Among them, the vehicle trajectory prediction technology can predict the vehicle position, speed, and other motion states in the predicted period according to the current and historical vehicle running state, and the prediction results can provide support for judging the vehicle safety in the predicted period.
With the development of technology, vehicle trajectory prediction and safety decision technology has become an important part of active safety technology.
conducted overtaking experiments at 30 established the straight track function of surrounding vehicles by using the quintic polynomial and predicted the collision time between the surrounding vehicles and the main vehicle on the straight and curve, respectively, by using the parameter track curve. carried out research on driver's lane-changing intention identification and realized the prediction of driver's lane-changing intention 0.6s in advance, with a recognition accuracy up to 89%. Based on SVM classification algorithm and based on steering angle, accelerator brake pedal, vehicle speed, acceleration and driver's line-ofsight information, Gao et al. established the MGHMM model for driver lane-changing intention recognition, determined the state number and Gaussian mixture number of the model through experimental analysis, and analyzed the relationship between the time point of model recognition of lanechanging intention and observation data interception method and observation data category selection under single and compound working conditions. The experimental results show that this method can effectively improve the efficiency of autonomous driving strategy learning and control the virtual vehicle for autonomous driving behavior decisions, and provide reliable theoretical and technical support for real vehicles in autonomous driving decision-making. Under different reward functions, the method in this paper obtains the highest cumulative reward value within 500 s, which improves 69 points compared with the reward function method based on the artificial potential field method, and has higher adaptability and robustness in different environments. The simulation results show that the completion rate of the virtual vehicle in the obstacle environment that generates penalty feedback is as high as 96.3%, which is 3.8% higher than the completion rate in the environment that does not generate penalty feedback. Finally, the APF-DPPO learning model is selected to train the driving strategy for the virtual vehicle, and the transfer learning method is selected to verify the comparison experiment. To solve the range repulsion problem of the artificial potential field method, which affects the optimal driving strategy, this paper proposes a directional penalty function method that combines collision penalty and yaw penalty to convert the range penalty of obstacles into a single directional penalty, and establishes the vehicle motion collision model. The ideas of target attraction and obstacle rejection of the artificial potential field method are introduced into the distributed proximal policy optimization algorithm, and the APF-DPPO learning model is established. To address the difficulty of obtaining the optimal driving strategy under the condition of a complex environment and changeable tasks of vehicle autonomous driving, this paper proposes an end-to-end autonomous driving strategy learning method based on deep reinforcement learning.
0 kommentar(er)
