A Secririty Verification Approach For Autonomous Maneuver Decision Of Vehicles

In order to sovel the problem that the traditional algorithm has low confidence level and incomplete verification results in vehicle autonomy decision security verification,the formalized accessibility method was used to investigate the feasibility of autonomous decision-making behavior state of vehicles and gives the complete set of autonomous decision-making behavior state of vehicles.The directed convex polyhedron algorithm is used to conservatively linearized the nonlinear hybrid system of autonomous vehicles,and the reachable set is calculated by iterative principle and Zonotope approximation method.This paper obtains the following three results of autonomous vehicle safety verification in different road environments:1.Simulation experiments of multiple autonomous vehicle avoidance on urban narrow roads show that the reference trajectories of the respective host vehicles are obtained after path planning using the A * algorithm.Under the initial state set and sensor noise boundary conditions,under the control of a simple estimation tracker,The position reachable sets of multiple autonomous vehicles give a high level of security assurance,and the unsafe coefficients of the respective host vehicles in this process are calculated after analyzing the intersection of the reachable sets and unsafe sets of the respective host vehicles.2.The simulation experiment of lane avoidance shows that the reference trajectories of the respective host vehicles are obtained after path planning using the optimized genetic algorithm.Under the initial state set and sensor noise boundary conditions,the autonomous vehicles are controlled by the horizontal and vertical trajectory trackers.The position reachable set and speed reachable set provide a high level of safety assurance,and prove that autonomous vehicles can safely and compliantly complete autonomous driving under the lane change and collision avoidance environment.3.Simulation experiments on collision avoidance of multiple autonomous vehicles show that the reference trajectories of the respective host vehicles are obtained after path planning using the A * algorithm.Under the initial state set and sensor noise boundary conditions,the trajectory tracking control of autonomous vehicles based on ADRC Under the control of the vehicle,the position reachable set and speed reachable set of the autonomous vehicle provide a high level of safety guarantee,which proves that under the circumstance of overtaking and collision avoidance,multiple autonomous vehicles can safely and compliantly complete autonomous driving.Innovations included in this article:First,this study uses the vehicle autonomous decision system as a hybrid system,and proposes to use a formal method to verify the safety of the vehicle autonomous decision system.Based on this,the reachable set is used to express the vehicle autonomous behavior state.By traversing and solving the complete state space of the system automatically,the completeness of system safety verification during autonomous driving of the autonomous vehicle is theoretically guaranteed;Second,this study proposes to use the zonotope expression to approximate the reachable set,and to visualize the reachable set,to conveniently and intuitively detect the parameters such as the position and speed of the autonomous vehicle;Third,based on the CORA toolbox,this study developed a calculation method that can be applied to the transformation of complex road geometries and vehicle autonomous control modes.Using this method,various operating parameters of autonomous vehicles can be well calculated;...