Performance Evaluation of Path Tracking Controllers for Scaled Robotic Cars

A versatile platform for testing automated driving algorithms is represented by scaled robotic cars, which can reduce costs, increase the repeatability of testing, and enhance safety, compared to experiments conducted on full-size vehicles. Therefore, scaled cars are used before testing control solutions on real vehicle demonstrators. Similarly to full-size autonomous vehicles, scaled robotic cars typically feature a hierarchical control structure that encompasses various functionalities, including path planning and path tracking, as well as state estimation algorithms. This work presents an experimental comparison of path tracking control methods applied to scaled robotic vehicles with front steering actuation. Six path tracking control strategies---namely, based on (i) model predictive control, (ii) pole placement control, (iii) linear quadratic regulation, (iv) sliding mode control, (v) proportional integral control, and (vi) the Stanley method---are designed, implemented, and experimentally evaluated on scaled robotic QCar vehicles along five paths at eight vehicle speeds. Performance is assessed through a set of key performance indicators, demonstrating the superiority of the model predictive and sliding mode controllers