System and method for vehicle trajectory prediction and trajectory deviation risk assessment

Abstract

The invention discloses a system and method for vehicle trajectory prediction and trajectory deviation risk assessment. Based on the vehicle dynamics model, a sensor filter fusion estimation algorithm and a Kalman predictor are designed to estimate and predict the motion and pose of the vehicle; combined Target trajectory, according to the actual trajectory controller, design the trajectory control algorithm for prediction, predict the next control action, input the Kalman predictor, obtain the next prediction result, then estimate and predict the vehicle model interference, further correct the trajectory prediction result, and repeat Until the trajectory prediction of the remaining steps is completed; according to the vehicle pose prediction and the target trajectory, calculate the predicted value of the distance between the vehicle corner point and the target trajectory lane line and the variance of the prediction error, and evaluate the risk of vehicle trajectory deviation. The invention can overcome the lack of accuracy caused by the strict limiting conditions of the traditional algorithm and the lack of consideration of closed-loop control, improve the accuracy of vehicle track prediction, and the obtained track deviation risk degree can be used as a basis for emergency intervention of automatic driving.

Description

technical field [0001] The invention belongs to the technical field of intelligent network-connected vehicles, and in particular relates to a system and method for predicting the driving trajectory of a vehicle and evaluating the risk of trajectory deviation. Background technique [0002] Driving in the expected lane position in a safe and comfortable state is the core task of driving, including the current Lane Departure Warning (LDW, Lane Departure Warning), Lane Keeping Assist (LKA, Lane KeepingAssist), and Lane Center Keeping Assist. (LCA, Lane CenteringAssist) and other advanced driving assistance systems (ADAS, Advanced Driving Assistance System) goals, is also one of the core goals of autonomous driving (ADS, Automated Driving system). In these ADAS and ADS systems, the estimation or prediction method of the vehicle's driving trajectory is the key for the system to realize deviation warning, deviation correction intervention and automatic tracking driving according to...

AI Technical Summary

Problems solved by technology

[0004] In comparison, the assumptions of the TLC-based method are too ideal. If the heading angle or vehicle speed of the vehicle changes within the prediction time domain, the decision-making result of the method will have errors; for the FOD-based method, although the method considers The driver's driving habits, experiments have also proved that this method can effectively reduce the false alarm rate of early warning, but this method is similar to the TLC method, based on the assumption that the vehicle speed remains constant, which may not match the reality; The method of size relationship, the effectiveness of this method depends on factors such as the pose of the vehicle, the shape of the lane line, etc., the versatility is poor, and it is impossible to predict the future situation

[0005] To sum up, the existing trajectory prediction methods have the above-mentioned deficiencies. In addition, they do not consider the closed-loop control algorithm carried by the vehicle's own ADAS / ADS system, including lane keeping control and lane tracking control. Therefore, there are serious problems in actual accuracy. Insufficient, it cannot be used as ADAS / ADS high-precision trajectory prediction, and it cannot be used to monitor whether the ADAS / ADS system is running safely

Images