If you’re trying to build a wheeled robot inexpensively or want to test out some algorithms, you will need a powerful microprocessor that is easy to use. The Nvidia Jetson, Raspberry Pi, and Arduino boards all are great for this purpose, but each vary slightly.
The Jetson Nano is our favorite computer due to its low cost, size, and functionality. It only costs $99 and can do anything an Ubuntu 18.04 computer can do. It has almost a gigabyte of CPU for medium level algorithms and enough GPU cores for processing camera data quickly or inference from Deep Learning networks. The main issues are that powering it without a wall outlet can be challenging, but we figured this out, and the community hasn’t had the time to answer most general questions quite yet. Go with the Nano if you are moderately familiar with robotics and want the best deal.
The Jetson TX2 is more expensive than the rest, but is very powerful. It has much more CPU, GPU, RAM, but disk space is limited to 32 GB; so you will need a USB flash drive to add more space. Go with the TX2 if you’re an expert at robotics and want to meet project requirements robustly.
The Raspberry Pi is probably the most famous microprocessor for beginners in robotics, as it has a large user base with many tutorials. Installing the OS is a bit more challenging, but afterwards, various hardware options makes it easy to fit it to your specific robot. Go with the Raspberry Pi if you’re moderately familiar with robotics but want to minimize costs.
The Arduino was one of the first microprocessors simplified for hobbyists but is still one of the cheapest and easiest ways to get started. Although it doesn’t run an operating system like Linux, you can easily create Arduino code to blink LED lights, learn circuitry, and even power motors for a robotic chassis. Go with the Arduino if you’re new to robotics.
The Arduino is for beginners, the Raspberry Pi and Jetson Nano are for mid-levels, and the Jetson TX2 is for experts. All mobile robots generally need Wi-Fi, enough processing, and a simple way to code or find help, and any of the boards here offer these. With a deeper understanding of engineering, more complex boards are available, too, for further customization.