Mode Estimation And Safety Control For Multiple Autonomous Vehicle System In Dynamic Environment

Intelligent Transportation System (ITS) uses intelligent vehicles as transport instrumentto improve the efficiency and security of the transportation system, which is one of the mostpopular research directions in home and abroad currently. The study of theory nature andsimulation of ITS can save resources and costs, and it also help to solve the traffic accidents,traffic congestion, energy saving and other practical problems. This paper primarilyresearches the state estimation and safety control strategies of autonomous vehicle system indynamic environments, the main content are as follows:(1)、Collision avoidance control of multiple autonomous vehicle system in dynamicenvironment is modeled based on hybrid automaton theory. The mode state of autonomousvehicles and switching relationships between them when under controlled were cleared,which supplied premise foundation for the following work. Besides, a collision avoidanceplanning method based on fuzzy logic control in crossroads environment was designed.Experimental results demonstrate the safety and effectiveness of the control algorithm.(2)、Vehicles following and lane changing process are studied, the states and switchinglaw of the controlled vehicles in this two move mode are describing. Artificial potential fieldmethod is applied to the following models. And impose field force to the following vehicleto achieve the status of the front one. Experiments verify the stability of this method.Besides, combine the hybrid automata model theory, the status of the various modes of thelane change process are defined, and gives some reference formulas for the lane changeconditions, trigonometric curve is used to portray the vehicle’s track to make it smooth. Thefeasibility and correctness of the changing model are verified by experiment.(3)、Let the vehicles have BDI reasoning ability to adjust the operating strategyreal-time under the driven of ultimate goal desire. The mode framework of the vehicle’soverall behavior is proposed. Take blackboard architecture to achieve informationinteractivity during the system running storage, make the cars can get every needinformation to realize rapid and safe operation. Model the entire multi-autonomous vehiclesystems, set a crossroads and T-shaped traffic environment to execute a systems simulationexperiment through Matlab programming.(4)、Multi-vehicles system task allocation problem is studied based on the fire-fightingenvironment and Nash equilibrium in game theory. According to the primary features of task model and Nash equilibrium, a task allocation algorithm based on game theory ispresented. Vehicles select their behaviors strategies according to the utility function andinduce them to extinguish the greater fire with larger publish value to get larger rewardvalue. The value of total task revenue，which is consistent with the realistic fire-fighting taskmodel, is used to evaluate the advantages and disadvantages of this algorithm, Experimentalresults show the effectiveness of this algorithm.In summary, this paper focused on the research of state estimation, control strategiesand task allocation issues for autonomous vehicles in dynamic environment, the simulationresults verify the effectiveness of the proposed algorithm.