Digital Twin Greenhouse with IoT sensors

Background

When it comes to discussions on digital transformation, these are the words that catch our sight a lot: IoT, Industrial IoT, AI, Big Data, and many more, among which I think the term “Digital Twin” is catching up a lot and getting my attention as well.

According to Wikipedia, “A digital twin is a real-time virtual representation of a real-world physical system or process (a physical twin) that serves as the indistinguishable digital counterpart of it for practical purposes, such as system simulation, integration, testing, monitoring, and maintenance.” It’s like creating a highly complex virtual model that is the exact counterpart (or twin) of a physical thing. The physical thing could be a house, a bridge, a school, or anything around us.

And this is exactly the inspiration for my project when I narrow the "physical thing” down to a greenhouse. How about building up a digital model of a greenhouse, where I can map out the data that I collect in a physical garden using some sensors? In this way, anyone looking at the digital twin greenhouse can now see crucial information about how the plants and trees in the physical garden are doing out there in the real world.

So, here is the Digital Twin Greenhouse that I built for a couple of months. It’s a digital model of a greenhouse, where I “planted” a bed of tomatoes (my favorite vegetable). It’s also an intelligent greenhouse powered by a sensors monitoring system and a gateway. The devices in the virtual model represent the 5 Seeed Studio SenseCAP LoRaWAN Sensors and Gateway that we deployed in our office rooftop garden. All the data collected in the garden (physical world) is mapped onto the screen dashboard of the virtual model. When you click the sensor devices in the model, it will automatically lead you to read the detailed data collected by the physical sensor in real time!

p.s. This project is now live on a Smart Village Solution Page. Visit here to experience it by yourself.

Want to learn how I built this metaverse project? Read the following step-by-step tutorial.

How I Built the Digital Twin Greenhouse

1.Modeling - Greenhouse

This part is mainly 3d modeling if you are familiar with any 3d software, such as 3ds Max, Maya, C4D, Unity, Blender, etc. You can build your own digital virtual scene.

I found some reference pictures of greenhouses, to make them look more like a "twin" of an actual greenhouse, and built it for growing tomatoes.

Greenhouse farming is the unique farm practice of growing crops within sheltered structures covered by a transparent, or partially transparent, material.

I also learned that tomato is a cash crop, the most common greenhouse crop, it can also be quite prolific and, thus, profitable.

• Modeling with Autodesk Maya.

2. Modeling - Sensors

• Determine the actual deployment of sensors on the scene, and place the corresponding industrial model of the equipment at each location in the scene to show the function of each sensor as much as possible.

all the sensor models are available on Sketchfab. DOWNLOAD.

• Export models to fbx format, convert fbx to gltf/glb format. 🔨: 3d-convert

3. Deploy devices

In the process of making it, it occurred to me that perhaps importing real-time data would be a good idea.

I communicated this idea with the engineers of Seeed and got their 100% support. As we had no public shed greenhouse-related data available, they even deployed a set of SenseCap sensors in the rooftop garden.

Hardware components

S2101 - Air Temperature and Humidity Sensor

S2102 - Light Intensity Sensor

S2103 - CO2, Temperature, and Humidity Sensor

S2104 - Soil Moisture and Temperature Sensor

S2105 - Soil Moisture, Temperature, and EC Sensor

Note:

The Air Temperature and Humidity Sensor (S2101) and CO2, Temperature, and Humidity Sensor(S2103) should be installed with the probe facing downwards.

The Light Intensity Sensor (S2102) should be installed with the probe facing upwards, and the Soil Temperature and Humidity (conductivity) node (S2104 & S2105) should be installed with the probe inserted horizontally near the plant root.

4. Interactive Part

Based on what we have above, we are now ready for the interactive part.

As we wanted to display interactive 3D models on the web, I found a friendly web component that makes rendering interactive 3D models - Model Viewer.

• Upload models to Modle Viewer for white model testing on the website.
• Give model material texture mapping on the model viewer and adjust the color, roughness, shading, and other material properties respectively.

• Add Hotspots to make models interactive.

This page showcases how you can add hotspots to your scene.

5.Realtime Data - API

Seeed Studio has a complete instruction document on how to make your first HTTP API call to SenseCAP HTTP API.

https://sensecap-docs.seeed.cc/httpapi_quickstart.html

6.Optimize

The web page loading model took 30 seconds. (Super Slow loading😢)

Solution

There are two downsize spaces in the production process:

1. Before fbx file convert to glTF file, delete part of the texture and models

2. After gltf converting, compressed it into a smaller gltf file. 🔨:gltf-pipeline

Finally, we compressed the model from 150M - 7M. 😎

Final Result 👉: Smart Village Solution Page.

Visit here to experience it!

GitHub

We open-sourced this project on GitHub.

and all the sensor models are available on Sketchfab. DOWNLOAD.

Feel free to contact me & leave your comments 🧡！