I finally started to get used to editing, shooting, etc a little. So, we can move on to more complex robotic projects.
At the same time, I want to start a new section dedicated to technical vision, machine learning and other funny moments.
I'll start with some basic things, like the movement of a robot along a line, and finish somewhere on neural networks and fuzzy logic.

This video is introductory. In it I'll talk about how to make a simple robot with a camera, with the help of which we will continue to test vision algorithms. The most part of the video is about setting up a single board computer (orange pi, in my case), as well as installing the OpenCV vision library.

By itself, the robot can also be used as a regular car with a remote control from a smartphone.

Links and materials from the video:
Project repository (there is a python script to run on orange, a sketch for arduino and a small description of the project): https://github.com/TrashRobotics/CVbot
Orange pi site: http://www.orangepi.org/
Armbian site: https://www.armbian.com/
SD card loader: https://www.balena.io/etcher/
Site with pytty: https://www.putty.org/

The main parts of the robot:
Orange Pi Zero 512MB;
Webcam;
usb splitter (or logic level converter);
usbttl converter;
XL4005 dc/dc converter;
Arduino Pro mini
MX1508 motor driver;
4x yellow arduino TT gear motors;
2x 18650 batteries and battery compartment for them;
toggle switch;
board or piece of plywood;