Research On Haptic Shared Human-machine Shared Control Steering Assistance System Based On Driver Characteristics

The current driverless technology is not perfect,and there are still safety and legal problems.Therefore,until a fully reliable driverless system matures,the human-machine cooperative driving model in which the driver and the assisted driving system jointly drive the car becomes a relatively good solution at present.Human-machine cooperative driving is divided into two types: haptic shared human-machine cooperative driving and input modified human-machine cooperative driving,in which haptic shared human-machine cooperative driving provides haptic feedback to the driver by applying auxiliary torque on the steering wheel to guide the driver to operate.This human-machine cooperative driving method allows the driver to continuously interact with the assisted driving system through haptic feedback,the driver’s degree of presence is higher,and in an emergency,the driver is allowed to override the auxiliary torque of the assisted driving system,retaining the driver’s ultimate control over the vehicle.However,the auxiliary torque of the driver assistance system in haptic shared human-machine cooperative driving may conflict with the driver’s torque,which may reduce the comfort of the driver.Therefore,it is particularly important to reduce the human-machine conflict of haptic shared human-machine cooperative driving,ensure driver comfort and acceptance of assisted driving systems.In order to solve the problem of human-machine conflict in haptic shared humanmachine cooperative driving,relying on the national key research and development program "Multi-system Efficient Integration of In-wheel Motor Action Module and Vehicle Torque Vector Distribution Technology"(Project Number: 2021YFB2500703)and the science and technology project of Jilin Provincial Department of Education,"Research on decisionmaking and control strategy of automatic lane change of X-by-Wire vehicle in intelligent transportation environment"(project number: JJKH20231148KJ),based on driver characteristics,this paper designs a haptic shared human-machine cooperative driving system.Firstly,the driver characteristics identification experiment in the obstacle avoidance scene is designed,and the drivers are clustered into three types: cautious,normal and aggressive according to the experimental results.Secondly,the machine learning algorithm was used to build the driver feature recognition model and verified the accuracy of the model recognition.Then,the path planning module of the assisted driving system is designed according to the identified driver type,and personalized obstacle avoidance paths are planned for different types of drivers.Finally,the path tracking module is designed to track the planned target path,and then calculate the deviation between the target steering wheel angle and the driver’s actual steering wheel angle,determine the obstacle avoidance assist torque of the assisted driving system,and adjust the size of the auxiliary torque according to different types of drivers,and realize personalized human-machine cooperative driving control for different types of drivers by superimposing the obstacle avoidance assist torque on the basis of the original torque of the electric power steering system.The main research work is as follows:(1)In order to build the driver characteristic recognition model,the driver characteristic recognition experiment in the obstacle avoidance scene was designed,and the driving simulator was used to collect the control parameters and vehicle status parameters in the process of obstacle avoidance and processed the collected data.According to the processed data,the drivers were clustered into three types: cautious,normal and aggressive,and the clustering results were used to train two machine learning algorithms,extreme gradient boosting and neural network,and driver feature recognition models based on extreme gradient boosting and neural network algorithms were built respectively,and finally the training speed and recognition accuracy of the two machine learning algorithms in driver feature recognition were compared.(2)The traditional artificial potential field path planning algorithm is modeled,and the two problems of local optimal solution and unreachable target point existing in the traditional artificial potential field method are reproduced through simulation,and then an improved artificial potential field method is proposed for the defects of the traditional artificial potential field method,and the simulation verifies the solution effect of the improved artificial potential field method proposed in this paper on the defects of the traditional artificial potential field method.Finally,an artificial potential field method based on driver characteristics is constructed,and the influence range of obstacles is adjusted according to the identified driver type,and personalized obstacle avoidance paths are planned for different types of drivers.(3)The obstacle avoidance path tracking controller based on model predictive control is built,and the three-degree-of-freedom vehicle dynamics model considering lateral,longitudinal and yaw is first established as the model prediction controller,and the prediction model is discretized.Then,the objective function and constraints of the model prediction controller are designed,and the simulation verifies the tracking effect of the path tracking controller on the obstacle avoidance path planned by the path planning algorithm.Finally,the obstacle avoidance assist torque is determined according to the deviation between the target steering wheel angle obtained by the path tracking controller and the actual steering wheel angle of the driver,and the size of the assist torque is adjusted according to the driver type,so as to provide personalized obstacle avoidance assistance for different types of drivers.