Research On Bionic Lane-changing Decision-making Model For Autonomous Vehicle Under Dynamic Urban Environment

Autonomous vehicle will exhibite the characteristics of anthropomorphic driving behavior and intelligentized decision-making in an era of information, and replace human drivers in the process of driving on the urban road and effectively and safety complete the driving task. The core of the self-driving technology is to establish driving decision-making model through extrcating experience driver’s driving rules using Rough Set theory. This technology will help autonomous vehicle to be able to adapt to the complex, dynamic and uncertain traffic environment and work well like human beings. The finding from this work could provide a theoretical basis for the in-depth research and application of autonomous vehicles driving under complex and uncertain environment.There are plenty of uncertain, dynamics information in the urban environment with consideration of varied traffic scenarios and unpredictable behavior of the participants, which has seriously affected the autonomous vehicle making accurate and timely decisions. In this paper, the lane-changing decision-making process considering the motion state of vehicles surrounding the autonomous vehicle was discussed under the dynamic urban environment. The virtual buildings around Beijing institute of technology were built by Google Sketchup software, then driving simulation model in urban environment was constructed based on Pre Scan /Matlab software. The motion state parameters about vehicles during the lanechanging process were obtained through the simulation, and then the effect factors of drivers’ decision were deliberated. Driver’s decision rules were extracted through the concept of Rough Set Theory. Through investigating the decision-making process of drivers and structuring the knowledge base about drivers’ driving rules, the velocity selection strategies based on the safety threshold were formulated. According to the different intentions of lane changing, an algorithm was presented for discretionary lane changing decision-making when velocity control operation is desirable and feasible. The development of the algorithm has established based on drivers’ experience, safety threshold and theories of acceptable gaps; Considering the gap determination conditions and the judgment of coordination driving behavior; the merging behavior which embodied the characteristics of mandatory lane changing was analyzed under the typical urban road environment, the decision-making model for merging behavior was proposed,which could determine the safety gap effectively during merging process and illuminate the bionic cognitive decision-making mechanism under complex dynamic urban environment. The algorithm was proven to provide satisfactoryvelocity control actions as well as to safely decide whether to conduct lane changing under real urban environment. The reliability and effectiveness of the decision models were validated by both simulation platform and real road experiments. The results provide driving knowledge and theoretical basis for the depth study of lane-changing behavior of uncertain decision-making.