AI Plant Guardian — The Smart Plant That Waters Itself, Sees with AI, and Talks to You on Telegram

LR7843 modulo FET

1,28 Pollici LCD TFT Display

USB Connector Type-C Female

Soil Moisture Measurement Sensor

YWBL-WH Modulo Telecamera USB OV5640

5 Ports USB C Hub to 4K HDMI

 Jumper Wire

GY-BME280 Barometric Pressure & Temp/Humidity Sensor

WS2812 5050 5V RGB 

AZDelivery 3 x Shield 

DC 5V Small 030 Motor Micro Water Pump

Arduino® UNO™ Q 4GB

Bambu Lab P2S

Arduino App Lab

<html><body><h3>AI Plant Guardian — The Smart Plant That Waters Itself, Sees with AI, and Talks to You on Telegram</h3><h3></h3><figure><img src="https://projects.arduinocontent.cc/7851d30c-4e85-44f7-9394-26f39ebe218f.png"/><figcaption></figcaption></figure><h3>What This Project Does</h3>AI Plant Guardian is an end-to-end smart plant, built from scratch around the built around the Arduino® UNO Q, combining three worlds into a single home-grade device: On-device Artificial Intelligence — A USB webcam observes the leaves in real time and an AI model (exported from Edge Impulse™) classifies them into 5 categories: Healthy, Powdery, Rust, Slug, Spot. Everything runs locally on the UNO Q — no cloud, no internet required. Automatic irrigation — Soil moisture, temperature and ambient humidity sensors monitor the plant 24/7. When it's thirsty, a small pump activates autonomously through a 3D-printed shower head with a pressure-balancing chamber. Remote control via Telegram bot — From anywhere in the world you can ask the plant how it's doing, get a live photo with AI classification, or trigger irrigation manually. The bot is locked to a personal chat ID for security. <blockquote>The project's goal is to demonstrate that, with Arduino® UNO Q, AI workloads previously requiring a server or cloud infrastructure can now run entirely on-device in a home IoT context </blockquote><h3>Watch the Full Tutorial on YouTube</h3>In the video I walk through the entire process — from the pencil sketch on paper to the AI model running on Arduino® : 3D design, sensor selection, PLA/PETG prints, hardware integration, RouterBridge, Edge Impulse™, dashboard and Telegram. Two months of work condensed into 20 minutes. <figure><iframe src="https://youtu.be/vUl222SlBfU"></iframe></figure><h3>Hardware Used</h3> <figure><img src="https://projects.arduinocontent.cc/64e15206-b4b3-4c3f-a35b-7e6722e139ae.png"/><figcaption></figcaption></figure>All links to the materials are available on my <a href="https://t.me/Ingeimaks">Telegram channel </a>organized and ready, no marketplace rabbit holes. <h3>Wiring Diagram</h3><figure><img src="https://projects.arduinocontent.cc/f158f66e-3084-49f5-b77c-7990737e2884.png"/><figcaption></figcaption></figure>Main connections: A0 → Soil moisture sensor (analog) D5 → LR7843 MOSFET gate → 5V pump D6 → WS2812B data-in (24 LEDs) I2C (SDA/SCL) → BME280 @ 0x76 SPI → GC9A01 display USB → Webcam (automatically recognized by Linux) <h3>Design &amp; 3D Structure</h3><figure><img src="https://projects.arduinocontent.cc/706ff387-e931-4c96-b3b7-c5d886bf5c53.jpg"/><figcaption></figcaption></figure>The entire structure is 3D-printed in PLA and PETG on a Bambu Lab® P2S, featuring: Integrated water reservoir in the base (tested for waterproofness: thick walls + 100% infill in critical zones) Balanced irrigation shower head with annular compensation chamber, throttle plate, and final distribution plate Dedicated housings for each sensor, webcam mount, hidden cable routing Special thanks to Attilio (designer) for the support on Blender aesthetic refinements. Download STL = <a href="https://makerworld.com/it/models/2853654-ai-plant-guardian-self-watering-smart-plant#profileId-3183332">https://makerworld.com/it/models/2853654-ai-plant-guardian-self-watering-smart-plant#profileId-3183332 </a><h3>Software Stack</h3>The entire project is managed through Arduino® App Lab, the dedicated IDE for UNO Q, which orchestrates microcontroller and Linux in a single environment via Bricks: <pre>bricks: </pre><pre>	 - arduino:video_image_classification: </pre><pre>	 variables: </pre><pre>	 EI_V_CLASSIFICATION_MODEL: /home/arduino/.arduino-bricks/ei-models/model.eim </pre><pre>	 - arduino:web_ui: {} </pre><h3>RouterBridge — The Communication Core</h3>Linux and STM32 talk to each other via RouterBridge (local RPC between the two brains). The Arduino® sketch exposes 4 methods: <pre>get_snapshot → returns all sensor data </pre><pre>do_water → starts the pump </pre><pre>set_led → switches LED mode </pre><pre>set_ai_alert → notifies the display of an AI alert </pre>The web dashboard, Telegram bot, and AI model all run on Python (Linux side) and call these methods on the microcontroller side. Clean, modular, maintainable. <h3>The AI Model (Edge Impulse™ + PlanTech-AI)</h3>Starting from the open source model PlanTech-AI by AhmAshraf1, the workflow was: 1) Local validation with Streamlit + uv venv 2) .tflite import into Edge Impulse™ Studio 3) Export as .eim for Linux aarch64 (UNO Q's ARM 64-bit CPU) 4) Deploy via scp + chmod +x 5) Integration into the video_image_classification brick <h3>Final Demo</h3><figure><img src="https://projects.arduinocontent.cc/24fa784a-0ed5-4dd0-be3c-607534d5f883.png"/><figcaption></figcaption></figure>On the display, in real time: - Ambient temperature and humidity (BME280) - Soil moisture with colored bar - Plant health status (healthy / thirsty / critical) - AI alert with detected issue type and confidence <figure><img src="https://projects.arduinocontent.cc/55e1e901-1d8a-4352-a990-16699c0bcaf6.png"/><figcaption></figcaption></figure> From the Telegram bot you can: Status → live sensor data Photo + AI → snapshot + automatic classification Water Now → manual pump command </body></html>

Smart Plant Guardian

AI Plant Guardian — The Smart Plant That Waters Itself, Sees with AI, and Talks to You on Telegram

AI Plant Guardian is an end-to-end smart plant, built from scratch around the brand new Arduino UNO Q