Decision-making Modeling Of Overtaking Behavior For Autonomous Vehicles On Freeway Environment

Based on the status and trends of domestic and international development of the autonomous vehicles,the behavioral decision-making problem on freeways is mainly studied in this paper.To make the autonomous vehicles adapt to real traffic flow,we use a humanoid research route to establish the decision-making system.In this paper,we present a novel framework for modeling the decision-making process of overtaking on freeways.As the input module,the microscopic scene model is to obtain abundant and valid information of the situation from the sensor data,which provides an effective basis for decision support.The decision-making framework is based on a Hierarchical State Machine(HSM),which models driver behaviors by discrete top states and substates.In this way the behavioral decision-making process is translated directly into the relationship between the states of HSM,which reduces the difficulty of the model and makes it easier to maintain.The overtaking process consists of two stages: the generation of overtaking intention based on the RBF neural network(RBFNN)model and the judgment of overtaking condition based on rules.It is closely related to their driving characteristics to generate an overtaking intention,and the RBFNN has been proved to be a general-purpose classifier which can be used in many cases.Therefore,a RBFNN model is used to learn and fit the person-specific driving habits.Then the rules of safety,efficiency,comfort and altruism are formulated as the determining condition for overtaking.It’s important to avoid dangerous consequences caused by the error outputs of RBFNN model by judging condition,which can ensure reasonability and validity of the decision result.Meanwhile,in order to acquire the driver behavior data and verify the feasibility of overtaking decision-making framework system by Man-in-the-Loop simulation,a co-simulation platform is built by Prescan and Matlab/Simulink.Then the real vehicle tests based on the autonomous vehicle Ray are carried out in the 3rd ring road of Beijing.The experimental results show that the autonomous vehicle can achieve overtaking on freeway with the decision-making framework established in this paper.