Research On Prediction Algorithm Of Front Vehicle Motion Based On Driving Intention Classification For Autonomous Vehicles

The driving prediction of surrounding vehicles has great potential to improve the driving safety,fuel consumption and traffic efficiency of the vehicle,and is widely valued.However,due to the uncertainty of its driving intention,vehicle dynamics characteristics,and the interaction between the predicted object and the surrounding environment,its driving prediction faces huge challenges.In order to break through the bottleneck in the prediction of the preceding vehicle movement,this paper proposes a preceding vehicle movement prediction algorithm based on the combination of driving intention recognition,environment and vehicle kinematics correction,and multi-model prediction results,in an effort to solve the unclear driving intention of the predicted object and the future of the vehicle.Non-linear problem between movement and environment.In response to the research and development needs of this article,first,a data acquisition and analysis platform was developed,and a 6.6m-long pure electric driverless bus was selected as the research object.Its operating route covers a variety of typical road conditions,and the design has been developed.An in-vehicle information terminal that can collect vehicle status information and vehicle environment information in real time,providing a good data basis for subsequent algorithm verification.Aiming at the problem that the longitudinal driving intention and vehicle dynamics characteristics of the target vehicle are not clear,a multi-model preceding vehicle speed prediction algorithm based on longitudinal driving intention classification is proposed.Using the fuzzy C-means method,using vehicle acceleration-related information,through offline training,the automatic recognition of longitudinal driving intentions is realized.The method of Gaussian process regression is used to predict future vehicle speeds under three different driving intentions by using historical and current vehicle speeds.According to the result of fuzzy classification of driving intention,the three kinds of predicted vehicle speeds are fused,and the rolling prediction of 1s is carried out through the iterative method.90.33% of the prediction errors are less than 1m/s,and the maximum speed prediction error is 2.45m/s.Aiming at the problems of lateral vehicle speed predicting driving intention,unclear vehicle dynamics and environmental interaction,a multi-model preceding vehicle trajectory prediction algorithm based on lateral driving intention recognition was developed.Based on the fuzzy C-means algorithm,using the information about the rate of change of the heading angle,the automatic recognition of lateral driving intentions is realized through offline training.Designing a trajectory prediction algorithm based on driving intention recognition,combining the future vehicle speed obtained in the previous article,developing a trajectory prediction algorithm based on long and short-term memory,and fusing the results of prediction models trained for different driving intentions to obtain the future trajectory of the vehicle.The artificial potential field method is used to calculate the longitudinal and lateral safety distance of the target vehicle,so as to summarize the potential field of the road boundary and the surrounding vehicles of the predicted object to achieve the effect of correcting the trajectory prediction result.The 1s rolling prediction is carried out by iterative method.Among the horizontal distances,94.52% of the prediction errors are less than 0.1m,and the maximum distance prediction error is 0.23 m.In the longitudinal distance,92.91%of the prediction errors are less than 0.5m,and the maximum distance prediction error is 0.87 m.In summary,the preceding vehicle motion prediction algorithm based on the combination of driving intention recognition,environment and vehicle kinematics correction,and multi-model prediction results proposed in this paper can effectively solve the predicted object driving intention,vehicle dynamics characteristics,and environmental interaction.Bottlenecks such as uncertainty have laid an important foundation for improving the driving safety,fuel consumption and traffic efficiency of the following vehicles.