Any ideas to create Autonomous mode for my robot. I am trying to implement Reinforcement Learning to do this. The only problem, is that I can’t use PyTorch or Tensorflow which is already a trained framework. So to solve this solution I must create my own neural network and train my robot. How would I do this?
I do not recommend using a neutral network as it is for more effecient to use simpler PID controllers, odometry, motion profiling etc. If you really wanted to use reinforcement learning you will probably have to use c++ for performance reasons. You could turn looking at this library: Copilot: An ML Library for PROS V5y
Are Neural Networks the same thing as PID controllers?
No, they are not the same.
I agree with EcstaticPilot on this, a neural network is very overkill. A simple PID controller is more than enough for 99% of VEX. You can read about PID here: PID Controller - BLRS Wiki
Is there a specific reason you wanted to use reinforcement learning?
Well I have the neural network coded and as of now I am training the robot. It works perfectly fine. As of now it tries to learn how to drive on its own. Is the implementation of a PID controller diffcult?
PID is extremely easy and simple to implement.
For some reason, I doubt that you have created a neural network the can consistently control the robot well. If so, I recommend you share some insight with everyone, and write a research paper lol.