Abstract
Keywords - Intelligent Transport Systems, Adaptive Cruise Control, experimental vehicle, extended Kalman smoother, image processing
Abstract - The purpose of this study is to develop an experimental vehicle in which several kinds of sensors and an integrated data recording system are installed, and to develop a software algorithm for obtaining dynamic characteristics of the vehicle and those of other vehicles having relation to the one, such as speeds, accelerations, spacings, longitudinal and lateral positions simultaneously. This paper describes the results of verification of the experimental vehicle and the software algorithm. With this algorithm, state variables of vehicle kinematics are identified with the Kalman smoother technique based on the maximum likelihood estimation. This vehicle is also equipped with video cameras and after an image processing in the laboratory, the movements of vehicles around the experimental vehicle can be observed. Therefore, the experimental vehicle and the software algorithm can provide not only the kinematics of the experimental vehicle but also those of the vehicles around this vehicle. Speed, heading direction, and angular velocity of the experimental vehicle are measured by each sensor and integrated with the GPS measurement. The Kinematic GPS (K-GPS) mode have very high accuracy but is not so stable, less accurate modes such as Differential GPS (D-GPS) or normal GPS modes sometimes take place. The authors propose a state equation of three-dimensional vehicle kinematics of parallel movement and angular movement with thirty state variables, and the extended Kalman smoother algorithm (E-KSA) is adopted. E-KSA can estimate vehicle state variables simultaneously and consider the every data collected in the past and the future even if a variable is not measured at the moment with maximum likelihood estimation. Practical data are applied to the algorithm. The validity of the algorithm is checked. This algorithm can produce very precise state variables of the experimental vehicle's kinematics. The experimental vehicle can also take the digital video image of diagonally front and diagonally behind of the vehicle on the right hand side. A semi-automated image processing technique is developed, and it can manage to identify every surrounding vehicle easily. The system shown here makes great advances in developing driving behaviour models because the dynamics of vehicles can be observed precisely enough. These models lead to the possibility for assessment of control strategy of vehicles with ACC (Adaptive Cruise Control systems) in terms of highway traffic safety and efficiency in the mixed-traffic operating condition.