Lesson 2 — How to Use Teachable Machine

By /

⚠️ IMPORTANT: SAVE YOUR WORK BEFORE YOU CLOSE THE PAGE

Your eLearning lesson status and your project files are saved separately.

The “In Progress” or “Completed” status only records your progress in the eLearning course. It does not save your work in MRT AI Studio or Teachable Machine.

Before closing MRT AI Studio:

  1. Select Save.
  2. Download the MRT AI Studio project file to your computer.
  3. Keep the file in a safe place so that you can open it and continue working later.

After creating and uploading your Teachable Machine model:

  1. Copy and save the model URL separately.
  2. Paste this URL into MRT AI Studio to continue using your model.
  3. Save your Teachable Machine project to Google Drive or download it to your computer if you want to edit or retrain it later.

MRT AI Studio and Teachable Machine do not automatically save your editable project on the server.

If you close the page without downloading your MRT AI Studio project file or saving your Teachable Machine project and model URL, you may need to create the project again.

1.

Click and add the extension.

Step 1

2.

Add the Teachable Machine extension.

Step 2

3.

First, take out the block that allows you to connect a trained model.
Then click the Teachable Machine button, create your own machine learning model, train it, upload it to the Google server, copy the web URL of your model, and paste it into this block.

Step 3

4.

First, practise with an image project.

Step 4

5.

Create three classes: Rock, Paper, and Scissors.
Record images directly using the video camera and create your own model.

Step 5

6.

Make sure the images are clear and accurate.
Use a simple background and good lighting, and take images from different angles.

Step 6

7.

Once you have collected enough images, click the Train button to train the model.

Step 7

8.

Test the model in the preview.
After checking that it works well, click the Export Model button.

Step 8

9.

Click “Upload my model.”
The model will be saved to the Google server.

Step 9

10.

Copy the URL of your model.

🔗 Copy and save your model URL now.
Keep it somewhere safe, then paste it into MRT AI Studio. Also save the editable Teachable Machine project to Google Drive or download it if you may need to retrain it later.

Step 10

11.

Paste the model URL into your block.

Step 11

12.

Add the Camera extension.

Step 12

13.

Code it as shown below and connect your machine learning model.
Check whether it works correctly.

💻 Coding Challenge students: If you are participating in the Coding Challenge only, skip Steps 14–15 and continue at Step 16.

Step 13

🤖 Robotics Challenge students only: Steps 14–15 are only for students participating in the Robotics Challenge. Coding Challenge-only students should skip these steps and continue at Step 16.

14.

You can control a robot in the same way.
First, add the Node Board.

Step 14

15.

Select the Node Board.

Step 15

16.

Now try controlling the robot in the same way.
Because the robot communicates through Bluetooth, there may be a delay of about 0.3 seconds.

💻 Coding Challenge students: If you are participating in the Coding Challenge only, you do not need to connect the red Node Board blocks.

Step 16

💡 Note: Bluetooth may have a ~0.3 second delay. This is normal.

17. 🧠 Use the Forever Loop to Keep Classifying Poses

To control your game with Teachable Machine, your code must continuously analyse and classify poses in real time.
This is why you must use the forever loop with Classify Pose inside — it keeps the camera running and checking your pose every frame.

The sample code below shows the full game control structure — use this as your starting template:

Sample Scratch Code — Full Teachable Machine Game Control

⚠️ Important — You MUST use this loop structure:

Forever → Classify Pose loop — required for continuous pose detection

📌 Why? The forever loop keeps the camera active and repeatedly runs Classify Pose so your game can react to your body movements instantly.
Without this loop, the pose is only checked once and your game will not respond to pose changes.

💡 Tip: Add your game sprites and actions inside the “When a pose is recognised as [Pose]” blocks — one block per pose you trained in Teachable Machine.

18. 🏁 Build Your Own Game & Join the Competition!

Now it’s your turn! Create your own Scratch game controlled by Google Teachable Machine to compete in the challenge.
Use pose recognition to control your game characters or objects — be creative!

🎮 Competition Rules

  • Your Teachable Machine model must recognise at least 3 different poses or gestures.
  • Your Scratch game must be playable and responsive to pose commands.
  • The game must have a clear goal or scoring system.
  • Be creative — design something fun and unique!


OFFICIAL RCC INFORMATION

📤 2026 RCC Project Submission

Learn and prepare here in MRT eLearning. Official rules and all final submissions are managed only through the RCC website.

💻 Coding · 20 October 2026Junior Path A: Presentation Video only. Paths B/C: Presentation Video + SB3 project/code file. Five minutes is recommended.
🤖 Robotics · 25 September 2026One combined Presentation & Robot Operation Video + SB3 code file. Six minutes is recommended.
⏱️ Flexible recordingLonger videos can be uploaded, but judges assess only the first 6 minutes.
📝 No required written reportThe Project Note is optional for other pathways and not required for Junior Path A. Explain the project in the video.
Upload files directly through the RCC Dashboard. Do not submit Google Drive, participant YouTube or GitHub sharing links. School entries are submitted by the teacher/coordinator; personal entries may be submitted by the participant or linked parent/guardian.
View Official Submission Guide Open RCC Submission Dashboard →

RCC policy 2026.5 · Updated 2026-08-20 · Finalised videos are uploaded by RCC as Unlisted and displayed in the RCC participant gallery.

🎤 Presentation Guide

Your presentation should cover the following points (2–3 minutes):

  1. 🎮 Game Design — What is your game about? What is the goal?
  2. 🧠 Teachable Machine Model — What poses/gestures did you train? Why those?
  3. ⚙️ How it Works — Explain how pose recognition connects to your Scratch game
  4. 🏁 Gameplay Demo — Show your game being played with pose controls
  5. 💡 What You Learned — What would you change or improve next time?
🏆

Ready to Compete?

Submit your Scratch game and Teachable Machine model at the RCC portal

Go to RCC Portal →

📊 Scoring Criteria (Total: 100 points)

Category Points
🎨
Creative Design
Creative and original Scratch game design 		20
🧠
Pose ML Model
Accuracy & efficiency of Google Teachable Machine pose model 		20
💻
Code Quality
Clean, efficient Scratch code 		20
🏆
Game / Story
Creativity & overall completeness of game / story 		20
🎥
Presentation
Video presentation 		20
🏅 Total 100

✅ BEFORE YOU FINISH

Please check that you have saved all three items:

☐ MRT AI Studio project file downloaded

☐ Teachable Machine model URL copied and saved

☐ Teachable Machine project saved to Google Drive or downloaded

Your eLearning “In Progress” status does not save these files automatically.

Scroll to Top