Search

We prove that quasi-static pushing with a sticking contact and ellipsoid approximation of the limit surface is differential flat. Both graphical and algebraic derivations are given. A major conclusion is that the pusher–slider system is reducible to the Dubins car problem where the sticking contact constraints translate to bounded curvature. Planning is as easy as computing a Dubins curve with the additional benefit of time-optimality. For trajectory stabilization, we design closed-loop control using dynamic feedback linearization or open-loop control using two contact points as a form of mechanical feedback. We conduct robotic experiments using objects with different pressure distributions, shape, and contact materials placed at different initial poses that require difficult switching action maneuvers to the goal pose. The average error is within 1.67 mm in translation and 0.5° in orientation over 60 experimental trials. We also show an example of pushing among obstacles using a RRT planner with exact steering.

Recent progress in computer vision has been driven by high-capacity deep convolutional neural network (CNN) models trained on generic large datasets. However, creating large datasets with dense pixel-level labels is extremely costly. In this paper, we focus on the problem of instance segmentation for robotic manipulation using rich image and depth features. To avoid intensive human labeling, we develop an automated rendering pipeline for rapidly generating labeled datasets. Given 3D object models as input, the rendering pipeline produces photorealistic images with pixel-accurate semantic label maps and depth maps. The synthetic dataset is then used to train an RGB-D segmentation model by extending the Mask R-CNN framework for depth input fusion. Our results open up new possibilities for advancing robotic perception using cheap and large-scale synthetic data.

We propose a polynomial model for planar sliding mechanics. For the force–motion mapping, we treat the set of generalized friction loads as the 1-sublevel set of a polynomial whose gradient directi...

How can a robot design a sequence of grasping actions that will succeed despite the presence of bounded state uncertainty and an inherently stochastic system? In this letter, we propose a probabilistic algorithm that generates sequential actions to iteratively reduce uncertainty until object pose is uniquely known (subject to symmetry). The plans assume encoder feedback that gives a geometric partition of the post-grasp configuration space based on contact conditions. An offline planning tree is generated by interleaving computationally tractable open-loop action sequence search and feedback state estimation with particle filtering. To speed up planning, we use learned approximate forward motion models, sensor models, and collision detectors. We demonstrate the efficacy of our algorithm on robotic experiments with more than 3000 grasp sequences using different object shapes, press-ure distributions, and gripper materials where the uncertainty region is comparable to the size of the object in translation and with no information about orientation.

A high-fidelity and tractable mechanics model of physical interaction is essential for autonomous robotic manipulation in complex and uncertain environments. This thesis studies several aspects of harnessing task mechanics for robotic pushing and grasping operations: mechanics model learning, pose and model uncertainty reduction, and planning and control synthesis in the minimal coordinate space. We begin with a study of the mechanics of planar sliding as captured by a proposed efficiently identifiable convex polynomial force-motion model. Given a position-controlled manipulator action, we derive a kinematic contact model that resolves the contact modes and instantaneous object motion. This enables a generic quasi-static planar contact simulation validated with extensive robotic grasping and pushing experiments. Based on the contact model, we prove that quasi-static pushing with a sticking contact is differentially flat using both graphical and algebraic derivations. The pusher-slider system is reducible to the Dubins car problem where the sticking contact constraints translate to bounded curvatures. For trajectory stabilization in the presence of model uncertainty, we design closed-loop control using dynamic feedback linearization or open-loop control using two contact points as a form of mechanical feedback. To reduce initial pose perception uncertainty, we propose a probabilistic algorithm that generates sequential uncertainty reduction actions based on finger encoder feedback until the object pose is uniquely known (subject to symmetry). Going beyond position-controlled planar manipulation, we demonstrate grasping a large-size book where the robot needs to perform a sequence of non-prehensile contact-rich actions to finish the task while obeying kinematic and hardware limitations. Such actions include reorienting the book by force-controlled twisting and pulling, and push-grasping in the gravity plane treating the table as a third supporting finger.

Based on the convex force-motion polynomial model for quasi-static sliding, we derive the kinematic contact model to determine the contact modes and instantaneous object motion on a supporting surface given a position controlled manipulator. The inherently stochastic object-to-surface friction distribution is modelled by sampling physically consistent parameters from appropriate distributions, with only one parameter to control the amount of noise. Thanks to the high fidelity and smoothness of convex polynomial models, the mechanics of patch contact is captured while being computationally efficient without mode selection at support points. The motion equations for both single and multiple frictional contacts are given. Simulation based on the model is validated with robotic pushing and grasping experiments., Comment: Robotics: Science and Systems 2017

Surveillance for Streptococcus pneumoniae resistant to penicillin and other antimicrobial agents is necessary to define the optimal empirical antibiotic therapy for meningitis in resource-poor countries such as Vietnam. The clinical and microbiological features of 100 patients admitted to the Centre for Tropical Diseases in Ho Chi Minh City, Vietnam, between 1993 and 2002 with invasive pneumococcal disease were studied. A penicillin-nonsusceptible pneumococcus (MIC, ≥0.1 μg/ml) was isolated from the blood or cerebrospinal fluid of 8% of patients (2 of 24) between 1993 and 1995 but 56% (20 of 36) during 1999 to 2002 ( P < 0.0001). Pneumococcal isolates resistant to penicillin (MIC, ≥2.0 μg/ml) increased from 0% (0 of 24) to 28% (10 of 36) ( P = 0.002). Only one isolate was ceftriaxone resistant (MIC, 2.0 μg/ml). Penicillin-nonsusceptible pneumococci were isolated from 78% of children younger than 15 years (28 of 36) compared with 25% of adults (16 of 64) ( P = 0.0001). Isolation of a penicillin-nonsusceptible pneumococcus in adults with meningitis was independently associated with referral from another hospital ( P = 0.005) and previous antibiotic therapy ( P = 0.025). Multilocus sequence typing showed that 86% of the invasive penicillin-resistant pneumococcus isolates tested (12 of 14) were of the Spain 23F -1 clone. The serotypes of >95% of the penicillin-nonsusceptible pneumococci were included in the currently available pneumococcal vaccines. Our findings point to the recent introduction and spread of the Spain 23F -1 clone of penicillin-resistant pneumococci in Vietnam. Simple clinical predictors can be used to guide empirical antibiotic therapy of meningitis. Pneumococcal vaccination may help to control this problem.

We propose a polynomial force-motion model for planar sliding. The set of generalized friction loads is the 1-sublevel set of a polynomial whose gradient directions correspond to generalized velocities. Additionally, the polynomial is confined to be convex even-degree homogeneous in order to obey the maximum work inequality, symmetry, shape invariance in scale, and fast invertibility. We present a simple and statistically-efficient model identification procedure using a sum-of-squares convex relaxation. Simulation and robotic experiments validate the accuracy and efficiency of our approach. We also show practical applications of our model including stable pushing of objects and free sliding dynamic simulations., Comment: 2016 IEEE International Conference on Robotics and Automation (ICRA)

Multilocus sequence typing (MLST) of 35 isolates of multidrug-resistant Streptococcus pneumoniae recovered in Poland during 1995-1996 distinguished 10 different sequence types (ST). The majority of the isolates were assigned to two Polish clones of serotypes 6B and 23F, although the international clones, Spain23F-1 and Spain9V-3, were also identified. Similar results were obtained using pulsed-field gel electrophoresis (PFGE), providing a direct comparison of these two typing methods.

We consider detecting objects in an image by iteratively selecting from a set of arbitrarily shaped candidate regions. Our generic approach, which we term visual chunking, reasons about the locations of multiple object instances in an image while expressively describing object boundaries. We design an optimization criterion for measuring the performance of a list of such detections as a natural extension to a common per-instance metric. We present an efficient algorithm with provable performance for building a high-quality list of detections from any candidate set of region-based proposals. We also develop a simple class-specific algorithm to generate a candidate region instance in near-linear time in the number of low-level superpixels that outperforms other region generating methods. In order to make predictions on novel images at testing time without access to ground truth, we develop learning approaches to emulate these algorithms' behaviors. We demonstrate that our new approach outperforms sophisticated baselines on benchmark datasets., Comment: to appear at ICRA 2015

The identification of clones within bacterial populations is often taken as evidence for a low rate of recombination, but the validity of this inference is rarely examined. We have used statistical tests of congruence between gene trees to examine the extent and significance of recombination in six bacterial pathogens. For Neisseria meningitidis , Streptococcus pneumoniae , Streptococcus pyogenes, and Staphylococcus aureus , the congruence between the maximum likelihood trees reconstructed using seven house-keeping genes was in most cases no better than that between each tree and trees of random topology. The lack of congruence between gene trees in these four species, which include both naturally transformable and nontransformable species, is in three cases supported by high ratios of recombination to point mutation during clonal diversification (estimates of this parameter were not possible for Strep. pyogenes ). In contrast, gene trees constructed for Hemophilus influenzae and pathogenic isolates of Escherichia coli showed a higher degree of congruence, suggesting lower rates of recombination. The impact of recombination therefore varies between bacterial species but in many species is sufficient to obliterate the phylogenetic signal in gene trees.

Multilocus sequence typing was used to characterize isolates of the major Spanish clones of penicillin-resistant and multiple-antibiotic-resistant Streptococcus pneumoniae . Isolates of the multidrug-resistant Spanish serotype 23F clone and serotype variants of this clone either had identical allelic profiles or their allelic profiles differed from this typical allelic profile at only one of the seven housekeeping loci. Similarly, isolates of the Spanish serotype 6B and 14 clones and the penicillin-resistant serotype 9V clone (and serotype variants of this clone) each had the same allelic profiles or profiles that differed at a single locus. Multilocus sequence typing therefore allows resistant pneumococci to be assigned to the Spanish clones if they have the typical allelic profile of the clone or if their profiles differ from that profile at a single locus. A few resistant isolates that had allelic profiles typical of that of a Spanish clone or whose profiles differed from that of the typical profile at only a single locus possessed penicillin-binding protein pbp1a , pbp2b , or pbp2x genes that differed from those that are characteristic of the clone. In most cases these isolates could be assigned as variant members of the clone. Since almost all serotype 9V isolates have very similar genotypes, independently emerging penicillin-resistant clones of this serotype will inevitably appear to be similar by molecular typing procedures. Analysis of the pbp genes, in addition to multilocus sequence typing (or any other molecular typing procedure), is therefore required to assign isolates unambiguously to the penicillin-resistant Spanish serotype 9V clone.

The sequences of the adenylate kinase gene (adk) and the RecA gene (recA) were determined from the same isolates of Neisseria gonorrhoeae, N. meningitidis, N. lactamica, N. polysaccharea, N. cinerea, N. mucosa, N. pharyngis var. flava, N. flavescens, and N. animalis. The patterns of sequence divergence observed at adk and recA were very different. Dendrograms constructed from the recA data using two different algorithms were statistically robust and were congruent with each other and with the relationships between the species previously proposed using other data. In contrast, the dendrograms derived from the adk data were noncogruent with each other, and with those from the recA data, and were statistically poorly supported. These results, along with the uniform distribution of pairwise sequence divergences between the species at adk, suggest there has been a history of interspecies recombination within the adk gene of the human Neisseria species which has obscured the phylogenetic relationships between the species. This view was supported by Sawyer's runs test, and the Index of Association (IA) between codons, which provided significant evidence for interspecies recombination between the adk genes from the human Neisseria species, but no evidence of interspecies recombination between the recA sequences.

Geographic Profiling has successfully assisted investigations for serial crimes. Considering the multi-cluster feature of serial criminal spots, we propose a Multi-point Centrography model as a natural extension of Single-point Centrography for geographic profiling. K-means clustering is first performed on the data samples and then Single-point Centrography is adopted to derive a probability distribution on each cluster. Finally, a weighted combinations of each distribution is formed to make next-crime spot prediction. Experimental study on real cases demonstrates the effectiveness of our proposed model., {"references":["D. K. Rossmo, Geographic Profiling, CRC Press, 2000.","D. Canter, T. Coffey, M. Huntley, and C. Missen, \" Predicting serial\nkillers- home base using a decision support system,\" J. Quantitative\nCriminology, vol. 16, pp. 457-478, 2000.","M. O-Leary, \"The mathematics of geographic profiling,\" J. Investig.\nPsych. Offender Profil, vol. 6, pp. 253-265, 2009.","V. Latora, and M. Marchiori, \"Efficient behavior of small-world networks,\"\nPhys. Rev. Lett., vol. 87. pp. 198701, 2001.","A. Arenas, A. Diaz-Guilera, J. Kurths, Y. Moreno, and C. Zhou, \"Synchronization\nin complex networks,\" Phys. Rep., vol. 469, pp. 93-153,\n2008.","M. Rosvall, and A. Trusina, and P. Minnhagen, and K. Sneppen, \"Networks\nand cities: An information perspective,\" Phys, Rev. Lett., vol. 94, pp. 28701, 2005.","C. Qian, Y. Wang, J. Cao, J. Lu, and J. Kurths, \"Weighted-trafficnetwork\nbased geographic profiling for serial crime location prediction,\" Europhysics Letters, vol. 93, pp. 68006, 2011.","M. O-Leary, \"Determining the optimal search area for a serial criminal,\"\nJoint Mathematics Meetings, Washington DC, USA, 2009.","S. S. Kind, \"Navigational ideas and the Yorkshire Ripper investigation,\"\nJournal of Navigation, vol. 40, pp. 385-393, 1987.\n[10] http://en.wikipedia.org/wiki/Peter Sutcliffe.\n[11] D. J. Icove, H. J. Crisman, \"Application of pattern recognition in arson\ninvestigation,\"Fire Techonology, 1975.\n[12] Bishop, C. M., Pattern recognition and machine learning, Springer New\nYork, 2006.\n[13] http://en.wikipedia.org/wiki/Jack the Ripper.\n[14] http://en.wikipedia.org/wiki/Zodiac\nKiller.\n[15] http://en.wikipedia.org/wiki/Milwaukee_North_Side_Strangler.\n[16] Duda, R.O. and Hart, P.E. and Stork, D.G., Pattern classification, wiley New York, 2001."]}

Background: Classifying bacteria into species is problematic. Most microbiologists consider species to be groups of isolates that share some arbitrary degree of relatedness of biochemical or molecular (such as DNA sequence) features and that, ideally, are clearly delineated from all other groups of isolates. The main problem in applying to bacteria a biological concept of species based on the ability or inability of their genes to recombine, is that recombination appears to be rare in bacteria in nature, as indicated by the strong linkage disequilibrium between alleles found in most bacterial populations. However, there are some naturally transformable bacteria in which assortative recombination appears to be so frequent that alleles are in, or close to, linkage equilibrium. For these recombining populations a biological concept of species might be applicable. Results: Populations of Neisseria gonorrhoeae and Neisseria meningitidis from Spain were analysed by multilocus enzyme electrophoresis. The data indicate that assortative recombination occurs frequently within populations, but not between populations. Similarly, the sequences of two house-keeping genes show no evidence of intragenic recombination between N. gonorrhoeae and N. meningitidis . Conclusions: N. gonorrhoeae and N. meningitidis represent extremely closely related ‘sexual' populations that appear to be genetically isolated in nature, and thus conform to the biological concept of species. The extreme uniformity of N. gonorrhoeae house-keeping genes suggests that this species may have arisen recently as a clone of N. meningitidis that could colonize the genital tract. Ecological isolation — of populations that can colonize the genital tract from those that can colonize the nasopharynx — may have been an important component in speciation, leading to a lower frequency of recombination between species than within species.

Attentiveness is one of the key factors in human intelligent behavior. Especially, we are interested in the attentiveness states of learners. In recent years, lots of methods were proposed for attentiveness assessment, including computer vision, speech recognition, physiology and other approaches, and some of them already shown exciting results. We believe that physiological approach is very suitable to detect learners' attentiveness. However, till now the performance of these methods were measured on the single testee in their experiments, which means their conclusions may not be generally valid. Although it is reasonable to restrict test subjects in early stage of research, generalized experiments involving multiple subjects are much more important to study. In this paper, we conducted a series of experiments that collected physiological data from 20 different subjects. Based on the experimental data, we revealed the huge individual differences of physiological features among those subjects. In order to smooth down such differences, we adopted continuous restricted Boltzmann machine to extract features from the original data. Finally we compared the method we used with other algorithms. The experimental result shows positive support towards generally applicable attentiveness detection by physiology approach.

Standard Natural Classroom (SNC) is a real-time classroom based on smart space and e-Learning technologies, which aims at creating face-to-face, interactive and pervasive learning experience for both local and remote students. In this paper, the phrase “interaction” refers to human-computer interaction, teacher-student interaction and student-student interaction in hybrid learning process. The characteristics of these interactions are multimodal and context aware. In this paper, we first draw an outline of SNC and Virtual Interaction Classroom (VIC). Second, we introduce a so called “Context Aware Multimodal Interaction” (CAMI) concept, and the multimodal and context information in SNC. Finally, we describe the implementation of a real-time Context Aware Multimodal Interaction Model (CAMIM) which is universal, extendable and efficient. We will illustrate the architecture of CAMIM, then we will demonstrate the Interaction Task Set (ITS) and CAMI fusion strategy. In particular, automated planning technology Hierarchical Task Network (HTN) is introduced into this model.

Summary Multiply antibiotic-resistant serotype 23F isolates of Streptococcus pneumoniae are prevalent in Spain and have also been recovered recently in the United Kingdom and the United States. Analysis of populations of these isolates by multilocus enzyme electrophoresis, and restriction endonuclease cleavage electrophoretic profiling of penicillin-binding protein (PBP) genes, has demonstrated that these isolates are a single clone (Munoz et al., 1991). Here we report studies of non-serotype 23F penicillin-resistant pneumococci isolated in Spain and the United Kingdom. One of the isolates expressed serotype 19 capsule but was otherwise indistinguishable from the serotype 23F clone on the basis of multilocus enzyme electrophoresis, antibiotic resistance profiling, and restriction endonuclease patterns of genes encoding PBP1A, PBP2B and PBP2X, a result which suggests that horizontal transfer of capsular biosynthesis genes had occurred. These same techniques revealed that six other resistant isolates, all expressing serotype 9 polysaccharide capsule, represent a clone. Interestingly, the chromosomal lineage of this clone is not closely related to the 23F clone; however, the serotype 9 and 23F clones harbour apparently identical PBP1 A, -2B and -2X genes. To explain these data, we favour the interpretation that horizontal gene transfer in natural populations has distributed genes encoding altered forms of PBP1A, -2B and -2X to distinct evolutionary lineages of S. pneumoniae.

Traditional and molecular typing schemes for the characterization of pathogenic microorganisms are poorly portable because they index variation that is difficult to compare among laboratories. To overcome these problems, we propose multilocus sequence typing (MLST), which exploits the unambiguous nature and electronic portability of nucleotide sequence data for the characterization of microorganisms. To evaluate MLST, we determined the sequences of ≈470-bp fragments from 11 housekeeping genes in a reference set of 107 isolates of Neisseria meningitidis from invasive disease and healthy carriers. For each locus, alleles were assigned arbitrary numbers and dendrograms were constructed from the pairwise differences in multilocus allelic profiles by cluster analysis. The strain associations obtained were consistent with clonal groupings previously determined by multilocus enzyme electrophoresis. A subset of six gene fragments was chosen that retained the resolution and congruence achieved by using all 11 loci. Most isolates from hyper-virulent lineages of serogroups A, B, and C meningococci were identical for all loci or differed from the majority type at only a single locus. MLST using six loci therefore reliably identified the major meningococcal lineages associated with invasive disease. MLST can be applied to almost all bacterial species and other haploid organisms, including those that are difficult to cultivate. The overwhelming advantage of MLST over other molecular typing methods is that sequence data are truly portable between laboratories, permitting one expanding global database per species to be placed on a World-Wide Web site, thus enabling exchange of molecular typing data for global epidemiology via the Internet.

Visual inspection showed clear evidence of a history of intraspecies recombinational exchanges within the neighbouring meningococcal shikimate dehydrogenase (aroE) and glutamine synthetase (glnA) genes, which was supported by the non-congruence of the trees constructed from the sequences of these genes from different meningococcal strains, and by statistical tests for mosaic structure. Many examples were also found of highly localized interspecies recombinational exchanges between the meningococcal aroE and glnA genes and those of commensal Neisseria species. These exchanges appear to have inflated the sequence variation at these loci, and have resulted in major distortions of the phylogenetic trees constructed from the sequences of the aroE and glnA genes of human pathogenic and commensal Neisseria species. Statistical tests for sequence mosaicism, and for anomalies within the Neisseria species trees, strongly supported the view that frequent interspecies recombination has occurred within aroE and glnA. The high levels of sequence variation, and intra- and interspecies recombination, within aroE and glnA did not appear to be due to a 'hitch-hiking' effect caused by positive selection for variation at a neighbouring gene. Our results suggest that interspecies recombinational exchanges with commensal Neisseria occur frequently in some meningococcal 'housekeeping' genes as they can be observed readily even when there appears to be no obvious selection for the recombinant phenotypes.

Studies of natural populations of Neisseria meningitidis using multilocus enzyme electrophoresis have shown extensive genetic variation within this species, which, it has been proposed, implies a level of sequence diversity within meningococci that is greater than that normally considered as the criterion for species limits in bacteria. To obtain a direct measure of the sequence diversity among meningococci, we obtained the nucleotide sequences of most of the argF, recA and fbp genes of eight meningococci of widely differing electrophoretic type (from the reference collection of Caugant). Sequence variation between the meningococcal strains ranged from 0-0.6% for fbp, 0-1.3% for argF, and 0-3.3% for recA. These levels of diversity are no greater than those found within Escherichia coli 'housekeeping' genes and suggest that multilocus enzyme electrophoresis may overestimate the extent of nucleotide sequence diversity within meningococci. The average sequence divergence between the Neisseria meningitidis strains and N. gonorrhoeae strain FA19 was 1.0% for fbp and 1.6% for recA. The argF gene, although very uniform among the eight meningococcal isolates, had a striking mosaic structure when compared with the gonococcal argF gene: two regions of the gene differed by greater than 13% in nucleotide sequence between meningococci and gonococci, whereas the rest of the gene differed by less than 1.7%. One of the diverged regions was shown to have been introduced from the argF gene of a commensal Neisseria species that is closely related to Neisseria cinerea. The source of the other region was unclear.

The two pathogenic species of Neisseria, N. meningitidis and N. gonorrhoeae, have evolved resistance to penicillin by alterations in chromosomal genes encoding the high molecular weight penicillin-binding proteins, or PBPs. The PBP 2 gene (penA) has been sequenced from over 20 Neisseria isolates, including susceptible and resistant strains of the two pathogenic species, and five human commensal species. The genes from penicillin-susceptible strains of N. meningitidis and N. gonorrhoeae are very uniform, whereas those from penicillin-resistant strains consist of a mosaic of regions resembling those in susceptible strains of the same species, interspersed with regions resembling those in one, or in some cases, two of the commensal species. The mosaic structure is interpreted as having arisen from the horizontal transfer, by genetic transformation, of blocks of DNA, usually of a few hundred base pairs. The commensal species identified as donors in these interspecies recombinational events (N. flavescens and N. cinerea) are intrinsically more resistant to penicillin than typical isolates of the pathogenic species. Transformation has apparently provided N. meningitidis and N. gonorrhoeae with a mechanism by which they can obtain increased resistance to penicillin by replacing their penA genes (or the relevant parts of them) with the penA genes of related species that fortuitously produce forms of PBP 2 that are less susceptible to inhibition by the antibiotic. The ends of the diverged blocks of DNA in the penA genes of different penicillin-resistant strains are located at the same position more often than would be the case if they represent independent crossovers at random points along the gene. Some of these common crossover points may represent common ancestry, but reasons are given for thinking that some may represent independent events occurring at recombinational hotspots.