Comparison Of Autonomous Driving Approaches

The task of developing autonomous driving systems is a very important problem at the present time. There are two main approaches to the development of such systems: modular approach and end-to-end approach. This paper presents a comparison of current autonomous driving systems based on different approaches in a simulator environment. We used the CARLA Leaderboard benchmark to evaluate the quality of the considered autonomous driving systems. Experimental studies show the advantage of the considered modular autonomous driving system algorithm in comparison with end-to-end one.