How to seriously engage in robotics NOT at the university?

Question

Robotics

I want to start doing serious robotics even before university. That is, not arduino, lego, etc. And create something really cool. What can be done? (I understand that I will have to start small)

Answer 1

No way. It is impossible to do serious robotics alone.
Serious robotics implies a team where programmers, engineer and mechanic are different people with work experience, not to mention APM.
It should be noted that the main thing is programming, programming is the biggest problem in robotics.
Tesla has been trying to program his autopilot for many years, there have been no questions with the mechanics for a long time, but the autopilot code has been written for several years.
You can try to assemble a banal robo-spider that will carry cola from the refrigerator, sweat with the calculation of kinematics and create a mathematical model, sweat with the design of leg mechanics, sweat with navigation and 3D space maps.

Answer 2

Hello. The word "robotics" is rather vague, what meaning do you put into it?
For example, you need to build a robot tank to compete. At a minimum, you need to understand the mechanics, how to count the gear travel, what gearboxes are, you may want to install a differential, how a stepper motor differs from a collector or servo drive, what to use to turn the tower, etc.
Next, you need to do management via the Internet. I had a task on raspberry pi to raise a node js web server and establish a web socket connection with another server. Transfer of control, plus video streaming. This is already pure web development, a huge layer of knowledge, which seems to have no direct relation to robotics, but in this case it was needed.
If I am going to build a drone, an airplane or an airship, then there is also a huge field of knowledge about aeronautics, gases, etc., which seems to be not robotics, but for a specific task you need to figure everything out.
Further, I don’t know what the arduino didn’t like - it’s a microcontroller that performs certain tasks. It is convenient to use it for prototyping, that is, first you build your robot on an arduino, you understand what exactly and how you use it, after which you can make your own board for the same task. In circuitry, there are also many rules of their own, and this is also a huge layer of knowledge. But if you know only circuitry separately, you cannot build a robot.
If you need to make computer vision, then there is also pure programming, which does not apply to mechanics or other areas at all, but in some robo projects it is necessary.
I hope I paint the big picture. You can create something really cool, just surf the Internet, look at existing projects and choose something to your liking, start sawing, while simultaneously swallowing educational videos from YouTube on related topics.
I can recommend the Amperka channel on YouTube, they have a bunch of projects and generally developing infa. In terms of mechanics, even Lego reviews from Cyril Desert Eagle Lego Technic may be enough.

Answer 3

All beginners, driving themselves into a certain robotic platform, under the influence, as it seems to them, of widespread "authoritative opinions", limit their professional perspectives to the hardware architecture under one or another OS, not realizing that the growth of their professional level will largely depend on the tools and hardware architecture for which this toolkit is intended, which will definitely affect the final result in terms of time and cost and other uncertainties at the initial stages of their projects.
Here are the main postulates that it makes sense to follow at the beginning of your path to robotics:
- robosoft works as an intermediate layer between a conventional OS and robot control programs / scripts;
- Robosoft has a modular structure that works on top of the base layer (framework);
— Robosoft has a distributed client-server structure;
“standard” robosoft is:
* a framework that provides typical operations, an abstraction layer from equipment, inter-thread interaction, etc.
* additional modules that implement various robotic algorithms (computer vision, SLAM, machine learning, kinematics)
* simulator (the simulator is an important part of the software for developing robots - it will allow you to debug algorithms in a virtual environment, without buying expensive equipment.)
* shell for visual programming and control of the robot (visual and accessible object-oriented programming, representation of the state of the robot in the form of a hierarchical finite state machine, consisting of a controlled set of final static machines, etc.)
I foresee here the possibility of reasoning regarding the ROS platform, which is actively popularized by most Russian universities, as well as courses with an army of hobi communities of student teams, etc. etc .
For clarity - To the question: "How ROS differs from other X platforms, there is no unequivocal answer, there is an axiom - if you decide to use X, you can still use many of the libraries distributed with ROS. It is important to understand that ROS only works on Unix platforms and is mostly experimental.
The topic has branches depending on what kind of robotics you want to do.
For example , industrial manipulators , logistics systems, biorobots, agrotechnical systems, etc. etc.

Answer 4

You can create a Pilesos robot on an arm microcontroller.

Answer 5

Start with avr, pic microcontrollers. But still I advise you to start with arduino. I am also a beginner. Started learning Arduino.