Thesis -> https://github.com/Magikis/bachelor-thesis/blob/master/docs/bachelor_thesis.pdf

• Python >=3.6
• TORCS 1.3.7 with SCRC patch
• numpy, sklearn, scikit, arrow

To run an agent use: python main.py Use --help to look for available arguments.

Agents shortcuts:

• line-follower,
• tree -> Decision Tree single model,
• mlp -> Neural network single model,
• dma -> double agent model,
• dma-sh - >double agent model - shared history,
• sma -> single model agent,
• tma -> triple model agent

To run race without graphics run: python ruPractice.py

To run algorithms which learns optimal speed limits fot specific track run: python speeed_limits_learner.py

All driving logic is in directory agents. All tracks XML's can be found in directory tracks All models learning was done with jupyter notebooks available in root directory

Trained models can be found in models directory