New Filtering Method for Trajectory Measurement Errors and Its Comparison with Existing Methods

Abstract

Dynamic traffic simulation tools are increasingly being used to help traffic managers and urban planners to make decisions. Therefore, simulation tool users require a validated methodology guaranteeing that simulation results can be trusted. This study contributes to the identification and correction of a possible deficiency in detailed calibration and validation of car-following models: the data errors of individual trajectory data. Some studies addressed the problem of filtering trajectory data. A new filtering technique to reduce the measurement errors on trajectories, speed profiles, and acceleration profiles is proposed here. This technique is based on some piecewise polynomials termed “splines.” The proposed technique is compared with a set of filtering techniques found in the literature. A complete trajectory data set available within the NGSIM program is used. As a quality indicator of the various filtering techniques, velocity distribution, acceleration distribution, and jerk analysis are used for the whole data set. Also, analyzing acceleration standard deviations for each trajectory of the data set is suggested. The main findings are as follows: (a) of the methods compared within this work, the I-spline method with the action points most reduces the spikes in the velocity distribution; (b) moreover, the I-spline method most reduces the percentage of jerk values higher than 15 m/s3 as well as the percentage of the 1-s windows with more than one sign inversion of the jerk; and (c) in some cases, this method increases the acceleration variability of smoothed trajectories.