Neural Netwok for classify 2D map position

Hello all,

I’m new to Neural Networks and my math is secundary school only.

Here is the problem I want to solve using PyTorch:

Im building a robot to explore my house rooms.
I need a way for the robot to identify its 2d map position.

The robot has 2 sensores: front distance sensor (tof) and compasso sensor - which unfortunately I can’t rely on because indoor I have too much magnetic intererence.
So basically I just have the distance sensor and the robot can rotate but I dont know the rotate angle.

Here is my idea:

I built a 2D map, a grid let say 40x40.
Each grid cell holds 8 distances for each 8 angles (0, 45, 90, …, 315 degrees) to neighborg cells.

The robot Will make 5 distance readings, ex:
60, 75, 60, 75, 90

So, 2D map Will have this data:
Cell 0: 30, 45, 60, 75, 90, 75, 60, 45
Cell 1: etc…

I want to Neural Network to learn the 2D map, and based on robot reading, It will classify which cell is more likely to be.

I have reed other papers like SLAM (Simultaneous Localization And Mapping) but I’m too far to reach this. I’m looking for a Simple PyTorch solutions to ML learn a 2D map based on cell parâmeters.

Any help will be very appreciated.
Marco