Research On New Driver Steering Model Based On Two-Point Preview

The global accident rate still remains high, more than90%is caused by the improper operation ofthe driver, which nearly30%is due to driver fatigue. Vehicle active safety technology can assist thedriver to complete a driving task, to alleviate driver’s physiological, psychological pressure. It caneffectively reduce the accident rate. For the same task of driving, the different driver will havedifferent reactions. Therefore, to establish a rational and precise driver model with fully human driveroperating behavior is very important for the study of active safety technology.In this paper, in connection with the steering control driver model for the intelligent vehicle, wemake a research on how to obtain driving environment information, how to learn and master the cardriving dynamic characteristics. The contents are as follows:(1) Make a design on steering driver model based on the two-point driver preview method on far,near points. The near angle, near angle change rate and the far angle change rate are used as the visualinput information, and the information-driven steering control process is regard as a second-orderdamping-spring dynamic system. Then, design the steering controller, so that the vehicle steeringcontrol process becomes smoother and smaller shocks. In addition, taking the system modelinguncertainties and vehicles outside interference into accent, an internal model controller based on Hinfinity optimization theory is designed, to restrain the uncertainty and interference. Finally, a singleline change simulation tests is made to verify the effectiveness of the steering controller.(2) Designed a steering controller based on the human self-adaptive steering behavior. It can befound that drivers can utilize the linear dynamics characteristic to acquire the skill of controlling thedriving direction. Besides, this skill can combine with physical limitations to form an internalreference model, which can be employed by drivers to adapt to the variable dynamics conditions.When the vehicle state is changed, the human driver can learn to adapt the changes in the vehicledynamics characteristics timely, to form an adaptive mechanism, and still be able to handle thetraveling well. Based on these studies, we designed a model reference adaptive steering driver model,combined with human driver preview mechanism. Single lane change simulation results show that:this method can effectively achieve the driver’s steering task and has good adaptability to the changesin the vehicle dynamics parameters.In conclusion, a human driver model with human driver preview mechanism as well as the behavior characteristics was designed, combined with robust internal model control algorithm. The model caneffectively restrain the modeling error, outside interference and the parameters change of vehicledynamics. Further research on the behavior characteristics of human driver, taking the driver learningadaptive characteristics into accent, the driver steering model with self-adaptive ability was designed.It can control the steering wheel more accurate and stable. The study has some guidance and referencefor further study on the human driver driving skills.