Comparative Analysis of Fusion Algorithms in a Loosely-Coupled Integrated Navigation System on the Basis of Real Data Processing

Bitar, N. Al; Gavrilov, A. I.

doi:10.17285/0869-7035.0001

Cited by 6 publications

(3 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Several authors use the term “information fusion” to refer to what we would consider standard statistical or machine-learning modeling steps, using Bayesian model averaging, 49 elastic-net regularization, 50 or Kalman filters. 51 , 52 This differing use of vocabulary is by no means incorrect, but underscores the difficulty encountered when summarizing a research area that is only nominally united and exceedingly broad. In contrast, in this study we specifically identify system fusion as the combination of model outputs involving no additional/subsequent data-driven training.…”

Section: Discussionmentioning

confidence: 99%

Ranks underlie outcome of combining classifiers: Quantitative roles for and

et al. 2022

View full text Add to dashboard Cite

Combining classifier systems potentially improves predictive accuracy, but outcomes have proven impossible to predict. Classification most commonly improves when the classifiers are ''sufficiently good'' (generalized as ''ACCURACY'') and ''sufficiently different'' (generalized as ''DIVERSITY''), but the individual and joint quantitative influence of these factors on the final outcome remains unknown. We resolve these issues. Beginning with simulated data, we develop the DIRAC framework (DIVERSITY of Ranks and ACCURACY), which accurately predicts outcome of both score-based fusions originating from exponentially modified Gaussian distributions and rank-based fusions, which are inherently distribution independent. DIRAC was validated using biological dual-energy X-ray absorption and magnetic resonance imaging data. The DIRAC framework is domain independent and has expected utility in far-ranging areas such as clinical biomarker development/personalized medicine, clinical trial enrollment, insurance pricing, portfolio management, and sensor optimization.THE BIGGER PICTURE It can be advantageous to combine multiple predictive models for power or robustness, but it is recognized that realizing these potential gains cannot be guaranteed, especially when the input models cannot be appropriately weighted a priori or the resulting fusion models cannot be cross-validated. This, and a series of mathematically related problems in different guises, fundamentally limits the ability to optimally use all available models to improve classification across essentially all domains in which more than one potentially useful model exists. We show that any fusion's outcome is fully predictable/explicable given characterization of the models to be fused. The ''mechanism'' described acts at the level of ranks, not scores, which extends our findings to all distributions and, functionally, to any domain of interest. We are elucidating the underlying math, following the framework's implications for data science, and using this approach on real-world problems.Concept: Basic principles of a new data science output observed and reported

show abstract

Section: Discussionmentioning

confidence: 99%

Ranks underlie outcome of combining classifiers: Quantitative roles for and

et al. 2022

View full text Add to dashboard Cite

show abstract

“…The solution of the estimation problem can also be simplified by searching for estimates within some limited class, for example, the class of linear estimates [23,24]. Among other procedures used in this field, we should mention unscented transformation (Unscented KF, UKF) [25][26][27] and cubature formulas [28].…”

Section: Introductionmentioning

confidence: 99%

“…It should be noted that nonlinearities in state and measurement equations are often represented by polynomials. From this point of view, polynomial filtering methods have been quite successfully used to solve estimation problems [25][26][27][28][29][30][31][32][33][34][35][36][37][38]. In particular, in [24,31], the authors consider problems whose specific feature is nonlinearity of measurements, whereas the equations of the shaping filter that describe the estimated sequence are linear.…”

Section: Introductionmentioning

confidence: 99%

Efficiency analysis of polynomial filtering algorithms in navigation data processing for a class of nonlinear discrete dynamical systems

Степанов

Vasiliev

Tupysev

et al. 2020

IET Control Theory & Appl

View full text Add to dashboard Cite

A polynomial filtering algorithm is designed for estimating a Markov sequence based on linear measurements. The feature of the estimation problem is that the Markov sequence is described by a nonlinear shaping filter, which is a second-order polynomial with respect to the state vector components. The algorithm efficiency is illustrated by three examples of navigation data processing. It is shown that the polynomial filter provides accuracy close to the best potential one calculated using the particle filter. At the same time, the amount of computation required to implement this algorithm is much smaller than that for the particle filter. In addition, the polynomial filter provides a consistent calculated accuracy characteristic.

show abstract

Strapdown inertial navigation systems readings correction based on navigational data of other sensors and systems with intelligent selection of the priority adjuster

2020

View full text Add to dashboard Cite

Strapdown inertial navigation systems (SINS) are one of the main components of the navigation systems of the drones and aircraft (including autonomous ones), but their readings need to be instanly corrected due to the constant accumulation of errors. This paper comprises the review of existing approaches to using one or more sensors or systems to correct the navigation data of SINS algorithms (herein after – correctors) using integrated information processing. A common disadvantage of the analysed approaches is the lack of flexibility concerning the types and the number of SINS correctors used, as well as the growth of computational burden due to the use of the measurement vectors of all the correctors in the process of forming the state vector of the system. This article proposes the use of the original adaptive scheme based on the selection of the least noisy data, taking into account environmental conditions, for the integrated processing of the SINS and the correctos’ navigation parameters. The essence of the approach is that the state vector is estimated on the basis of the most reliable corrector. This allows reducing the correlation of errors in the correctors’ measurement of navigational parameters, since only the measurement vectors (or vector) with best navigational data signal/noise ratio (received from the corresponding correctors) are used in forming the state vector. Furthermore, the proposed navigational data fusion scheme has a modular structure and greater flexibility in comparison with the loosely coupled systems, and also implies the use of an arbitrary number of correction sensors and systems regardless of the physical nature of their measurements.

show abstract

Comparative Analysis of Fusion Algorithms in a Loosely-Coupled Integrated Navigation System on the Basis of Real Data Processing

Cited by 6 publications

References 0 publications

Ranks underlie outcome of combining classifiers: Quantitative roles for and

Ranks underlie outcome of combining classifiers: Quantitative roles for and

Efficiency analysis of polynomial filtering algorithms in navigation data processing for a class of nonlinear discrete dynamical systems

Strapdown inertial navigation systems readings correction based on navigational data of other sensors and systems with intelligent selection of the priority adjuster

Contact Info

Product

Resources

About