Intelligent Platoon Operating Environment Driving Behavior Cognition And Collaborative Control Optimization Method

With the development of technologies such as perception,communication,control decision-making,and artificial intelligence,intelligent transportation systems play an important role in improving traffic safety,efficiency,and optimizing energy consumption.Automated driving is an important development direction for intelligent transportation.One of its main application scenarios is the intelligent operation of the autonomous platoon.At present,Europe,the United States,and Japan have successively launched research programs related to the safe operation of smart platoons such as SARTRE,PATH,and AHS.China is also conducting in-depth research in areas such as precise positioning,trusted communications,and collaborative control of smart platoon operations,but since autonomous vehicles will not completely replace human driving,autonomous vehicles and human-driving vehicles will coexist,the issue of lacking cognition of drivers' driving behavior need to be solved.Therefore,this paper research on the optimization method of the platoon operation control strategy,based on the operation features of the intelligent platoon,also,this paper establish a cognitive model that can obtain the driver's continuity,sporadic and stressful driving behavior characteristics,and optimize the platoon formation mode in scenarios such as road segments and signalized intersections.The main research work of the thesis includes the following:(1)This paper proposed the drivers' continuity,sporadic,and stressful driving behavior feature acquisition algorithm,modeled driving behavior features by leveraging non-parametric Bayesian method,solved the problem of cognition of the driver's driving behavior characteristics by decomposing the driver's driving operation process into different primitives and using the critical operation primitive to obtain the information of the driver's driving behavior characteristics.(2)This paper proposed an algorithm for acquiring driver's following behavior characteristics based on the different driving styles of the preceding vehicles.This algorithm obtained the driver's following behavior characteristics by leveraging the raw data of vehicle following events,classified the acquired following behavior characteristics,and made them feedback to the original vehicle-following model so that the drivers' driving behavior characteristics when following different styles preceding vehicles are acquired,and the problem of operation differences caused by the vehicle order changed in the platoon is solved.(3)A step-by-step single-vehicle overtakes platoon algorithm based on the characteristics of the drivers' lane-changing behavior is proposed.The relationship between the velocity of the vehicle before the safety slot,the velocity of the vehicle behind the safety slot,the velocity of single-vehicle joining a platoon and the required safety slot was studied,and the speed matching scheme with the minimum safety slot required for the vehicle to overtake a platoon was analyzed.The Hierarchical Constrained Multi-objective Optimization Method based on Improved Particle Swarm Optimization is proposed,which provides the optimized three levels speed guidance strategy for the SVOP algorithm.Also,the strategies which could be accepted by the drivers were filtered by leveraging the drivers' lane-changing behavior features,so that the problem of singlevehicle overtakes platoon on the two-lane two-way road and the problem that drivers could not accept the advanced driver assistance strategy were solved.(4)This paper provided a platoon vehicle separation approach with unsupervised learning to learn the driving patterns of the human-driven vehicle at intersections with SEC(safety,efficiency,and energy consumption)requirements to select the optimized separation strategy.The Bayesian nonparametric learning was employed to segment the drivers' driving raw data into driving primitives,and selecting the separated vehicle by considering the safety,efficiency,and energy consumption.The problem that vehicles have to cross the intersection in the next green light phase or separate from the platoon since the limited green light duration was solved.(5)This paper proposed a multi-vehicle motion pattern acquisition algorithm based on the Dirichlet process-mixed Gaussian process.By considering the multi-vehicle motion pattern around the platoon as the mixed Gaussian process and considering the Dirichlet process as the prior distribution of Gaussian mixture weights,this method establishes the multi-vehicle motion pattern velocity field around the platoon.By comparing the operation efficiency of vehicles in different multi-vehicle motion pattern,the platoon's perception of the driving environment could be realized.The problem of how to select the optimal parameters for the macroscopic traffic flow from the optimal microscopic traffic behavior is solved.This paper obtained the raw data of drivers' driving behavior by leveraging the virtual driving simulation test platform.For verifying the capability of the proposed method,the microscopic traffic operation optimization parameters were injected into the macroscopic traffic simulation environment.The development and design of the driving control system of the autonomous platoon provide a high theoretical reference value.