Порівняння методів керування безпілотними автомобілями

Abstract

The purpose of this work is to analyze different approaches of controlling driverless cars and provide personal sight on the ways to improve them. There are 6 levels of autonomy of driving automation according to The Society of Automotive Engineers. Currently only level 3 is reached by Mercedes and Tesla (count starts from level 0). Research was conducted by teaching different existing neural networks that try to simulate inner workings of different existing systems and comparing their results to choose the best approach and ways of improvement