Waveshare Jetbot Complete Kit

Waveshare JetBot Kit

Updated: Aug 19 2020

Nvidia's Jetson Nano made a bit of storm when it was first released. Here was an affordable GPU computer that let you do AI processing on the Edge. The Jetbot kit was designed to take advantage of the Nano’s capabilities, letting aspiring roboticists build and experiment with autonomous AI robots.

The open source design for Jetbot is available on Github, and anyone can source the parts themselves, download the software, build the Jetbot and start experimenting.


The Waveshare Kit

Waveshare is one of Nvidia's certified partners for the Jetbot and their kit is one of the most complete, easy to use (no soldering) and well priced packages available.


Whats in the kit

Jetson Nano Developer kit

Wifi module

Camera

Chassis, motors and wheels

Power / Motor board with OLED display

Gamepad

2 power adapters (for Nano & Jetbot)

SD card & card reader


Also provided are all the tools, screws and attachments you need to put this kit together. Waveshare have a detailed assembly guide you can find on their wiki.

Unlike Nvidia's spec, the Waveshare Jetbot has a metal chassis making it slightly sturdier than the 3D printed option. We like how they stuck with the green color scheme of the Jetbot.

The Waveshare Jetbot is powered by 3 X 16340 batteries instead of a mobile battery charger. These batteries are not included, probably due to import restrictions in some countries. Care should be taken when handling 18650s as they need to be charged with a compatible charger and you can't let the voltage in each cell drop below 2.7V. The Waveshare power board takes care of the charging for you, all you have to do is plug it into a socket.

Once assembled, the Waveshare Jetbot feels sturdy, but it is really only designed for AI experiments. The naked frame has no covering or protection for your Jetson Nano, and the wheels are quite basic. Looks wise, its akin to a motherboard on wheels. Don't expect to drive it fast or over rough terrain.

Jetpack & Jetbot Software

Once the hardware is built you will have to install both the Jetpack & Jetbot software onto the Jetson Nano. Nvidia has made this easy by providing images which you can burn onto an SD card, so you don't need to muck around with installing modules yourself.


Installing Requirements

During the initial installation you will need to connect and external display and keyboard to set the wifi network on the Nano. Since the nano has HDMI output it was plugged into the TV and an old USB keyboard was dusted off for this purpose. Once the network is set, you can operate the Nano in ‘headless mode’ and there is no further need for the display or keyboard.

Waveshare provide their own Jetbot image which you need to install instead of Nvidia's image. This custom image is designed to work with their expansion board. One cool thing about Waveshare's image is that it shows you the the remaining charge on the batteries on the OLED display. The installation instructions can be found here.


Programming the Jetbot

When you are ready to get programming, switch the Jetbot on and wait for it to connect to the network. Once connected, you will be able to access the Jetbot's Jupyter notebooks from a browser on your computer.

Jupyter notebooks are interactive webpages which are served from your Jetbot. The great thing about these notebooks is that they function as documentation and programming environment all at the same time. Letting you code and learn all within the browser. The Jetbot notebooks guide you through different topics and AI experiments you can run on the robot. A little knowledge of Python is handy, but you can learn as you go.

Once you have completed the notebooks, you can go further by trying out different AI projects on the Jetson Nano. Nvidia's Hello AI world tutorials are good starting point. There is also a showcase of what other people have built with their Jetson Nanos.


Conclusion

Nvidia have made AI robotic experiments accessible with the Jetbot project. With a little coding know-how, you can purchase one of these kits and build your own AI powered robot at home. The Waveshare kit was well priced, and the fact that you didn't need to do any soldering, with all components supplied was a huge plus for us. If however you would like to build your own robot from scratch, you can find a list of the recommended components on Nvidia's Jetbot wiki.