Comparison of Control Approaches for Autonomous Race Car Model

Abstract

Autonomous driving is an interesting but challenging field. To make it accessible to college students, several universities worldwide joined to create an F1/10 Autonomous Racing Competition, where students compete in a so-called battle of algorithms. When developing these algorithms or combining a few of them in more complex control approaches, it is not always easy to decide which approach is better. In this paper, we present a comprehensive methodology for comparing the control approaches based on four metrics: racing performance, obstacle avoidance slowdown, CPU load, and tuning effort. We apply the methodology to three conceptually different control approaches by performing an experimental evaluation with an autonomous race car model on two tracks. The results show that the difference between reactive and optimal approaches is small on a simple track, whereas on more complex tracks, the optimal approach has significant advantages. While these results were expected, the comparison shows the gap we will aim to close with our future control approaches.