Sampling Rate Selection for Trajectory Tracking Control of Autonomous Vehicles

Trajectory tracking is a basic yet highly important function for autonomous driving, and its performance differs with different sampling rates. That is, the performance usually improves at the expense of high computer cost, and reducing the sampling rate may drive the system to instability. In this paper, the influence of different sampling rates for autonomous vehicle control is discussed and a sampling rate design method is proposed. Firstly, a 2-DOF linear vehicle bicycle model with a preview model for trajectory tracking is introduced. With the combined model, a certain bandwidth expression is derived from the model transfer function, which is then utilized to deduce the range of possible sampling rates. Impacts of sampling rate on tracking performance is demonstrated by both simulations and experiments.