Research On Lane-changing Strategy Of Unmanned Vehicles Based On Hidden Markov Prediction Algorithm

During the normal driving of a vehicle,changing lanes is one of the common driving behaviors.Traffic accidents occur frequently due to improper lane changes.In order to reduce the occurrence of traffic accidents,it is very important to develop a reasonable lane changing strategy for driverless cars.This paper mainly studies the autonomous lane changing algorithm of unmanned vehicles.The lane-changing behavior of the preceding vehicle is one of the influencing factors of the decision-making of the vehicle’s lane-changing.Based on the dynamic grid map and the Gaussian Hidden Markov Model,this paper establishes a lane-changing intention recognition and prediction algorithm for the preceding vehicle.In the dynamic grid map,the center of the vehicle head is the origin,and each grid unit is given environmental state information,and the four attributes of the horizontal and vertical coordinates and the horizontal and vertical speed in the grid are used as the observation of the mixed Gaussian Hidden Markov Model parameter.Divide the hidden states of the hybrid Gaussian Hidden Markov Model into left lane change,keep lane and right lane change,process the I-80 segment vehicle driving data in the US NGSIM data set,and extract the four attribute data of the preceding vehicle before training and verification Vehicle lane change intention recognition and prediction algorithm.The results show that the model can achieve high recognition and prediction accuracy,and the effect is good.In addition,the actual vehicle test results show that the prediction result of the model are accurate and can achieve better results.For the one-way three-lane unmanned vehicle lane changing scene,the factors affecting the unmanned vehicle’s lane changing intention,namely the accumulation of speed dissatisfaction and the vehicle distance,are analyzed,and the selection rule of the lane changing target lane is established.Then,this paper establishes the safe distance of lane-changing for unmanned vehicles.Based on safety constraints,comfort constraints,lane-changing efficiency constraints,traffic rules,and lane-changing intentions of the preceding vehicle,a reasonable autonomous vehicle lane-changing strategy is formulated,and analyzed the characteristics of commonly used lane-changing trajectories,selected a fifth-degree polynomial to plan the lane-changing trajectory of unmanned vehicles.At the same time,an objective function of optimal lane changing time is established according to evaluation factors such as safety,comfort,lane changing efficiency,etc.,so as to plan an optimal lane changing trajectory that satisfies the constraints.In order to make the unmanned vehicle change lanes according to the planned expected trajectory,this paper uses the model predictive control algorithm to track the unmanned vehicle’s lane change trajectory control.First,a two-degree-of-freedom vehicle dynamics model is established based on the assumption of a small front wheel deflection angle and a linear tire model,and the model is linearized and discretized to improve the controller’s real-time performance and vehicle stability.Then,based on the current state quantity,the control quantity at the previous time and the control increment,an optimized objective function with relaxation factor is established,and the problem is transformed into a quadratic programming problem to solve,and the vehicle front wheel angle control quantity is obtained to achieve Tracking control of unmanned vehicles when changing lanes.Finally,the Pre Scan/Car Sim/Simulink joint simulation verifies the effectiveness of the autonomous vehicle lane changing algorithm.