Updated: Jan 22, 2021
Artificial intelligence or A.I. is all the rage these days. Everyone wants to know more about it and apply it to their work and personal projects. However, learning A.I. can be tedious, especially for younger kids. There are many products on the market that want to solve this problem. On the one end, many A.I. hardware kits offer instructions for building vehicles/robots and provide some limited programming ability. They are great for introductory projects but seldomly provide enough depth to keep people interested longer than 2-3 days. On the other hand, there are many university level tutorials and online courses. They are excellent sources of technical information and give people a clear look at what's under the hood of these A.I. algorithms. Still, they can be overwhelming for beginners and casual enthusiasts.
We are pleased to introduce our open source project, the Cortic A.I. Toolkit (or CAIT), which aims at providing a gentle introduction to A.I. for beginners while offering a lot of customization possibilities and depth so that users can progressively grow with it. The best part about using CAIT is that there is no need to buy an expensive computer. Our deep learning algorithms work really well on the $35 Raspberry Pi 4B. Because of the open hardware design of the RPi, it is compatible with literally hundreds of off-the-shelf hardware components, ranging from different sensors, various motors to even LiDARs. There is just an endless combination of things that the users can build with it.
For Beginner, setting up CAIT is very simple. Just follow a short screencast after cloning or unpacking our release from Github. After a few quick configuration steps, the user is greeted with our visual programming environment shown below.
There are many artificial intelligence blocks above that users can experiment with right away. We also offer a simple programming guide and several tutorials to get people started.
As users gain more experience using these blocks or with programming in general, they may want to have more power and flexibility in building their projects. At this point, they can just click on the Python icons on the toolbar to translate the current visual program into Python code and then continue to build their project from a code editor like Microsoft VSCode shown below.
Alternatively, the user may want to explore how a Python program works. This can easily be done by importing the code into Jupyter notebook using the link http://<hostname>.local/8000 as shown.
Our goal is to minimize friction that people experience when they try to learn A.I. The CAIT project eliminated the need to manually install numerous software packages just to get a simple A.I. example program running. It allows users to instantly dive into the most enjoyable part of learning and building with A.I.
We are very excited to put out this initial release of CAIT and eager to hear from our community about how people are using it.