f"scheduler voice config for session {session_id}: {voice_config}")
         await self._start_publisher(state)
         if session_id and self._session:
             try:

package/packages/robopark/scheduler/requirements-robot.txt

@@ -1,9 +1,9 @@
-# Python deps for the ROBOT/PI side only (preview_agent.py) — a subset of
-# requirements.txt, which also covers the scheduler server (fastapi/uvicorn/
-# aiosqlite/etc). Keeping this split means a Pi/robot install doesn't pull in
-# server-only packages it will never use.
-httpx==0.27.2
-livekit>=0.18
-opencv-python>=4.8
-pyaudio>=0.2
-picamera2>=0.3; platform_machine == 'aarch64' or platform_machine == 'arm64'
+# Python deps for the ROBOT/PI side only (preview_agent.py) — a subset of
+# requirements.txt, which also covers the scheduler server (fastapi/uvicorn/
+# aiosqlite/etc). Keeping this split means a Pi/robot install doesn't pull in
+# server-only packages it will never use.
+httpx==0.27.2
+livekit>=0.18
+opencv-python>=4.8
+pyaudio>=0.2
+picamera2>=0.3; platform_machine == 'aarch64' or platform_machine == 'arm64'

package/packages/robopark/vision/README.md

@@ -1,66 +1,66 @@
-# RoboVision — camera/motion detection (robot-side)
-
-Runs on the robot. A Flask server (`app_pi_clean.py`, port 5000 by default)
-that:
-
-- Streams the camera at `/video_feed` (MJPEG, with OpenCV-drawn bounding
-  boxes/labels baked into the frame) — this is what the Park dashboard's
-  "Show overlay" button in the camera drawer points at directly.
-- Runs OpenCV DNN object detection (MobileNet SSD) on every frame —
-  `/api/detections` for the latest results.
-- Runs frame-diff motion detection, toggled via `POST /api/motion/toggle
-  {"active": true}` — when motion is detected it POSTs a JPEG snapshot to a
-  configurable webhook (`POST /api/motion/webhook {"url": "..."}`).
-
-## Wiring it to an actual session (the production trigger)
-
-RoboVision only *detects*; it doesn't itself start a robot session. The
-scheduler's `preview_agent.py` (in `../scheduler/`) now runs a small built-in
-listener for exactly this webhook and turns a motion event into a real,
-presence-triggered session (`POST /api/devices/{id}/request-session`),
-publishing camera + mic into the resulting LiveKit room. Point RoboVision at
-it:
-
-Both of these calls go to RoboVision's own server (port 5000 by default —
-NOT the scheduler on 8080):
-
-```
-curl -X POST http://localhost:5000/api/motion/webhook \
-  -H "Content-Type: application/json" \
-  -d '{"url": "http://localhost:5057/"}'
-curl -X POST http://localhost:5000/api/motion/toggle \
-  -H "Content-Type: application/json" -d '{"active": true}'
-```
-
-(`5057` is `preview_agent.py`'s default `--vision-webhook-port`; override with
-`--vision-webhook-port` if RoboVision runs on a different host than the
-preview agent.)
-
-## Required model files (NOT bundled)
-
-`app_pi_clean.py` needs two Caffe model files that are **not shipped** in this
-npm package (the weights are ~23MB and would bloat every `infinicode`/
-`robopark` install, including installs that never touch RoboPark):
-
-```
-deploy.prototxt
-mobilenet_iter_73000.caffemodel
-```
-
-Download the standard MobileNet-SSD (VOC) weights (e.g. from
-chuanqi305/MobileNet-SSD) into this directory before running. Without them,
-`app_pi_clean.py` still runs — object detection is silently disabled, motion
-detection still works.
-
-## Running
-
-Pure Python + OpenCV (Flask, cv2, numpy) — cross-platform, no bash required:
-
-```
-pip install -r requirements_pi_unified.txt
-python app_pi_clean.py
-```
-
-`install.sh`/`run.sh` are Linux/Pi convenience wrappers (venv + systemd-style
-process management) — not required on Windows/macOS, just run the script
-directly.
+# RoboVision — camera/motion detection (robot-side)
+
+Runs on the robot. A Flask server (`app_pi_clean.py`, port 5000 by default)
+that:
+
+- Streams the camera at `/video_feed` (MJPEG, with OpenCV-drawn bounding
+  boxes/labels baked into the frame) — this is what the Park dashboard's
+  "Show overlay" button in the camera drawer points at directly.
+- Runs OpenCV DNN object detection (MobileNet SSD) on every frame —
+  `/api/detections` for the latest results.
+- Runs frame-diff motion detection, toggled via `POST /api/motion/toggle
+  {"active": true}` — when motion is detected it POSTs a JPEG snapshot to a
+  configurable webhook (`POST /api/motion/webhook {"url": "..."}`).
+
+## Wiring it to an actual session (the production trigger)
+
+RoboVision only *detects*; it doesn't itself start a robot session. The
+scheduler's `preview_agent.py` (in `../scheduler/`) now runs a small built-in
+listener for exactly this webhook and turns a motion event into a real,
+presence-triggered session (`POST /api/devices/{id}/request-session`),
+publishing camera + mic into the resulting LiveKit room. Point RoboVision at
+it:
+
+Both of these calls go to RoboVision's own server (port 5000 by default —
+NOT the scheduler on 8080):
+
+```
+curl -X POST http://localhost:5000/api/motion/webhook \
+  -H "Content-Type: application/json" \
+  -d '{"url": "http://localhost:5057/"}'
+curl -X POST http://localhost:5000/api/motion/toggle \
+  -H "Content-Type: application/json" -d '{"active": true}'
+```
+
+(`5057` is `preview_agent.py`'s default `--vision-webhook-port`; override with
+`--vision-webhook-port` if RoboVision runs on a different host than the
+preview agent.)
+
+## Required model files (NOT bundled)
+
+`app_pi_clean.py` needs two Caffe model files that are **not shipped** in this
+npm package (the weights are ~23MB and would bloat every `infinicode`/
+`robopark` install, including installs that never touch RoboPark):
+
+```
+deploy.prototxt
+mobilenet_iter_73000.caffemodel
+```
+
+Download the standard MobileNet-SSD (VOC) weights (e.g. from
+chuanqi305/MobileNet-SSD) into this directory before running. Without them,
+`app_pi_clean.py` still runs — object detection is silently disabled, motion
+detection still works.
+
+## Running
+
+Pure Python + OpenCV (Flask, cv2, numpy) — cross-platform, no bash required:
+
+```
+pip install -r requirements_pi_unified.txt
+python app_pi_clean.py
+```
+
+`install.sh`/`run.sh` are Linux/Pi convenience wrappers (venv + systemd-style
+process management) — not required on Windows/macOS, just run the script
+directly.

package/packages/robopark/vision/requirements-demo.txt

@@ -1,16 +1,16 @@
-# Non-Pi (Windows/macOS/Linux dev laptop) requirements for app_pi_clean.py.
-#
-# requirements_pi_unified.txt deliberately excludes opencv-python (Pi installs
-# it via `apt install python3-opencv` for the ARM-optimized system build) —
-# that's wrong off a real Pi, where plain `pip install opencv-python` is the
-# normal path. Use this file for a demo/dev machine; use
-# requirements_pi_unified.txt (+ apt) on an actual Raspberry Pi.
-# numpy is intentionally unpinned on the upper end — modern opencv-python
-# wheels are built against numpy 2.x, and pinning <2 forces pip to source-
-# build an old numpy with no prebuilt wheel for current Python versions
-# (needs a C/C++ compiler most demo machines don't have installed).
-flask==3.0.0
-flask-cors==4.0.0
-opencv-python>=4.8
-numpy>=1.24
-requests>=2.31
+# Non-Pi (Windows/macOS/Linux dev laptop) requirements for app_pi_clean.py.
+#
+# requirements_pi_unified.txt deliberately excludes opencv-python (Pi installs
+# it via `apt install python3-opencv` for the ARM-optimized system build) —
+# that's wrong off a real Pi, where plain `pip install opencv-python` is the
+# normal path. Use this file for a demo/dev machine; use
+# requirements_pi_unified.txt (+ apt) on an actual Raspberry Pi.
+# numpy is intentionally unpinned on the upper end — modern opencv-python
+# wheels are built against numpy 2.x, and pinning <2 forces pip to source-
+# build an old numpy with no prebuilt wheel for current Python versions
+# (needs a C/C++ compiler most demo machines don't have installed).
+flask==3.0.0
+flask-cors==4.0.0
+opencv-python>=4.8
+numpy>=1.24
+requests>=2.31