KALMAN FILTER MODELING FOR TRAJECTORY PREDICTION OF CAR TRAVELING THROUGH STRAIGHT AND CURVED LANE BOUNDARIES

Abstract

Due to rise in fatalities and serious injures because of car accidents many people have paid attention to doing research on Intelligent Transport System (ITS). These researches especially deal with lane detection system using combination of machine vision with the fusion of sensors for obtaining data of road lane boundary detection and vehicle movement in every state to improve data from noisy environment. And hence these data can be used for future work whether for lane departure warning system or automatic car system. This paper proposes to develop a precised lane boundary estimation model which can predict the lateral distance while a vehicle is moving along the road lane boundary, whether curve or straight lane boundary, with that of a different vehicle’s movement including straight and circular lanes. This model is different from previous models which were only able to estimate if the lane is straight and flat only. The model is developed under assumption of noisy data due to environment and the Extended Kalman Filter (EKF) is applied to improve the accuracy of the data. Experiments were conducted to measure the efficiency of the model using RMSE. The results of our simulation indicated high performance because the model can reduce noise more than 50%. Moreover, the model can still smooth noisy data even if obtained data contains both noise and missing data. However, the estimation performance of the model decreases as the number of missing data increase.