MoonCar (AMB82 MINI) New

What this kit teaches

Course outline

01 Basics (2 class periods)

Hardware setup, pin mapping, motor control, OLED and buttons, and battery charging. Read the tutorial

02 Data collection (2 class periods)

Setting up the track and signs, RTSP streaming and photo-capture code, and taking training photos. Read the tutorial

03 Data labeling (3 class periods)

Labeling four target types in Roboflow, splitting the dataset, augmenting data, and getting the download code. Read the tutorial

04 Model training and conversion (2 class periods)

Training YOLOv7-tiny on Google Colab, understanding mAP, and converting the model for Realtek. Read the tutorial

05 Autonomous driving (3 class periods)

Putting the model on the SD card, line following with a P controller, sign triggers with distance filtering, and adjustable parameters. Read the tutorial

Kit contents

• MoonCar (A+) bare car, 18650 version × 1 set: 18650 battery holder, micro metal gear motor × 2, RGB LED, buzzer, color sensor, infrared reflective sensor × 2, infrared transceiver, and ultrasonic distance sensor
• Realtek AMB82 MINI development board × 1: RTL8735B (built-in NPU), HD camera, and dual-band Wi-Fi
• AMB82:bit breakout expansion board × 1: 1.3-inch OLED (128×64, I2C), A/B buttons, and a micro:bit edge connector slot

18650 Li-ion battery and extra connecting cables are not included; actual contents may vary at the time of shipment. A battery can be added: Sanyo 18650 Li-ion Battery 2550 mAh.

Already have a MoonCar?

If your school already has the MoonCar (micro:bit) or the MoonCar (LinkIt), you don’t need to buy a whole new car. Add the AMB82 Breakout expansion board (SKU 368030502162) and the Realtek AMB82-mini AI Camera development board (368030501864), then plug them into your existing MoonCar to follow this course.

The expansion board does not include the AMB82 MINI development board, a Li-ion battery, or connecting cables. Please purchase the development board separately.

What your school needs to provide

• 18650 Li-ion battery × 1: the kit does not include a battery; you can add Sanyo 18650 Li-ion Battery 2550 mAh (368090500255)
• Computer: one that can install the Arduino IDE, use a Google account (Colab training), and run a web browser (Roboflow labeling)
• MicroSD card, 8–32 GB (FAT32), and a card reader: to store the captured photos and the model
• Internet-connected 2.4 GHz Wi-Fi: for the RTSP live view and program connections
• VLC player: to watch the live view from the car’s camera
• Track materials: two large sheets of black poster paper and white and yellow electrical tape, or a floor that tape can be applied to
• Traffic signs: print the sign.pdf provided with the tutorial in color and attach it to boxes or cardboard that can stand upright

FAQ

Can I train the model without an NVIDIA graphics card?

Yes. The tutorial uses the free T4 GPU on Google Colab, so with a Google account you can train within the usage quota.

Does the kit include a battery?

No. The kit is the bare car, 18650 version, and includes only the battery holder. You can add Sanyo 18650 Li-ion Battery 2550 mAh (368090500255); the micro USB port on the car can charge it directly.

Can I upgrade my existing MoonCar?

Yes. Add the AMB82 Breakout expansion board (368030502162) and the Realtek AMB82-mini AI Camera development board (368030501864), then plug them into your existing MoonCar.

How many photos do I need to take?

About 50–100 photos each of the road and the signs are enough. The key is to vary the angle, position, or background in every photo.

What if the car doesn’t run steadily?

Unit 5 includes a complete guide to the adjustable parameters, so you can tune the speed, steering gain, and sign trigger distance for your track.