| dc.description.abstract | In order to lower human drivers’ driving load and to enhance their systematic performance during driving, driver assistant systems have been introduced during the past few decades. Unfortunately, a large proportion of existing lane keeping techniques only focus on how to hold the car in the center of the lane, which may be contrary to the driver's natural motion sense. This research focuses on developing a rational and precise driver model with fully human driver operating behavior, which is crucial for the study of active safety technology and can provide drivers with a comfortable motion by imitating driving habits and trajectory.
Modeling a naturalistic lane keeping control requires understanding of how a driver operates the vehicle, analysis from vehicle lateral dynamics perspective, and knowledge of the combination of driver’s physical limitation. Another requirement to build an adaptive steering control model is to regard driver’s steering behavior as a reciprocal process between anticipation and compensation. Based on two angles (near and far angles) mechanism and experimental data recorded by the SIMULINK and dSpace co-platform, a close-loop system is designed. The whole system is a combination of a PI (proportional–integral) controller driver model and a vehicle model, which integrates vehicle lateral dynamic characteristics and upcoming road information. Moreover, a nonlinear steering driver model is designed. This open loop driver model can effectively correct steering wheel angle by minimizing the error between recorded driving data and that of the simulated model.
The simulation outcome shows that the proposed model captures human drivers’ behavior well and has an excellent adaptability towards the change of vehicle dynamic parameters and external disturbances. | en |
