Research On Human-machine Sharing Control Strategy Based On Traffic Situation

As the future development direction of automobile,automatic driving technology can not only liberate the hands of drivers to reduce the burden of drivers,but also improve the driving safety.However,due to the limitations of current technology,imperfect infrastructure and social acceptance,there is still a long way to go before the commercialization of fully automatic driving.As a transition stage towards full automation,human-machine co-driving realizes the coexistence of human driver and low-level intelligent driving function.However,when the driver is "out-of-the-loop",there will be problems such as excessive dependence,loss of real-time grasp of external information and reduction of trust in intelligent driving.And in the research of human-machine co-driving,when the human-machine goals are inconsistent,how to deal with the cooperation and conflict relationship between the human driver and the intelligent system is a key issue.Therefore,in response to the above problems,this article focuses on the driver is always "in-the-loop" human-machine sharing control strategy.The specific content is as follows:(1)Design of driving rights allocation strategy based on traffic situationIn order to solve the problem that the driver’s trust in intelligent driving is reduced,this paper designs a driving right allocation strategy based on the traffic situation based on the idea of changing the driver’s trust according to the complexity of the scene.The specific content draws on the idea of artificial potential field,and models the current driving risk field of the vehicle.According to the characteristics of various targets in the traffic scene,the driving risk field is divided into two parts,dynamic field and static field,and the speed is used for the two parts.And relative distance are two variables,and the single variable of relative distance is modeled as the core variable.Then,based on the fuzzy rules,a driver’s weight distribution strategy is designed based on the modelling results of the driving risk field and the driver’s characteristic parameters.In the man-machine shared control strategy.(2)Lane changing strategy design of intelligent control systemIn order to obtain the expected path of the intelligent system in human-machine co driving,this paper designs a simple and effective lane changing strategy of the intelligent system,analyzes the traffic vehicle information in the current traffic scene,and obtains the expected path of the intelligent system through three steps: lane changing decision-making mechanism,lane changing feasibility analysis and lane changing trajectory planning.The decision-making mechanism of lane changing is based on the collision time margin calculated by the relative speed and distance.The feasibility of lane changing is guaranteed by establishing the safe distance model between each vehicle and its own vehicle.The lane changing trajectory is determined by the unary quintic polynomial derived from the lateral displacement and initial position of lane changing.(3)Design of man-machine sharing control strategy based on Nash gameIn order to solve the problem that the driver is " out-of-the-loop ",the system design is carried out by adopting the man-machine sharing control method that the driver is always "inthe-loop",and the Nash game theory is added to describe the characteristics of man-machine interaction,which solves the problem of man-machine conflict and cooperation.In the humanmachine shared controller,aiming at the lateral control problem in the human-machine codriving system,the derivation and construction of the control problem are completed in the framework of model predictive control,and then the vehicle model including lateral position,lateral speed,yaw rate and yaw angle is established,and the intelligent system and the driver’s corner are taken as the inputs;finally,the expected path of the driver is assumed to be known The optimization problem of model predictive controller is reconstructed by Nash game method,so that the optimization problem can obtain two control quantities of driver angle and controller angle at the same time.The driving weight obtained before is taken as the weight of optimization function,and the convex iteration method is used to solve the optimization problem.(4)System integration verification and driver-in-the-loop testIn order to verify the effectiveness of this strategy,a driver simulation test platform was built.The hardware part includes the d SPACE tool SCALEXIO for real-time simulation,Senso Wheel torque steering wheel and Kistler MSW sensor,and the software part includes Matlab/Simulink for algorithm building,and Car Sim and Control Desk for vehicle dynamics and scene construction can realize the visualization of traffic scenes and driver input and data collection.According to the strategic requirements,some lane-changing scenes including static obstacles and dynamic obstacles were designed,and driving tests of drivers in different scenes and working conditions were carried out,and data collection and data processing were completed.In the system integration simulation verification and driver-in-the-loop test,test scenarios are designed according to the characteristics of the strategy,and it is verified that the system can not only allocate dynamic driving rights according to traffic situation assessment in static and dynamic scenarios,but also realize the shareing control of Human-machine game under dynamic variable weight.