Self-Driving Cars: Simulation, Issues and Possible Solutions for Implementation in India

Self-driving cars being the new technology are gaining attention from all over the globe. This paper deals with the simulation results of an autonomous car trying to learn from its environment which includes static blocks using machine learning. Learning is performed using Deep Q-learning. The neural network computes the Q-values on the basis of the rewards corresponding to the action that the car takes. The autonomous system in the car chooses that particular action that has a maximum reward. The actions are the angles through which the car can steer at a fixed speed. Also, difficulties related to the implementation of autonomous self-driving cars in India have been discussed and possible solutions to them have been presented.