Lockheed Martin (Australia)

Efficient wide-area surveillance of the maritime environment requires automated detection and classification techniques. Deep learning can be exploited for this purpose, but is susceptible to adversarial attacks that can cause misclassifications. Diffusion models have previously been shown to improve robustness to adversarial perturbation in electro-optic imagery. The aim of this work is to investigate how this adversarial defence technique protects against white-box attacks on a high-resolution Synthetic Aperture Radar (SAR) dataset. The benefits of the diffusion model are demonstrated using a maritime SAR dataset, and include an improvement in the classification performance and a reduction of the sea clutter.

Autonomous robots for gathering information on objects of interest has numerous real-world applications because of they improve efficiency, performance and safety. Realizing autonomy demands online planning algorithms to solve sequential decision making problems under uncertainty; because, objects of interest are often dynamic, object state, such as location is not directly observable and are obtained from noisy measurements. Such planning problems are notoriously difficult due to the combinatorial nature of predicting the future to make optimal decisions. For information theoretic planning algorithms, we develop a computationally efficient and effective approximation for the difficult problem of predicting the likely sensor measurements from uncertain belief states. The approach more accurately predicts information gain from information gathering actions. Our theoretical analysis proves the proposed formulation achieves a lower prediction error than the current efficient-method. We demonstrate improved performance gains in radio-source tracking and localization problems using extensive simulated and field experiments with a multirotor aerial robot.

It has been roughly a decade since the first papers using deep neural networks (DNNs) for radar applications were published. Deep learning has revolutionized almost every technical area, from computer vision and natural language processing to health, finance, and biology—any field where data can be analyzed to provide insight. However, in radar applications, deep learning faces unique challenges due to the phenomenology of radio frequency (RF) propagation that creates essential differences in the data itself and impacts the design of DNNs for radar signal analysis [1], [2], [3].

Reduced rank adaptive detection techniques are employed in airborne radar when the clutter returns have insufficient sample support due to the non-stationarity and non-homogeneity of the data. A number of approaches to combat this problem exist, including the single data set algorithms which eliminate the need for training data, and reduced rank detectors such as those employing the principal components, cross-spectral metric and the multistage Wiener filter (MWF). This latter approach is superior to other rank reduction techniques in terms of computational efficiency and sample support requirements. The benefit of the MWF is maximised when the number of stages is kept at a minimum. In this paper, we propose an algorithm to determine the number of stages that is formulated as a model order estimation problem and utilises the minimum description length (MDL). This overcomes the main drawback of the Ritz-value estimation which requires a threshold to be set. We give a computationally efficient implementation of the MDL and demonstrate its effectiveness using simulations.

With the current emphasis on Information Assurance and security of Automatic Test Equipment (ATE), security of the TPS is often not considered. Since Test Program Sets (TPSs) often come from disparate sources, it must be considered that they or items associated with them could become an attack vector into an ATE system. In addition, some TPSs include executables which are run during the execution of the TPS itself. Those executables themselves may also consist of external dependencies which may not be from the TPS developer such as Free Open-Source Software (FOSS). Therefore, it must be surmised that given the quantity of TPSs developed that it is theoretically possible for one to contain malware. This malware could compromise the ATE system that executes the TPS. One way to help control the vulnerability of the ATE is to run the TPS in some sort of sandbox or virtual machine to limit the access the TPS has to the ATE. This has the additional benefit of forcing the TPS developer to use the Application Programming Interfaces (APIs) provided by the ATE itself which is always a desired solution. All of these thing help control the TPS and provide a safer experience for the users of the ATE.

Semantic segmentation of aerial point cloud data can be utilised to differentiate which points belong to classes such as ground, buildings, or vegetation. Point clouds generated from aerial sensors mounted to drones or planes can utilise LIDAR sensors or cameras along with photogrammetry. Each method of data collection contains unique characteristics which can be learnt independently with state-of-the-art point cloud segmentation models. Utilising a single point cloud segmentation model can be desirable in situations where point cloud sensors, quality, and structures can change. In these situations it is desirable that the segmentation model can handle these variations with predictable and consistent results. Although deep learning can segment point clouds accurately it often suffers in generalisation, adapting poorly to data which is different than the training data. To address this issue, we propose to utilise multiple available open source fully annotated datasets to train and test models that are better able to generalise. In this paper we discuss the combination of these datasets into a simple training set and challenging test set. Combining datasets allows us to evaluate generalisation performance on known variations in the point cloud data. We show that a naive combination of datasets produces a model with improved generalisation performance as expected. We go on to show that an improved sampling strategy which decreases sampling variations increases the generalisation performance substantially on top of this. Experiments to find which sample variations give this performance boost found that consistent densities are the most important.

Earth and Space Science Open Archive PosterOpen AccessYou are viewing the latest version by default [v1]The Long Period of 3He-rich Solar Energetic Particles Measured by Solar Orbiter 2020 November 17–23AuthorsRadoslavBucikiDGlennMasonRaulGomez-HerreroiDDavidLarioiDLauraBalmacedaNariakiNittaVratislavKruparNinaDresingGeorgeHoiDRobertAllenFernandoCarcaboso-MoralesiDJavierRodriguez-PachecoFredericSchulleriDAlexanderWarmuthRobertWimmer-SchweingruberiDJohanFreiherr von ForstnerGAndrewsLarsBergerIgnacioCernudaiDFranciscoEspinosa LaraiDW.LeesCesarMartin-GarciaDanielPachecoManuelPrietoiDSebastianSanchez PrietoiDCharlesSchlemmHelmutSeifertKushTyagiMilanMaksimoviciDAntonioVecchioAlexanderKollhoffPatrickKühlZigongXuiDSandraEldrumSee all authors Radoslav BucikiDCorresponding Author• Submitting AuthorSouthwest Research InstituteiDhttps://orcid.org/0000-0001-7381-6949view email addressThe email was not providedcopy email addressGlenn MasonJohns Hopkins University Applied Physics Laboratoryview email addressThe email was not providedcopy email addressRaul Gomez-HerreroiDUniversity of AlcaláiDhttps://orcid.org/0000-0002-5705-9236view email addressThe email was not providedcopy email addressDavid LarioiDNASA Goddard Space Flight CenteriDhttps://orcid.org/0000-0002-3176-8704view email addressThe email was not providedcopy email addressLaura BalmacedaGeorge Mason University Fairfaxview email addressThe email was not providedcopy email addressNariaki NittaLockheed Martin Advanced Technology Centerview email addressThe email was not providedcopy email addressVratislav KruparUniversity of Maryland Baltimore Countyview email addressThe email was not providedcopy email addressNina DresingUniversity of Turkuview email addressThe email was not providedcopy email addressGeorge HoiDJohns Hopkins University Applied Physics LaboratoryiDhttps://orcid.org/0000-0003-1093-2066view email addressThe email was not providedcopy email addressRobert AllenJohns Hopkins University Applied Physics Laboratoryview email addressThe email was not providedcopy email addressFernando Carcaboso-MoralesiDUniversity of AlcaláiDhttps://orcid.org/0000-0003-1758-6194view email addressThe email was not providedcopy email addressJavier Rodriguez-PachecoUniversity of Alcaláview email addressThe email was not providedcopy email addressFrederic SchulleriDLeibniz Institute for Astrophysics PotsdamiDhttps://orcid.org/0000-0002-2609-1604view email addressThe email was not providedcopy email addressAlexander WarmuthLeibniz Institute for Astrophysics Potsdamview email addressThe email was not providedcopy email addressRobert Wimmer-SchweingruberiDUniversity of KieliDhttps://orcid.org/0000-0002-7388-173Xview email addressThe email was not providedcopy email addressJohan Freiherr von ForstnerUniversity of Kielview email addressThe email was not providedcopy email addressG AndrewsJohns Hopkins University Applied Physics Laboratoryview email addressThe email was not providedcopy email addressLars BergerUniversity of Kielview email addressThe email was not providedcopy email addressIgnacio CernudaiDUniversity of AlcaláiDhttps://orcid.org/0000-0001-8432-5379view email addressThe email was not providedcopy email addressFrancisco Espinosa LaraiDUniversity of AlcaláiDhttps://orcid.org/0000-0001-9039-8822view email addressThe email was not providedcopy email addressW. LeesJohns Hopkins University Applied Physics Laboratoryview email addressThe email was not providedcopy email addressCesar Martin-GarciaUniversity of Kielview email addressThe email was not providedcopy email addressDaniel PachecoUniversity of Kielview email addressThe email was not providedcopy email addressManuel PrietoiDUniversity of AlcaláiDhttps://orcid.org/0000-0003-3050-3445view email addressThe email was not providedcopy email addressSebastian Sanchez PrietoiDUniversity of AlcaláiDhttps://orcid.org/0000-0002-6729-7932view email addressThe email was not providedcopy email addressCharles SchlemmJohns Hopkins University Applied Physics Laboratoryview email addressThe email was not providedcopy email addressHelmut SeifertJohns Hopkins University Applied Physics Laboratoryview email addressThe email was not providedcopy email addressKush TyagiUniversity of Colorado at Boulderview email addressThe email was not providedcopy email addressMilan MaksimoviciDLESIA, Observatoire de Paris, Université PSL, CNRS, Sorbonne Université, Université de ParisiDhttps://orcid.org/0000-0001-6172-5062view email addressThe email was not providedcopy email addressAntonio VecchioLESIA, Observatoire de Paris, Université PSL, CNRS, Sorbonne Université, Université de Parisview email addressThe email was not providedcopy email addressAlexander KollhoffUniversity of Kielview email addressThe email was not providedcopy email addressPatrick KühlUniversity of Kielview email addressThe email was not providedcopy email addressZigong XuiDUniversity of KieliDhttps://orcid.org/0000-0002-9246-996Xview email addressThe email was not providedcopy email addressSandra EldrumUniversity of Kielview email addressThe email was not providedcopy email address