How to design a neural network?

Question

Neural networks

Is it possible to design such a network with a teacher, which, for example, will learn to determine the angle at which the line is located in the picture? To be more precise - to determine the angle on which car you need to turn. It is possible to scale 180 degrees conventionally by [-1:1] gap as an output, but will she be able to learn to determine the angle normally, that is, those angles that were not in training.

Answer 1

You have multiple tasks mixed up.
The first is to define the "line in the picture". I don’t know what kind of picture you have, I don’t know what you mean by “line”. but if this is, for example, a photo, and the line is an element of some design, then the solution to the problem lies in the recognition of objects in the picture. Today, neural networks solve such problems quite effectively.
And if your task is to "determine the angle on which car you need to turn," then, generally speaking, the neural network has nothing to do with it. There is a direction of movement, there is a new direction. The "definition" of an angle is the definition of the difference between the two directions. What else needs to be determined there, especially with the help of a neural network ???
Well, the sentence "scale (???) 180 degrees conditionally (???) by [-1: 1] interval (???) as an output (???)" is generally a set of some loosely connected between words. If you meant "display 180 degrees in the interval [-1:1]" - then yes, it's elementary to do. But why, and what does this have to do with the previous two questions?