447 research outputs found
Security mechanisms in Prolog-based IMAGO system
No full textA mobile agent is a piece of software which is able to migrate and execute on a remote host. The host may accept the agents without knowing the result in advance of executing the agents. Malicious mobile agents or malicious third parties may launch attacks and do harm to the host. In addition, poorly coded agents in some cases may also exhibit malicious behaviors to the host. Mobile agents are coded in a variety of scripting or interpreted languages, and by simply using static semantic analysis for pre-detecting potential hazards cannot provide a generic solution. To effectively discern, and thereby protect the hosts against such potential threats, is a challenge in the mobile agent research community. Our research at IMAGO (an intelligent mobile agent system) group, is seeking ways of protecting hosts against such possible threats with regard to the system resources available. This thesis presents a mobile agent system called IMAGO (Intelligent Mobile Agent Gliding Online) System and the related security architecture. It introduces the ways of protecting hosts against such possible threats. It also demonstrates how the security services are integrated into the IMAGO system and achieve an insignificant overhead by the means of several innovative designs and techniques
6D Rigid Object Pose Estimation Using Deep Learning
No full text6D object pose estimation is the task of determining an object’s 3D rotation and translation with respect to a camera, and plays a critical role in applications such as robotic manipulation, autonomous navigation, and augmented reality. While recent advances in deep learning have substantially improved performance, many existing methods still face limitations in learning robust and generalizable representations. Factors such as variations in object appearance, occlusion, sensor noise, and domain shifts can degrade model accuracy, highlighting the need for more effective representation learning strategies that capture rich geometric and semantic cues for reliable pose estimation across diverse conditions.
This dissertation investigates 6D pose estimation from geometric information, emphasizing the development of robust and generalizable representation learning techniques to address both instance-level and category-level settings. Instance-level pose estimation refers to the task of predicting the 6D pose of specific, known objects for which exact 3D models are available during both training and testing. For instance-level pose estimation, this dissertation presents a depth-only fusion framework that converts depth images into normal vector angle maps to explicitly embed geometric cues, and combines them with point cloud features for accurate 3D keypoint localization and semantic segmentation. This approach achieves state-of-the-art performance on the LineMod and Occlusion-LineMod datasets, and delivers competitive results on YCB-Video without post-processing.
Category-level pose estimation, on the other hand, aims to estimate 6D poses for previously unseen object instances that belong to a predefined category. For category-level pose estimation, this dissertation introduces a contrastive learning framework that learns pose-aware point cloud representations while preserving the intrinsic continuity of 6D poses. Specifically, we present two frameworks: the first is a two-phase one that combines pose-aware and geometry-aware representations to estimate target object poses, and the second is an end-to-end hierarchical ranking contrastive learning architecture, eliminating the need for a separate geometric encoder and enhancing the pose estimation modules. The resulting model achieves state-of-the-art accuracy among depth-only methods on the REAL275 and CAMERA25 datasets, while maintaining real-time inference speed.
In addition, we conduct an exploratory study on applying diffusion-based generative modeling to category-level pose estimation. The method generates canonical partial-view point clouds from observed depth-based point clouds before estimating poses via the Umeyama algorithm. While preliminary results reveal limitations in generation fidelity and pose consistency, the study highlights key challenges and opportunities for integrating generative models into pose estimation pipelines.
Overall, this dissertation contributes novel geometric representation learning frameworks for both instance-level and category-level 6D pose estimation, supported by extensive experiments on widely used benchmarks. The findings not only advance geometric-based pose estimation methods but also open pathways toward unified, generative–discriminative approaches for robust object pose estimation in real-world environments. Such capabilities are critical for enabling reliable robotic manipulation in cluttered or unstructured settings, enhancing perception for autonomous navigation in dynamic scenes, and improving interaction in augmented and mixed reality systems. By bridging fundamental representation learning with practical deployment, this work moves closer to making 6D pose estimation an integral component of real-world intelligent systems
Recommended from our members
On Shallow Cumulus Cold Pools
No full textObservations of organized precipitating trade-wind cumuli show convective invigoration at their cold pool boundaries. Convection and cold pools are studied using nested-WRF simulations of a day from the Rain in Cumulus over the Ocean experiment. The sub-cloud updrafts downwind and near the cold pool boundaries are statistically compared to updrafts not associated with cold pools. Cold pools affect cloud growth through altering the thermodynamic properties of the updrafts to include moister, cooler air and by enhancing the updraft speed; the latter is particularly important for encouraging deeper, high-liquid water path clouds. It is examined in this study how different microphysics schemes affect the characteristics of simulated trade-wind cumulus convection and cold pools. The initial simulation is compared to an additional nested-WRF simulation using a different microphysics scheme. The differences in autoconversion and accretion rate lead to different precipitation production. The parameterization of raindrop terminal fall speed causes the discrete evaporation efficiencies and different surface cold pool properties under the same surface rainrate. The shallow cumulus cold pools observed in trade-wind region are compared to observation over the tropical Indian Ocean. Data from the Dynamics of the Madden-Julian Oscillation field campaign is analyzed to identify and characterize shallow cumulus cold pools. For the same surface temperature depression in the cold pools, the cold pools in the tropical Indian Ocean show smaller changes in surface specific humidity and wind speed. The comparisons of cold pool characteristics between these two regions help understand the sensitivity of cold pool generation to environmental conditions.</p
Emerg Infect Dis
No full textDuring February 2011-June 2012, invasive infection with Neisseria meningitidis serogroup W was identified in 11 persons in southeastern China. All isolates tested had matching or near-matching pulsed-field gel electrophoresis patterns and belonged to multilocus sequence type 11. The epidemiologic investigation suggested recent transmission of this clonal complex in southeastern China
Emerg Infect Dis
No full textSerogroup B Neisseria meningitidis strains belonging to sequence type 4821 clonal complex (CC4821), a hyperinvasive lineage first identified for serogroup C in 2003, have been increasingly isolated in China. We characterized the outer membrane protein genes of 48 serogroup B and 214 serogroup C strains belonging to CC4821 and analyzed the genomic sequences of 22 strains. Four serogroup B strains had porin A (i.e., PorA), PorB, and ferric enterobactin transport (i.e., FetA) genotypes identical to those for serogroup C. Phylogenetic analysis of the genomic sequences showed that the 22 CC4821 strains from patients and healthy carriers were unevenly clustered into 2 closely related groups; each group contained serogroup B and C strains. Serogroup B strains appeared variable at the capsule locus, and several recombination events had occurred at uncertain breakpoints. These findings suggest that CC4821 serogroup C N. meningitidis is the probable origin of highly pathogenic CC4821 serogroup B strains
Dynamic transfer partial least squares for domain adaptive regression
No full textThe traditional soft sensor models are based on the independent and identical distribution assumption, which are difficult to adapt to changes in data distribution under multiple operating conditions, resulting in model performance deterioration. The domain adaptive transfer learning methods learn knowledge in different domains by means of distribution alignment, which can reduce the impact of data distribution differences, and effectively improve the generalization ability of the model. However, most of the existing models established by domain adaptation methods are static models, which cannot reflect the dynamic characteristics of the system, and have limited prediction accuracy when applied to dynamic system modeling under multiple operating conditions. The dynamic system modeling methods can effectively extract the dynamic characteristics of the data, but they cannot deal with the concept drift problem caused by the change of data distribution. This paper proposes a new dynamic transfer partial least squares method, which maps the high-dimensional process data into the low-dimensional latent variable subspace, establishes the dynamic regression relationship between the latent variables and the labels, and realizes the systematic dynamic modeling, at the same time, the model adds regular terms for distribution alignment and structure preservation, which realizes dynamic alignment of data distribution difference. The effectiveness of the proposed method is validated on three publicly available industrial process datasets.Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Transport Engineering and Logistic
Spatial defects nanoengineering for bipolar conductivity in MoS2
No full textUnderstanding the atomistic origin of defects in two-dimensional transition metal dichalcogenides, their impact on the electronic properties, and how to control them is critical for future electronics and optoelectronics. Here, we demonstrate the integration of thermochemical scanning probe lithography (tc-SPL) with a flow-through reactive gas cell to achieve nanoscale control of defects in monolayer MoS2. The tc-SPL produced defects can present either p- or n-type doping on demand, depending on the used gasses, allowing the realization of field effect transistors, and p-n junctions with precise sub-μm spatial control, and a rectification ratio of over 104. Doping and defects formation are elucidated by means of X-Ray photoelectron spectroscopy, scanning transmission electron microscopy, and density functional theory. We find that p-type doping in HCl/H2O atmosphere is related to the rearrangement of sulfur atoms, and the formation of protruding covalent S-S bonds on the surface. Alternatively, local heating MoS2 in N2 produces n-character
An analysis of pion photoproduction
No full textA partial-wave analysis of pion photoproduction data up to a photon lab energy of 1.8 GeV has been performed. Both energy-dependent and energy-independent solutions have been obtained. The energy-dependent parametrization incorporates the recently determined elastic pion nucleon scattering amplitudes in such a way as to satisfy unitarity and utilize the resonance structure contained in the pion nucleon elastic amplitudes. Starting from the energy-dependent solution, energy-independent partial-wave solutions are obtained at a set of energies from threshold to 1.8 GeV. The data base used in the analysis contains 11,911 data from the reactions. The predictions of our solution are compared with the experimental data and previous analyses. Suggestions are made for future experiments.
A total of sixteen resonances exist in the energy range from threshold to 1.8 GeV. These resonance states are studied using our energy-independent solutions. Photon decay couplings to the sixteen resonances are extracted. These couplings are also compared with previous solutions and quark model predictions.Ph. D
- …
