autoFeeder

This is Clover

My beloved eastern mud turtle. She was given to me as a gift by my girlfriend in 2019 during our senior year of high school. Fast forward to 2026 — I've graduated college and work as a full-time software engineer, spending most of my time working from home. Since graduating I've been taking more and longer trips, and currently my mom handles feeding when I'm away. However, that won't always be the case, and when that time comes I need a solution.

autoFeeder is an automated device that not only drops food on a set schedule, but also records a video when it does and live streams a view of the tank.

Demo

1. System Update

sudo apt update
sudo apt upgrade

2. Camera Setup

Verify camera is detected

dmesg | grep -i imx708

Look for a line like: imx708 10-001a: camera module ID 0x0382 The "fixed dependency cycle" messages are harmless.

Install camera apps

sudo apt install -y rpicam-apps

Test capture

rpicam-jpeg --output test.jpg

Verify the image by pulling it off with rsync.

3. Clone the Repository

sudo apt install git
git clone https://github.com/HansChaudry/autoFeeder.git
cd autoFeeder

4. Python Virtual Environment

python -m venv venv --system-site-packages
. venv/bin/activate

--system-site-packages allows the venv to access system packages like picamera2.

5. Install Dependencies

sudo apt install build-essential python3-dev
pip install --upgrade pip setuptools wheel
pip install -r requirements.txt

If you get a libcap error during pip install:

sudo apt install libcap-devel

6. Install ffmpeg

sudo apt install ffmpeg

7. Configure Environment Variables

nano .env

Ensure the following variables are set:

Variable	Description
HOSTNAME	MQTT broker hostname or IP
PORT	MQTT broker port
inTopic	MQTT topic for incoming commands
streamTopic	MQTT topic for camera stream
streamURL	Full RTSP URL e.g. rtsp://host:port/tankCam

8. MinIO Client Setup

Download the MinIO client binary

curl --progress-bar -L https://dl.min.io/aistor/mc/release/linux-arm64/mc \
  --create-dirs \
  -o ~/aistor-binaries/mc

Make it executable

chmod +x ~/aistor-binaries/mc

Set up an alias for your MinIO/AIStor instance

mc alias set hansololabMinio <url> <username> <password>

Verify with ~/aistor-binaries/mc alias list

Create the media directory for recordings

mkdir -p /home/clover/autoFeeder/media

9. MediaMTX Streaming

The Pi streams to a MediaMTX instance running on the homelab server via RTSP. MediaMTX then serves the stream over HLS for the client app.

HLS playback URL (client)

http://<server-ip>:8888/tankCam/index.m3u8

MediaMTX server config (mediamtx.yml)

logLevel: info
api: yes
apiAddress: :9997
rtsp: yes
rtspAddress: :8554
hls: yes
hlsAlwaysRemux: yes
hlsAddress: :8888
webrtc: yes
webrtcAddress: :8889
rtspTransports: [tcp, udp]
paths:
  tankCam:
    source: publisher

Make sure UFW allows port 8554 (RTSP) and 8888 (HLS) from your local network.

10. n8n Automation & Discord Notifications

Two n8n workflows handle automated feeding and notifications:

Workflow 1 — Scheduled Feeding

• Cron trigger → publish MQTT message to inTopic
• Discord notification: "Feeding triggered"

Workflow 2 — Feeding Confirmation

• MQTT trigger on inTopic/done
• Discord notification: "Feeding complete"

11. Run the Application

python main.py

Expected output

Press CTRL+C to exit...
Connected to broker!
Message received on topic: <inTopic>
feeding
Stream started successfully
Recording started successfully
...Feeding.mp4: ████████████████████████████████ 100.0%
Stopping stream. PID: XXXX
Stream stopped successfully
Feeding done published

Architecture Overview

n8n (schedule feed trigger)
    → MQTT publish → Pi Zero 2W
                        → stepper motor rotates drum → food drops
                        → rpicam-vid + ffmpeg → RTSP → MediaMTX
                        → ffmpeg records from RTSP → lastFeeding.mp4
                        → mc uploads to MinIO
                        → MQTT publish done message
                            → n8n → Discord notification
Client App (Expo)
    → MQTT subscribe (start stream)
    → HLS stream from MediaMTX
    → Display the recording from the last feeding