文章基本信息

标题：Utilizing Hidden Markov Models to Classify Maneuvers and Improve Estimates of an Unmanned Aerial Vehicle
本地全文：下载
作者：Amy K. Strong ; Scott M. Martin ; David M. Bevly 等
期刊名称：IFAC PapersOnLine
印刷版ISSN：2405-8963
出版年度：2021
卷号：54
期号：20
页码：449-454
DOI：10.1016/j.ifacol.2021.11.214
语种：English
出版社：Elsevier
摘要：AbstractEstimating the states of a Unmanned Aerial Vehicle (UAV) without the use of onboard sensors can be difficult, particularly if the UAV is performing high dynamic maneuvers. This paper examines if data driven modelling can assist in estimating UAV states, as well as classification of UAV maneuvers. A standard Extended Kalman Filter (EKF) that uses radar measurements and a constant acceleration dynamic model is used as the baseline estimation technique for dynamic UAV maneuvers. The UAV maneuvers are then modelled as Hidden Markov Models (HMM), which classify maneuvers and generate additional state information in the form of acceleration and jerk estimates. These HMM estimates are incorporated into an EKF to create a fusion EKF+HMM. This paper evaluates the robustness of the HMM classification accuracy and compares the EKF+HMM to a standard EKF using both simulated and experimental data.
关键词：KeywordsEstimationFilteringTime Series ModellingMechanicalAerospace EstimationHidden Markov ModelGaussian Mixture Model