| This dissertation presents a new approach for automated vehicle steering control, i.e. the autonomous vehicle following approach. The goal of vehicle steering control is to achieve lane-keeping performance, i.e. to keep the vehicle close to the road centerline, by adjusting the steering angle at the tires in real time. Current vehicle steering control algorithms rely on road reference systems and appropriate on-board sensors to detect the vehicle's deviation from the road centerline. Therefore, these methods are referred to as road following algorithms. Autonomous vehicle following allows a vehicle to automatically follow its preceding vehicle based on real-time information of the relative distance between the two vehicles. If the leading vehicle tracks the road centerline reasonably well, this approach can achieve good lane-keeping performance of the controlled vehicle. The main benefit is that no road reference system is required for implementation of this new control scheme.; In this dissertation, the relative distance is measured by an on-board laser scanning radar (LIDAR) sensor. The sensing mechanism of LIDAR is based on the "Time of Flight" principle. The sensor emits laser beams to scan in the horizontal plane in search for the leading vehicle or any other reflective object in the environment. In each sampling period, the sensor returns 80 sets of measurements, most of which are caused by environmental clutter. A probabilistic data association based data processing method is used to extract the position information of the leading vehicle from the LIDAR measurements.; One issue in autonomous vehicle following is a propagation of errors from one vehicle to another. When errors increase in the upstream direction of a vehicle platoon, the platoon is string unstable. Analysis of string stability is performed, and inter-vehicle communication is suggested as a means to solve the string stability problem. Experimental results are used to show the effectiveness of the proposed solution.; Autonomous vehicle following may work as a back-up system for road following systems when the road following sensors are under partial failure. An integrated vehicle steering control scheme, which essentially combines the use of LIDAR and the remaining road following sensors, is introduced. The optimal combined use of LIDAR and a set of magnetometers mounted under the rear bumper of the test vehicle is considered. In the combined use of these two independent sensors, the interaction between the two system outputs is a major concern. Effects of a disturbance associated with one sensor output on the other sensor output are investigated. This leads to solving the problem of minimizing the interaction between the two system outputs. The solution is verified by both simulations and real-time testing. |
