Design and Evaluation of Robust M-estimators for GNSS Positioning in Urban Environments

Abstract

Lausanne (EPFL) BIOGRAPHIES Omar García Crespillo holds a M.Sc. in Telecommunication Engineering from the University of Malaga in Spain. In 2013, he joined the Navigation department of the German Aerospace Center (DLR) where his current field of research includes multi-sensor fusion algorithms, GNSS, inertial sensors and integrity monitoring for safe ground and air transportation systems. Since 2015, he is also a PhD student at the Swiss Federal Institute of Technology (EPFL) in Lausanne. Alice Andreetti is currently a last year Master student in Environmental Engineering at the Swiss Federal Institute of Technology in Lausanne (EPFL). She chose the specialization in environmental monitoring and modeling and she is currently working on the master thesis in the field of GNSS positioning. She received in 2017 the bachelor degree from the same university in Environmental Science and Engineering. In 2019 she stayed at the German Aerospace Center (DLR) for a 4 months internship.

Robust Filtering Techniques for RTK Positioning in Harsh Propagation Environments

“…Baseband processing Interference mitigation [11][12][13][14] PVT processing Snapshot code-based positioning [15][16][17][18][19] Recursive code-based positioning [20][21][22] Recursive RTK/PPP [23,24] Variational Inference…”

“…These functions present a tuning parameter that controls efficiency in the normal case (i.e., when all the observations are normally distributed) or, in other words, their sensitivity in detecting outliers. An interested reader might refer to [19,42] for more details on classical and modern robust functions. Within the robust statistics-based filtering solutions, one can distinguish between resilience against outliers in the correction step (for robust information filters) or against outliers in both the prediction and correction steps (for the generalized M-estimator KF).…”

“…These functions present a tuning parameter that controls efficiency in the normal case (i.e., when all the observations are normally distributed) or, in other words, their sensitivity in detecting outliers. An interested reader might refer to [ 19 , 42 ] for more details on classical and modern robust functions.…”

Design and Implementation of Opportunity Signal Perception Unit Based on Time-Frequency Representation and Convolutional Neural Network

“…However, it also has some problems: (1) The signal landing power is about −130 dBm, which is easy to be interfered and spoofed. (2) The signal is easy to be blocked by obstacles, making it difficult to use in dense urban areas or indoor environments [ 1 ]. In perspective of the above-mentioned problems of GNSS, an ever-increasing amount of researchers have begun to explore reliable positioning methods that do not rely on GNSS systems.…”

Review of Factor Graphs for Robust GNSS Applications

“…M-estimators [26] have also been recently tested within the GNSS framework in batch form [27] and found to perform better than non-robust estimators. Mestimators assume a loss function that is different from the squared loss function.…”