Adaptive Fuzzy Tuning Framework for Autonomous Vehicles: An Experimental Case Study

Achieving precise trajectory tracking performance from an autonomous vehicle requires a carefully tuned controller. However, such a task is arduous which necessitates iterative testing. Furthermore, changes in traction condition render the offline tuned gains less viable. Hence, this paper proposes an adaptive tuning strategy to improve the performance of lateral trajectory tracking. In essence, the tuning framework utilizes fuzzy inference to update the controller gains online. The underlying rules are based on intuitive ideas that facilitate easy deployment. Moreover, the efficacy of the tuning strategy has been experimentally evaluated in multi-scenario conditions. The obtained results validate that the adaptive fuzzy-based tuning strategy consistently improves the tracking performance with a decrease in the tracking error with values of up to 73%. This paper is an effort to showcase the importance of a reliable tuning strategy towards motion control of autonomous vehicles.