Robust fusion of location information

Abstract

The purpose of this paper is to examine a sensor fusion problem for location data using statistical decision theory (SDT). The contribution of this paper is the application of SDT to obtain: (i) a robust test of the hypothesis that data from different sensors is consistent; and (ii) a robust procedure for combining the data which pass this preliminary consistency test. Here, robustness refers to the statistical effectiveness of the decision rules when the probability distributions of the observation noise and the a priori position information associated with the individual sensors are uncertain. Location data refers to observations of the form: 2 = B + V, where V represents additive sensor noise and B denotes the "sensed" parameter of interest to the observer. While the theory developed in this paper applies to several uncertainty classes, the focus of this presentation is on r-contamination models, which d o w one to account for heavytailed deviations from nominal sampling distributions.

A New View of Multisensor Data Fusion: Research on Generalized Fusion

“…Limited description ① Obtain (more comprehensive/complete and higher quality) [15,16] information that is greater than the sum of each contribution part ① Source limitation: that is, limiting the data source, such as data or information from sensors ② Accurately understand and describe the given scene [17] ③ Realize inferences that cannot be achieved with a single sensor [18] ② Scenario limitation: that is, the application type or decisionmaking situation is limited, such as decision-making with strict timing requirements ④ Infer events related to the observed object [15,19] ⑤ Improve state estimation, prediction, and risk assessment [13,16,20,21] ⑥ Realize precise positioning, tracking, and identification [20,[22][23][24] ⑦ Realize accurate (accuracy, robustness, qualitative, and quantitative) decision-making and action [23,25] ③ Characteristic limitation: that is, to limit the fusion characteristics, such as continuous refinement ⑧ Maximize useful information, improve reliability or recognition ability, and minimize the amount of retained data [14,[26][27][28] Feature information Feature information (2) According to the attributes of fusion data, multisensor data fusion can be divided into homogeneous data fusion and heterogeneous data fusion.…”

Sensor Fusion for Mobile Robot Navigation

“…10.2-3], [15]. 4) Robustness: Data consistency and data combining were further studied in a series of papers by Mintz and his co-workers on robust sensor fusion-using statistical decision theory [45], [56], [57]. Mintz et al have developed a robust test of the hypothesis that data from different sensors is consistent, and a robust procedure for combining the data which pass this preliminary test.…”

“…The observation model was where represents additive sensor noise and is the sensed parameter of interest. Robustness in the context of this work refers to "the statistical effectiveness of the decision rules when the probability distribution of the observations noise and the a priori position information associated with the individual sensors are uncertain" [45]. Mintz et al note that most studies of data fusion (such as the ones that we have quoted so far) assume that sensor noise can either be adequately modeled by Gaussian distributions with known means and covariances, or by distributions characterized only by specified first and second moment (using procedures that are equivalent to making Gaussian-distribution assumptions).…”

Estimation of the absolute position of mobile systems by an optoelectronic processor

“…where Fn and CYn are the estimated distance and the orientation of the mobile system, respectively; w, T and w: are the weights at point i , and rr and ay are the projections for the coordinates at point n, which are defined by equations similar to (15). Equation (18) can be expressed in the following recursive form:…”

“…where r," and a; are the new measurements, and ?=E-l and are the projections from the last estimate, i.e, Minimizing the Variance of the New Estimate: The algorithm described by (18) is not based on an optimization procedure. In contrast, our second algorithm is based on minimizing the variance of the new estimate.…”