Doppler velocity estimation based on the Random Sample Consensus considering a prior dynamic model
Jing Guo,Ziyu Fan,Zhen Li,Qile Zhao
DOI: https://doi.org/10.1016/j.asr.2024.12.027
IF: 2.611
2024-12-14
Advances in Space Research
Abstract:Typically, centimeters per second (cm/s) accuracy can be achieved for velocity estimation based on time-differenced positioning, Doppler velocity, and the time-differenced carrier phase approach in open air conditions. However, velocity estimation performance can deteriorate due to non-line-of-sight (NLOS) signals, multipath effects, and other sources of interference in urban areas. We discuss a velocity estimation method based on the RANdom SAmple Consensus (RANSAC) algorithm and a prior dynamic model is proposed. RANSAC is used to isolate the observations affected by NLOS or multipath, while a prior dynamic model is implemented to provide proper velocity constraint. Vehicle kinematic experiments conducted in both open and complex urban environments demonstrate that the velocity accuracy in all three directions can be improved by over 40% and centimeter-level velocity accuracy can be obtained in open areas. Moreover, the accuracy can be improved by 78.5%, 78.5%, and 68.4% in the north, east, and up directions compared to that obtained by traditional single-point Doppler velocity, and 10.5 cm/s, 11.0 cm/s, and 14.3 cm/s accuracies are achieved in complex urban areas. Moreover, this method is also reasonably applicable to low-cost receivers.
geosciences, multidisciplinary,meteorology & atmospheric sciences,astronomy & astrophysics,engineering, aerospace