On the use of robust design methods in vehicle controller design

A robustness-based methodology for designing an automated vehicle longitudinal controller is presented. The resulting longitudinal controller, which is intended for point-following operations, is applied to velocity-dependent (slowly varying) automobile dynamics. The results obtained effectively demonstate how robust design methods may be applied to this class of dynamics. Further, similar approaches might be taken to controller designs for vehicles under different operating modes, e.g., under carfollowing operations. The design consists of a cascade compensator, which is selected to achieve small tracking errors, and an observer/controller compensator. Kharitonov-related robustness methods are used to design a fixed (constant coefficient) observer/controller, such that the resulting closed-loop pole-clusters are confined to an acceptable area of the s-plane. The performance of the resulting overall system is evaluated using a large-signal, entry merging command and some small-signal mainline commands. The results are comparable to ones obtained by utilizing more complex controllers, designed to achieve a velocity-invariant response and based on parameter scheduling[1],[2] and nonlinear design methods[3].