Trajectory Prediction For Vehicle In The Slope Condition

Predicting vehicle trajectory accurately is one of the most crucial task for anautonomous vehicle or intelligent control system. The previous trajectory predictionalgorithm mostly is carried out in the level road. When driving in the large longitudinalslope conditions, such as viaduct or mountain road, prediction result will appear largerdeviation. Therefore, a reasonable and efficient algorithm of vehicle trajectory predicting cannot only describe vehicle trajectory well on a level road in good condition, but also playswell in a slope condition. Only in this way, can we reduce the interference of the expectedthe unreasonable trajectory point, so as to ensure the validity of the analysis and evaluationin the future.This paper mainly studies the vehicle trajectory prediction algorithm in the longitudinalslope conditions. Firstly, the longitudinal slope of the road accurately and reliably need to beacquired. At the same time, to meet the real-time requirements of intelligent control system,road slope recognition algorithm also need to guarantee the real-time performance. Theaccuracy, real-time, and reliability of the road slope estimation are particularly important forintelligent vehicle in any condition. However, current estimation method is not able to satisfythe demands. Compared with the traditional estimation method, we propose an architectureof longitudinal slope estimation, which meets the requirement. In this method, we analysisthe advantage and dis advantage of the traditional estimation method, and then adoptInteracting Multiple Model and Probability Data Association Filter to get the effectiveinformation of slope for trajectory prediction algorithm. Without considering the roadconditions, the previous trajectory prediction algorithm is mainly based on the rigid bodykinematics. This paper put forward a trajectory prediction algorithm according to vehiclemotion characteristics and the spatial Multi-body dynamics theory.Finally, the main contents of intelligent vehicle predicting system were realized andcertificated. Firstly, we use CARSIM-SIMULINK to test the slope estimation algorithm. Andthen, the developed algorithm is verified and evaluated through experimental and casestudies using a real-time embedded system. Secondly, the algorithm of vehicle trajectoryprediction is identified using simulation and experiments compared to GPS/INSnavigation result. The results show that the performance and efficiency of the road slope estimationalgorithm is accurate and reliable enough for autonomous vehicle application. And theprediction trajectory shows great similarities to the actual location andnavigation information using GPS system, which proves trajectory prediction algorithm inthe longitudinal slope condition accurately and effectively.