Driver State Supervision And Extension Decision Based Lane Departure Assistance System

With the growth of car ownership,the number and proportion of traffic accidents caused by lane departure has increased year by year.More and more attention has been paid to researches of lane departure assistance system.In this paper,the decision method and lateral control of lane departure assistance system were studied.First,the 7 DOFs model and 2 DOFs linear model for vehicle were established.Based on the characteristics of lane departure assistance system,a vehicle-road model and a driver model were established,and a steer-by-wire system model was established,which laid a foundation for designing and verifying the control method in this paper.Then the decision method of lane departure assistance system was studied.First,lane departure decision based on extension was proposed.Time to lane crossing(TLC)and distance to lane center(DLC)were used as features to establish the extension set.Dynamic TLC threshold was designed to divide the set into classical domain,extension domain and non-domain,which were represented for free lane departure risk,lane departure risk and lane departure danger.The correlative function was calculated to divide patterns.The free driving mode,assistant driving mode and active steering by-wire mode were adopted separately in different domains.Then it was considered that the driver’s state has a direct impact on the lane departure,driver state supervision based on Gauss Mixed Hidden Markov model(GM-HMM)was designed.Driver’s state were divided into normal driving,improper operation and no operation.Driving samples were collected in a driving simulation test bench.GM-HMM was built to train and recognize driver’s status.Finally,the decision strategy of lane departure based on driver state supervision and extension was designed,and the decision of driving mode was made.Then the lateral control method of lane departure assistance system was studied.An angle controller based on TLC potential field method was designed for active steering by-wire mode.The free driving mode was driven by the driver alone.The assistant driving mode uses the input of driver and active steering controller were used as input of the assistant driving mode,and the weight of the control was determined by the correlation function in the extension decision.Finally,the simulations were made to verify the effectiveness of the control method under the improper operation and no operation working condition of the straight path and the bend path.Further,the test was done on the CarSim/LabVIEW hardware in loop test and a real car platform to verify the reliability of the control method.