Laterally String Stable Control at Large Following Distances Using DRTK and TDCP

Abstract

This thesis examines the lateral string stability of vehicle convoys. String stability is a type of stability that relates specifically to interconnected systems. In the case of vehicle convoys, string stability examines how the convoy, or "string", as a whole reacts to disturbances applied to the lead vehicle. When a convoy is considered string unstable, the disturbances at the lead vehicle are propagated down the stream. This occurs even if each vehicle is locally stable. When a convoy is considered string stable, those disturbances are dampened out along the string of vehicles. The idea of string stability may be formulated as both longitudinal control and lateral control problems. A longitudinally unstable string has the possibility of a vehicle wrecking into its preceding or following vehicle. A laterally unstable string has the possibility of a vehicle running off the road or wrecking into a vehicle next to it. This thesis addresses ways to prevent string instability in a lateral sense.

A classical cascaded control strategy is presented which uses feedback of lateral position error and vehicle heading error. The measurements of lateral position error and heading error are acquired using dynamic base real-time kinematic positioning solution (DRTK) and time-differencing of the carrier phase measurement for odometry (TDCP). This methodology for generating measurements allows the vehicles in the convoy to follow at much greater distances than if a camera/radar was used for measurement generation. With this architecture, a baseline control strategy where each vehicle in the convoy follows the ultimate lead vehicle is employed. This control strategy is compared against another control strategy where each vehicle in the convoy follows the immediate lead vehicle. The control strategies are compared for multiple simulation scenarios using the industry standard vehicle simulation software, CarSim. These scenarios examine a manually driven or an autonomously driven ultimate lead and three driving scenarios: a single lane change, a double lane change, and driving on the NCAT test track. Evaluations are made based on the lateral error along the string. The results show that the immediate lead following strategy is able to achieve lateral offsets which are nearly equal to the ultimate lead following strategy; therefore, the requirements of the convoy itself should be the deciding factor for which following strategy is employed.