Research On Safety Decision-making And Planning Of Unmanned Vehicle Behavior In Urban Road Traffic Environment

With the popularization of cars,it not only facilitates people’s life and improves work efficiency,but also causes a series of social problems such as environmental pollution,road safety and traffic congestion.With the emergence of driverless vehicles,the artificial intelligence system can replace the driver to conduct driving operations,which can avoid traffic accidents caused by improper driver operations from the root and effectively improve road traffic safety.The decision-making and planning system is the core embodiment of the intelligence of unmanned vehicle and an important guarantee for the safety,efficiency and stability of unmanned vehicle.Therefore,based on the National R&D Projects "Development and Application of New Multi-functional Intelligent Vehicle Terminal(2018YFB1600701)",this paper studies the safety decisionmaking and planning system for driverless vehicles in urban road traffic environment.Primarily,research on the decision-making method of collision avoidance of driverless vehicles based on different road environments.In this paper,the application scenarios of driverless vehicles are selected,which are one-way lanes and intersections,and the potential collision risk of vehicles is analyzed.According to the critical conflict area of vehicles,a safe driving distance model is established to make collision avoidance decisions for driverless vehicles in different risk scenarios.Then,the prediction method of driving intention of other traffic vehicles in road traffic environment is studied.The real-time speed information and position information of the vehicle were selected as the observable sequence of the model,and the future driving intention of the vehicle was taken as the hidden state to be predicted by the model.In the Pre Scan simulation platform,the predictive effect test was carried out for the vehicle driving intentions in the one-way straight lane environment and the road environment at the intersection.On this basis,a method for predicting vehicle trajectory based on driving intention is studied.In order to improve the accuracy of trajectory prediction,considering the influence of the driver’s driving style on the trajectory of the target vehicle,GMM algorithm was used to cluster the selected sample trajectory according to the driver’s different degree of aggressiveness in the process of driving.The trajectory regression prediction model based on gaussian process was established according to the driver type classification,and the polynomial function was selected as the model mean function to describe the vehicle trajectory.Finally,simulation experiments are carried out to verify the trajectory prediction of vehicles under lane changing and turning driving behavior.Finally,a local path planning method for driverless vehicles based on improved RRT algorithm is studied.The method of setting the goal guiding threshold can reduce the generation of invalid exploring paths and improve the efficiency of path planning.In order to reduce the curvature fluctuation of the planned path,the traditional RRT algorithm is improved from the selection of the nearest secondary node and the Angle of path turning.Simulation experiments are carried out to verify the path planning algorithm in the static obstacle environment and the complex traffic environment.