SUMMARY
Increased production of electric cars due to climate change and global warming, and the ability of self-driving systems in electric cars made its popularity soar. The problem with the current implementation of self-driving cars is the need for large datasets. In this study proposed a system that does not require datasets for its training by applying Reinforcement Learning and NeuroEvolution of Augmented Topologies (NEAT), NEAT is a form of genetic algorithm used as the brain of self-driving cars and Reinforcement Learning is used as a set of rules and actions to train self-driving cars in simulation environments with the help of Unity Engine. The results of this simulation prove that the Reinforcement Learning and NEAT methods are able to complete tracks with an average of 12 minutes 20 seconds faster than neat simulations alone.