| Adaptive Cruise Control (ACC) systems have been proposed as an enhancement over the existing cruise controllers on ground vehicles. ACC systems control the vehicle speed to follow a driver's set speed closely when no lead vehicle is in sight. When a slower leading vehicle is present, the ACC controlled vehicle will follow the lead vehicle at a safe distance. ACC research first began in the 1960's, and has received ever-growing attention in the last decade. Their commercial implementation was not possible until recently with significant progresses in sensors, actuators, and other enabling technologies.; Most of the existing ACC research has focused on the range and range rate performance. The effect of ACC on traffic flow, however, has not received much attention. A key issue, the so-called “string stability”, needs to be carefully examined to ensure that an ACC design does not affect traffic flow adversely. A string-stable ACC design ensures that inter-vehicular tracking errors decrease as they propagate along a vehicle string. In this dissertation, we provide a framework for the study of string stability. Both uniform and mixed vehicle strings are analyzed. Necessary and sufficient conditions for string stability are presented. The stability degradation is shown in the context of the “string stability margin” (SSM) which is an index defined to measure the string stability of a vehicle. Two ACC control methods, the distributed optimal synthesis method and the sub-optimal H∞ Control, are proposed to explicitly address the string-stability issue in ACC designs. Results show that string stability can be guaranteed when the ACC control problem is formulated appropriately.; In the past, ACC designs were evaluated either using a complex vehicle model in a two-vehicle-platoon format (a lead vehicle followed by a controlled vehicle) or a simplified vehicle model with a string of identical vehicles. In this research, we use a microscopic traffic simulator. In this simulator, an autonomous driver model is constructed such that this model is capable of performing approaching, following, and lane changing maneuvers. This complex driver behavior model enables us to study safety and traffic-flow characteristics more accurately. |
