The Research On Typical Behavior Identification Method Of Intelligent Vehicle

As an important part of the intelligent vehicle environment perception, the behavior identification of intelligent vehicle provides the necessary data for controlling and decision-making of the intelligent vehicles. It is the foundation to ensure the safety and reliability of intelligent vehicle operation. This paper sets up a system to research for the issue of intelligent vehicle typical behavior identification. The main works of this paper are as follows:(1)After studying the intelligent vehicle architecture and functional structure, this paper built a system of typical behavior identification of intelligent vehicle, analyzed the working principles and the key technologies of each module in the system,and then completed the hardware selection.(2)After studying the estimation algorithm of intelligent vehicles pose, chose rotation vector algorithm as vehicle attitude estimation method of gyroscope, selected Gauss-Newton method as vehicles pose estimation method of accelerometers and magnetometers, utilized the extended Kalman filter algorithm to fuse the results of the two kinds estimation method to complete the vechicle pose estimation. Simulations show that this method has good vehicles pose estimation results.(3)Proposed an intelligent vehicle speed estimation method based on modified acceleration to solve the problem of the cumulative error occurred during the process of integrating vehicle acceleration when getting the vehicle speed. Simulation results show that this approach reduced the impact of error and the noise of the vehicle acceleration when estimating the vehicle speeds, and the method has good vehicle speed estimation results.(4)Classified the typical behavior into different vehicle behavior types by analyzing the performance characteristics of vehicle attitude and velocity in the typical intelligent vehicle behavior. Extracted the FFT coefficient of vehicle attitude and the digital signature of vehicle speed curve respectively as vehicle behavior characteristics according to the type of vehicle behavior, and it’s a support vector machine algorithm input to achieve intelligent vehicle typical behavior identification.(5)Built a platform of intelligent vehicle typical behavior identification,completed the software design and algorithm selection. Verified the effectiveness of the proposed intelligent vehicle typical behavior identification method by the experiment.