Research On Lane Changing Decision Model For The Intelligent Vehicle Considering Driving Style

With the advancement of connected vehicle technology and driverless technology,complex traffic scenarios involving multiple subjects represented by artificial driving,connected vehicles and self-driving vehicles are gradually becoming a reality.In human-vehicle co-driving,the decision-making behaviour of intelligent vehicles has been mainly aimed at efficiency and safety,lacking the problem of consistency between its decision-making behaviour and that of the driver;and in the past,the driving style of the driver,once confirmed,no longer changed and was steady-state constant,while more and more research shows that there is a certain volatility in the driver’s style with the surrounding traffic environment.In order to seek a lane change decision-making behaviour of intelligent vehicles in the intelligent networked hybrid traffic environment that is more consistent with the driver’s benefit-seeking logic,this study takes the lane change behaviour of intelligent vehicles as the main research object,considers indicators such as vehicle lane change utility evaluation and vehicle lane change safety evaluation,incorporates dynamic driving style character methods,and improves the intelligent driving vehicle lane change decision-making model to achieve a more efficient and safe lane change behaviour decision.Specifically,the following three main parts of work were carried out.（1）In order to dissect the changing patterns of driving styles in different driving environments,the volatility of driving behaviour with changes in the driving environment was identified through the analysis of driving behaviour characteristics in the time domain for two different traffic environments,namely steady-state and oscillatory.Therefore,a framework for the character of dynamic driving styles within the long-and-short term time domains was established based on a focus on the fluctuating characteristics of driving behaviour as a means of quantitative character of driving styles.（2）The prediction of neighbouring driver behaviour can provide more adequate decision support for intelligent driving vehicles’lane change decisions.To improve the safety of lane change decisions for intelligent driving vehicles,the study character dynamic driving style as a driving style portrait and real-time traffic environment as a driving environment map,and constructs a complex convolutional neural network structure with two-channel input for training based on deep learning techniques.It was found that the designed method could better control the occurrence of incorrectly predicted lane-changing behaviour while maintaining high prediction accuracy,providing a guarantee for the safety of the driver’s own lane-changing behaviour prediction.（3）In order to make the driver’s perception of driving conditions consistent with that of the intelligent vehicle during human-vehicle co-driving,and to enable the intelligent vehicle to imitate the driver’s driving habits,the maximum utility theory of lane choice and the dynamic driving style character method are incorporated on the basis of the traditional game-theoretic lane change model,and a lane change safety risk evaluation model is proposed for lane change safety constraints,and a vehicle lane change decision considering the fluctuating characteristics of driving style is established.model.To validate the adaptability of the model,the experiments were carried out using SUMO simulation platform and Python language for vehicle driving control simulation.Three highway environments with different traffic densities of1800veh·h^-1,2400 veh·h^-1 and 3000 veh·h^-1 were selected for the traffic scenarios.The model is compared with the LC2013 and MOBIL models and it is found that the proposed model in this study performs better in terms of decision making behaviour at both the micro and macro levels,especially in the higher density traffic environments.In summary,this paper conducts theoretical analysis and modelling research at several levels,including driving behaviour pattern analysis,dynamic driving style character,driver behaviour prediction and intelligent driving lane change behaviour decision making,with a view to improving driving efficiency and safety through more reasonable lane change decision making behaviour of intelligent driving vehicles in an intelligent networked hybrid traffic environment.