Mechanical Engineering

For a mobile robot that interacts with humans such as a home assistant or a tour guide robot, activities such as locating objects, following specific people, and distinguishing among different people are fundamental, yet challenging robotic vision tasks. For complex object recognition and tracking tasks such as person recognition and tracking, we use the method of inter-classifier feedback to track both uniquely identifying characteristics of a person (e.g. face), and more frequently visible, but perhaps less uniquely identifying characteristics (e.g. shirt). The inter-classifier feedback enables merging multiple, heterogeneous sub-classifiers designed to track and associate different characteristics of a person being tracked. These heterogeneous sub-classifiers give feedback to each other by identifying additional online training data for one another, thus improving the performance of the overall tracking system. We implement the tracking system on a Segway base that successfully performed aforementioned activities to a second place finish in the RoboCup@Home 2007 competition. The main contribution is a complete description and analysis of the robot system and its implemented algorithms.

In Fall 2007, the US Defense Advanced Research Projects Agency (DARPA) held the Urban Challenge, a street race between fully autonomous vehicles. Unlike previous challenges, the Urban Challenge vehicles had to follow the California laws for driving, including properly handling traffic. This article presents the modular algorithms developed largely by undergraduates at The University of Texas at Austin as part of the Austin Robot Technology team. We emphasize the aspects of the system that are relevant to multiagent interactions. Specifically, we discuss how our vehicle tracked and reacted to nearby traffic in order to allow our autonomous vehicle to safely follow and pass, merge into moving traffic, obey intersection precedence, and park.

In this paper a new on-Mars calibration procedure of a planetary lander vision system is presented. It is based on recent developments in computer vision. The calibration is retrieved from the images of the Mars terrain (the same images from which the 3D measurements of the terrain are obtained). The procedure is based on the relations between multiple views of the same scene. The main advantage of this calibration procedure is that it allows to retrieve the calibration on mars without any additional requirements on the system. Once the system has been calibrated, the same images can be used to estimate a digital elevation map of the environment around the lander. This means that the terrain based calibration causes no overhead on transmission. These images are first processed pairwise using a stereo algorithm yielding sub-pixel disparity maps. In order to perform the path planning a digital elevation map is required. A new algorithm was developed to efficiently extract this from the disparity maps. This algorithm can generate a regular digital elevation map at any desired resolution (interpolating if necessary) and easily takes occlusions into account.

The problem addressed is the distributed reconfiguration of a metamorphic robotic system composed of any number of identical hexagonal modules. A metamorphic robotic system is a collection of independently controlled, smaller robots that can move around adjacent robots to change the shape of the overall system. The shape-changing ability of these systems makes them more versatile than conventional mobile robots. Because the modules are homogeneous, the systems are also capable of self-repair.

It is possible to broadly characterize two approaches to probabilistic modeling in terms of generative and discriminative methods. Provided with sufficient training data the discriminative approach is expected to yield superior accuracy as compared to the analogous generative model since no modeling power is expended on the marginal distribution of the features. Conversely, if the model is accurate the generative approach can perform better with less data. In general it is less vulnerable to overfitting and allows one to more easily specify meaningful priors on the model parameters. We investigate multi-conditional learning -a method combining the merits of both approaches. Through specifying a joint distribution over classes and features we derive a family of models with analogous parameters. Parameter estimates are found by optimizing an objective function consisting of a weighted combination of conditional loglikelihoods. Systematic experiments in the context of foreground/background pixel classification with the Microsoft-Berkeley segmentation database using mixtures of factor analyzers illustrate tradeoffs between classifier complexity, the amount of training data and generalization accuracy. We show experimentally that this approach can lead to models with better generalization performance than purely generative or discriminative approaches.

Automated recognition of features from CAD models has been attempted for a wide range of application domains. In this paper we address the problem of representing and recognizing a complete class of features in alternative i n terpretations for a given design.

Smooth integration of computer-aided design (CAD) and manufacturing (CAM) is vital for companies' survival in the present competitive marketplace. One key component in this effort is the representation of product designs in terms of features, elements that are semantically higher-level than the pure geometric elements typically used in CAD and CAM systems.

We propose a novel approach to the problem of simultaneous localization and mapping (SLAM) based on incremental smoothing, that is suitable for real-time applications in large-scale environments. The main advantages over filter-based algorithms are that we solve the full SLAM problem without the need for any approximations, and that we do not suffer from linearization errors. We achieve efficiency by updating the square-root information matrix, a factored version of the naturally sparse smoothing information matrix. We can efficiently recover the exact trajectory and map at any given time by back-substitution. Furthermore, our approach allows access to the exact covariances, as it does not suffer from under-estimation of uncertainties, which is another problem inherent to filters. We present simulation-based results for the linear case, showing constant time updates for exploration tasks. We further evaluate the behavior in the presence of loops, and discuss how our approach extends to the non-linear case. Finally, we evaluate the overall non-linear algorithm on the standard Victoria Park data set.

We describe a reinforcement learning system that transfers skills from a previously learned source task to a related target task. The system uses inductive logic programming to analyze experience in the source task, and transfers rules for when to take actions. The target task learner accepts these rules through an advice-taking algorithm, which allows learners to benefit from outside guidance that may be imperfect. Our system accepts a human-provided mapping, which specifies the similarities between the source and target tasks and may also include advice about the differences between them. Using three tasks in the RoboCup simulated soccer domain, we demonstrate that this system can speed up reinforcement learning substantially.

We introduce the Marco Polo Localization approach, where we apply sound as a tool for gathering range measurements between robots, and use those to solve a range-only Simultaneous Localization and Mapping problem. Range is calculated by correlating two recordings of the same sound, recorded on a pair of robots, after which the resulting time delay estimate is converted to a range measurement. The algorithmic approach we use is a straightforward application of the Bayesian estimation framework. We also present two complementary views on the associated optimization problem that provide insight into the problem and allows one to devise initialization strategies, indispensable in a range-only scenario. We illustrate the approach with both simulated and experimental results. ©

This paper studies the effect of post-auction market interaction and asymmetric information on the design of a revenue maximizing mechanism. In this situation, bidders can have incentives to signal their types to influence the outcome of the postauction market game. Thus a revenue-maximizing mechanism not only specifies who wins the object but also describes the amount of information that is revealed about the bidders. The form of the bidders' utility functions determines the format of a revenue maximizing mechanism. This paper shows that if bidders have additively separable utility functions in their true and signaled types, then revealing information is revenue maximizing if the utility function is convex in the signaled type, while hiding information is revenue maximizing if the utility function is concave. Previous literature suggested that the key factor is the sign of the first derivative of the utility function with respect to the perceived type, which is contradicted by our findings. * We thank Abstract: This paper studies the effect of post-auction market interaction and asymmetric information on the design of a revenue maximizing mechanism. In this situation, bidders can have incentives to signal their types to influence the outcome of the postauction market game. Thus a revenue-maximizing mechanism not only specifies who wins the object but also describes the amount of information that is revealed about the bidders. The form of the bidders' utility functions determines the format of a revenue maximizing mechanism. This paper shows that if bidders have additively separable utility functions in their true and signaled types, then revealing information is revenue maximizing if the utility function is convex in the signaled type, while hiding information is revenue maximizing if the utility function is concave. Previous literature suggested that the key factor is the sign of the first derivative of the utility function with respect to the perceived type, which is thus contradicted by our findings.

Experimentalists now measure intense rotations of Lagrangian particles in turbulent flows by tracking their trajectories and Lagrangian-average velocity gradients at high Reynolds numbers. This paper formulates the dynamics of an orthonormal frame attached to each Lagrangian fluid particle undergoing three-axis rotations, by using quaternions in combination with Ertel's theorem for frozen-in vorticity. The method is applicable to a wide range of Lagrangian flows including the three-dimensional Euler equations and its variants such as ideal magneto-hydrodynamics. The applicability of the quaterionic frame description to Lagrangian averaged velocity gradient dynamics is also demonstrated.

Numerous mesh algorithms such as parametrization, radiosity, and collision detection require the decomposition of meshes into a series of clusters. In this paper we present two novel approaches for maintaining mesh clusterings on dynamically deforming meshes. The first approach maintains a complete face cluster tree hierarchy using a randomized data structure. The second algorithm maintains a mesh decomposition for a fixed set of clusters. With both algorithms we are able to maintain clusterings on dynamically deforming surfaces of over 100K faces in fractions of a second.

Pearling is an interactive technique for extracting the centerline and varying thickness of (hand-drawn) curves in images. It works in realtime, is capable of tracing complex curve networks, and supports gesture-driven interactive editing to trim or extend the selection.

: The deformation of a source image (a), described by tracing and moving feature curves (b, top), can unrealistically stretch texture details, e.g. the ear fur (b, bottom). We preserve texture detail by growing curvilinear coordinates around both the source and target feature curves (c), and resynthesizing the texture on the target coordinates from the original image at the source coordinates. Texture synthesis occurs on small patches (d), joined by GraphCut and Poisson Image Editing to yield the final result (e), preserving texture frequency and orientation.

This paper presents a kinematic model of the human arm in which the workspace of the elbow is modeled as a triangular B ezier spline surface. It is also explained how this model is used for solving forward and inverse kinematics in computer animation systems. In order to solve inverse kinematics problems, it is necessary to nd a curve on the surface modeling the elbow workspace. This curve is obtained as the intersection of the surface and a sphere. We present an algorithm for computing this curve, using the fact that triangular B ezier patch can be e ciently subdivided, and using an oct-tree data structure that prunes the search space e ciently.

Many standard functions, such as the logarithm and square root functions, cannot be de ned continuously on the complex plane. Mistaken assumptions about the properties of these functions lead computer algebra systems into various conundrums. We discuss how they can manipulate such functions in a useful fashion.

We introduce a new method for stereo visual SLAM (simultaneous localization and mapping) that works in unstructured, outdoor environments. Unlike other gridbased SLAM algorithms, which use occupancy grid maps, our algorithm uses a new mapping technique that maintains a posterior distribution over the height variance in each cell. This idea was motivated by our experience with outdoor navigation tasks, which has shown height variance to be a useful measure of traversability. To obtain a joint posterior over poses and maps, we use a Rao-Blackwellized particle filter: the pose distribution is estimated using a particle filter, and each particle has its own map that is obtained through exact filtering conditioned on the particle's pose. Visual odometry provides good proposal distributions for the particle pose. In the analytical (exact) filter for the map, we update the sufficient statistics of a gamma distribution over the precision (inverse variance) of heights in each grid cell. We verify the algorithm's accuracy on two outdoor courses by comparing with ground truth data obtained using electronic surveying equipment. In addition, we solve for the optimal transformation from the SLAM map to georeferenced coordinates, based on a noisy GPS signal. We derive an online version of this alignment process, which can be used to maintain a running estimate of the robot's global position that is much more accurate than the GPS readings.

DRAFT low (from a few dollars to a few cents depending on the application), sensors will need to last for years or even decades without battery replacement, and the network will need to organize without significant human moderation. Traditional localization techniques aren't well suited for these requirements. Including GPS on each device is cost and energy prohibitive for many applications, not sufficiently robust to jamming for military applications, and limited to outdoor applications. Local positioning systems (LPS)

We present a new tactile display for rendering rotational friction at the fingertips. To inform the design and control of this device, we have conducted an initial set of perceptual experiments to investigate perceptual thresholds for rates of rotational slipping at the fingertips. It was found that humans can discriminate a change in rotational sliding rate of 4.13-28.04 rad/sec from a reference rate of 1.57-25.13 rad/sec. An initial device design and prototype are presented. The device utilizes a small friction disk housed in a thimble, driven remotely from a motor worn on the user's wrist. The device directly integrates with a robotic force feedback arm, preserving the force feedback functionality while adding the tactile information on frictional sliding.

This paper investigates the problem of perceived instability during haptic texture rendering. We focus on the perceptual analysis of the stability of textured surfaces rendered with a force-reflecting device. The method of limits is used to assess the detection thresholds for perceived instability in terms of the stiffness of the virtual textured surfaces. We varied texture rendering method, exploration mode, and the amplitude and spatial wavelength of a sinusoidal surface texture model. Our results show that the maximum stiffness value for perceptually stable texture rendering is quite small (< 0.45 N/mm) for the range of textured surfaces tested. We also found that the stiffness thresholds depend on many factors such as rendering method, exploration mode, inter-subject difference, and surface model. Our current and future work focuses on the sources of perceived instability during haptic texture rendering. With these studies, we hope to develop strategies that can mitigate the problem of perceived instability during texture and other types of haptic rendering.

One challenge in multimodal interface research is the lack of robust subsystems that support multimodal interactions. By focusing on a chair-an object that is involved in virtually all human-computer interactions, the sensing chair project enables an ordinary office chair to become aware of its occupant's actions and needs. Surface-mounted pressure distribution sensors are placed over the seatpan and backrest of the chair for real time capturing of contact information between the chair and its occupant. Given the similarity between a pressure distribution map and a grayscale image, pattern recognition techniques commonly used in computer and robot vision, such as principal components analysis, have been successfully applied to solving the problem of sitting posture classification. The current static posture classification system operates in real time with an overall classification accuracy of 96% and 79% for familiar (people it had felt before) and unfamiliar users, respectively. Future work is aimed at a dynamic posture tracking system that continuously tracks not only steady-state (static) but transitional (dynamic) sitting postures. Results reported here form important stepping stones toward an intelligent chair that can find applications in many areas including multimodal interfaces, intelligent environment, and safety of automobile operations.

Motivated by some concrete examples that have been studied in other contexts (Example 3), we generalize the concept of subnormality for operators on Hilbert space to include the possibility that the space on which the normal extension is defined is a Krein space instead of a Hilbert space. In this paper, we develop some of the basic properties of this new class, the Krein subnormal operators. In recently published work [?], Wu Jingbo studies a different generalization of subnormality, called J-subnormality, and proves the striking theorem that every bounded operator on Hilbert space is J-subnormal. Our definition is much more restrictive; Krein subnormal operators have many properties analogous to subnormal operators.

Motivated by the viscous sintering of amorphous materials in the presence of a distribution of rigid platelet inclusions added to increase the strength of the final sintered medium, a simple model for the shrinkage of compressible pores close to a wall in Stokes flow is proposed. The model assumes that the pore remains elliptical/ellipsoidal in shape at all times. The model is expected to be valid provided the pore is not too far from spherical and not too close to the wall. It relies on a separation of the flow into 'inner' and 'outer' problems. The inner problem is to assume that the ellipsoid evolves in a linear ambient flow given by the first two terms in a local expansion of an outer flow produced by the image singularity distribution of a point-sink near a planar no-slip boundary. The focus of the present paper is to test the viability of the model in the planar case. Using a spectral method based on analytic functions and conformal mappings, the results of a full numerical simulation are compared with the predictions of the planar model. The effects of the proximity of the pore to the wall, the anisotropy in the pore shape and its relative orientation to the wall are all examined. It is observed that, as they shrink, pores drift towards the wall becoming elongated in the direction perpendicular to the wall.

This paper presents a combined numerical and analytical investigation into various problems involving two symmetric interacting constant-pressure bubbles evolving in two-dimensional Stokes flow. The bubbles have constant surface tension on their boundaries and are taken to be in an ambient straining flow. First, a novel numerical method based on conformal mappings is presented to compute the free-surface evolution. Then, a special class of time-evolving exact solutions to the problem is derived and used to check the numerical code. These solutions reveal that, for bubbles with shrinking area, a competition between the imposed strain and surface tension can lead to either a slit or a point as the limiting shape. Numerical solutions of fixed-area bubbles are then computed and reveal that when they are forced together by a straining flow, a thin lubrication layer forms. In the absence of surface tension, large-curvature regions develop at the bubble edges and these are smoothed out by capillary effects. Further, motivated by the viscous sintering application, a study of interaction effects on the pure surface-tension-driven shrinkage of circular bubbles is investigated and compared, in an appropriate limit, to a recently derived 'ellipticalpore model'.

The classic exact solution due to Lagally Die reibungslose strömung im aussengebiet zweier kreise. Z. Angew. Math. Mech. 9, 299-305.) for streaming flow past two cylindrical aerofoils (or obstacles) is generalized to the case of an arbitrary finite number of cylindrical aerofoils. Given the geometry of the aerofoils, the speed and direction of the oncoming uniform flow and the individual round-aerofoil circulations, the complex potential associated with the flow is found in analytical form in a parametric pre-image region that can be conformally mapped to the fluid region. A complete determination of the flow then follows from knowledge of the conformal mapping between the two regions. In the special case where the aerofoils are all circular, the conformal mapping from the parametric pre-image region to the fluid domain is a Möbius mapping. The solution for the complex potential in such a case can then be used, in combination with the Blasius theorem, to compute the distribution of hydrodynamic forces on the multi-aerofoil configuration.

A numerical algorithm is presented for the computation of the Schottky-Klein prime function on the Schottky double of multiply connected circular domains in the plane. While there exist classical formulae for the Schottky-Klein prime function in the form of infinite products over a Schottky group, such products are not convergent for all choices of multiply connected circular domains. The prime function itself, however, is a well-defined function for any multiply connected circular domain. The present algorithm facilitates the evaluation of this prime function when the planar domains are such that the classical infinite product representation is either not convergent or so slowly convergent as to be impracticable.

The use of visual displays is seen as a supportive tool for solving design problems.Throughout the design process, and particularly in the early stages of the process, designers are exposed to vast collections of visual displays. These are external representations, i.e., pictures, diagrams, or sketches, which provide designers with helpful explicit and non explicit references. Despite the importance of this pictorial
material, only a small number of researches have dealt with the spontaneous use of visual displays as an aid in design problem solving when no instruction to use analogy is given. Furthermore, no studies have been conducted to examine whether visual displays play a more significant role in solving ill-defined or well-defined design problems. The main goal of this work is to empirically research the use of visual displays in these two problem contexts, by studying the design process of groups of designers with different levels of expertise. Findings showed that both experts and
novices profited from the use of visual displays in ill-defined design problem solving, resulting in a significant enhacement of the quality of design solutions. Additional results showed that visual displays did aid experts, but not novices, to improve their performance in solving well-defined design problems. These findings may have Design Aided By Visual Displays: A Cognitive Approach consequences for design education. It is suggested that practicing with a large number
of within-domain visual displays (which belong to the same or very close realm of the problem) and a large collection of between-domain visual sources (which belong to a different or remote realm of the problem) can help designers in general, and novices in particular, to spontaneously retrieve meaningful information, and to enhance their design abilities.

This paper presents a self-improving reactive control system for autonomous robotic navigation. The navigation module uses a schemabased reactive control system to perform the navigation task. The learning module combines case-based reasoning and reinforcement learning to continuously tune the navigation system through experience. The case-based reasoning component perceives and characterizes the system's environment, retrieves an appropriate case, and uses the recommendations of the case to tune the parameters of the reactive control system. The reinforcement learning component refines the content of the cases based on the current experience. Together, the learning components perform on-line adaptation, resulting in improved performance as the reactive control system tunes itself to the environment, as well as on-line learning, resulting in an improved library of cases that capture environmental regularities necessary to perform on-line adaptation. The system is extensively evaluated through simulation studies using several performance metrics and system configurations.

We establish the existence and uniqueness of finite free resolutions -and their attendant Betti numbers -for graded commuting d-tuples of Hilbert space operators. Our approach is based on the notion of free cover of a (perhaps noncommutative) row contraction. Free covers provide a flexible replacement for minimal dilations that is better suited for higher-dimensional operator theory.

The eigenvalues of a self-adjoint n × n matrix A can be put into a decreasing sequence λ = (λ1, . . . , λn), with repetitions according to multiplicity, and the diagonal of A is a point of R n that bears some relation to λ. The Schur-Horn theorem characterizes that relation in terms of a system of linear inequalities.

Let X = {λ1, . . . , λN } be a finite set of complex numbers and let A be a normal operator with spectrum X that acts on a separable Hilbert space H. Relative to a fixed orthonormal basis e1, e2, . . . for H, A gives rise to a matrix whose diagonal is a sequence d = (d1, d2, . . . ) with the property that each of its terms dn belongs to the convex hull of X. Not all sequences with that property can arise as the diagonal of a normal operator with spectrum X.

A challenge of design education is the question of how to help designers develop skills in design problem-solving. How can designers be taught to use relevant prior knowledge to solve new design problems? To answer this question we must know more about differences between experts and novices regarding the use of prior knowledge to solve ill defined problems. In design, visual analogy is used as a powerful problem-solving strategy; the evidence, however, is hitherto mostly anecdotal. In this study our objective is to determine empirically whether, and how, the use of visual analogy can improve design problem-solving by both novice and expert designers. Our results indicate that the use of visual analogy improves the quality of design across the board, but is particularly significant in the case of novice designers. These findings lead to conclusions regarding design training and education.

Live bacterial vaccines have great promise both as vaccines against enteric pathogens and as heterologous antigen vectors against diverse diseases. Ideally, room temperature stable dry formulations of live bacterial vaccines will allow oral vaccination without cold-chain storage or injections. Attenuated Salmonella can cross the intestinal wall and deliver replicating antigen plus innate immune activation signals directly to the intestinal immune tissues; however, the ingested bacteria must survive firstly gastric acid and secondly the antimicrobial defences of the small intestine. We found that the way in which cells are grown prior to formulation markedly affects sensitivity to acid and bile. Using a previously published stable storage formulation that maintained over 10% viability after 56 days storage at room temperature, we found dried samples of an attenuated Salmonella serovar Typhimurium vaccine lost acid and bile resistance compared to the same bacteria taken from fresh culture. The stable formulation utilised osmotic pre-conditioning in defined medium plus elevated salt concentration to induce intracellular trehalose accumulation before drying. Dried bacteria grown in rich media without osmotic pre-conditioning showed more resistance to bile, but less stability during storage, suggesting a trade-off between bile resistance and stability. Further optimisation is needed to produce the ultimate room temperature stable oral live bacterial vaccine formulation.

The home advantage is a widely acknowledged sporting phenomenon, especially in association football. Here, we examine the second leg home advantage, an effect that is discussed in the public domain but which has received very little scientific attention. The second leg home advantage effect occurs when on average teams are more likely to win a two-stage knock-out competition when they play at home in the second leg. That is, both teams have a home advantage but this advantage is significantly greater for the team that plays at home second. Examining data from three different European Cup football competitions spanning 51 years, we show that the second leg home advantage is a real phenomenon. The second leg home team has more than a 50% probability to qualify for the next round in the competition even after controlling for extra time
and team ability as possible alternative explanations. The second leg home advantage appears, however, to have decreased significantly over the past decade. Possible reasons for its existence and subsequent decline are presented.

Some probabilistic aspects of the number variance statistic are investigated. Infinite systems of independent Brownian motions and symmetric α-stable processes are used to construct new examples of processes which exhibit both divergent and saturating number variance behaviour. We derive a general expression for the number variance for the spatial particle configurations arising from these systems and this enables us to deduce various limiting distribution results for the fluctuations of the associated counting functions. In particular, knowledge of the number variance allows us to introduce and characterize a novel family of centered, long memory Gaussian processes. We obtain fractional Brownian motion as a weak limit of these constructed processes.

Numerical simulations reveal the formation of singular structures in the polymer stress field of a viscoelastic fluid modeled by the Oldroyd-B equations driven by a simple body force. These singularities emerge exponentially in time at hyperbolic stagnation points in the flow and their algebraic structure depends critically on the Weissenberg number. Beyond a first critical Weissenberg number the stress field approaches a cusp singularity, and beyond a second critical Weissenberg number the stress becomes unbounded exponentially in time. A local approximation to the solution at the hyperbolic point is derived from a simple ansatz, and there is excellent agreement between the local solution and the simulations. Although the stress field becomes unbounded for a sufficiently large Weissenberg number, the resultant forces of stress grow subexponentially. Enforcing finite polymer chain lengths via a FENE-P penalization appears to keep the stress bounded, but a cusp singularity is still approached exponentially in time.

We study the behavior of an elastic loop embedded in a flowing soap film. This deformable loop is wetted into the film and is held fixed at a single point against the oncoming flow. We interpret this system as a two-dimensional flexible body interacting in a two-dimensional flow. This coupled fluid-structure system shows bistability, with both stationary and oscillatory states. In its stationary state, the loop remains essentially motionless and its wake is a von Kármán vortex street. In its oscillatory state, the loop sheds two vortex dipoles, or more complicated vortical structures, within each oscillation period. We find that the oscillation frequency of the loop is linearly proportional to the flow velocity, and that the measured Strouhal numbers can be separated based on wake structure.

The dynamics of slender filaments or fibers suspended in Stokesian fluids are fundamental to understanding many flows arising in physics, biology and engineering. Such filaments can have aspect ratios of length to radius ranging from a few tens to several thousands. Full discretizations of such 3D flows are very costly. Instead, we employ a non-local slender body theory that yields an integral equation, along the filament centerline, relating the force exerted on the body to the filament velocity. This hydrodynamical description takes into account the effect of the filament on the fluid, and is extended to capture the interaction of multiple filaments as mediated by the intervening fluid. We consider filaments that are inextensible and elastic. Replacing the force in the slender body integral equation by an explicit expression that uses Euler-Bernoulli theory to model bending and tensile forces yields an integral expression for the velocity of the filament centerlines, coupled to auxiliary integro-differential equations for the filament tensions. Based on a regularized version of these slender body equations that is asymptotically equivalent to the original formulation, we construct a numerical method which uses a combination of finite differences, implicit time-stepping to avoid severe stability constraints, and special quadrature methods for nearly singular integrals. We present simulations of single flexible filaments, as well as multiple interacting filaments, evolving in a background shear flow. These simulations show shear induced buckling and relaxation of the filaments, leading to the storage and release of elastic energy. These dynamics are responsible for the development of positive first normal stress differences, commonly associated with visco-elastic fluids that are suspensions of microscopic elastic fibers.

Using slender-body hydrodynamics, we study the flow-induced deformation of a high-aspect-ratio elastic filament. For a filament of zero rest curvature rotating in a viscous linear shear flow, our model predicts a bifurcation to shape instabilities due to compression by the flow, in agreement with experimental observations. Further, nonlinear simulations of this shape instability show that in dilute solutions, flexibility of the fibers causes both increased shear thinning as well as significant nonzero first-normalstress differences. These stress differences are positive for small-to-moderate deformations, but negative for large-amplitude flexing of the fibers.

A mechanism by which smooth initial conditions evolve towards a topological reconfiguration of fluid interfaces is studied in the context of Darcy's law. In the case of thin fluid layers, nonlinear PDEs for the local thickness are derived from an asymptotic limit of the vortex sheet representation. A particular example considered is the Rayleigh-Taylor instability of stratified fluid layers, where the instability of the system is controlled by a Bond number B. It is proved that, for a range of B and initial data ''subharmonic'' to it, interface pinching must occur in at least infinite time. Numerical simulations suggest that ''pinching'' singularities occur generically when the system is unstable, and in particular immediately above a bifurcation point to instability. Near this bifurcation point an approximate analytical method describing the approach to a finite-time singularity is developed. The method exploits the separation of time scales that exists close to the first instability in a system of finite extent, with a discrete spectrum of modes. In this limit, slowly growing long-wavelength modes entrain faster short-wavelength modes, and thereby, allow the derivation of a nonlinear evolution equation for the amplitudes of the slow modes. The initial-value problem is solved in this slaved dynamics, yielding the time and analytical structure of a singularity that is associated with the motion of zeros in the complex plane, suggesting a general mechanism of singularity formation in this system. The discussion emphasizes the significance of several variational principles, and comparisons are made between the numerical simulations and the approximate theory.

We present a space robotic system capable of capturing a free-flying satellite for the purposes of on-orbit satellite servicing. Currently such operations are carried out either manually or through discrete-event scripted controllers. The manual approach is costly and exposes astronauts to danger, while the scripted approach is tedious and brittle. Consequently, there is substantial interest in performing these operations autonomously, and the work presented here is a step in this direction. To our knowledge, ours is the only satellite-capturing system that relies on vision and cognition to deal with an uncooperative satellite. Our innovative system combines visual perception (object identification, recognition, and tracking) with high-level reasoning in a hybrid deliberative/reactive computational framework. The reasoning module, which encodes a model of the environment, performs deliberation to control the perception pipelineit guides the vision system, validates its performance, and suggests corrections when vision is performing poorly. Furthermore, it advises the behavioral controller to carry out its tasks. Reasoning and related elements, among them intention, context, and memory, are responsible for the robustness and reliability of the overall system. We demonstrate our prototype system controlling a robotic arm that autonomously captures a free-flying satellite in a realistic laboratory setting that faithfully mimics on-orbit conditions.

As an important feature, orientation field describes the global structure of fingerprints. It provides robust discriminatory information other than traditional widely-used minutiae points. However, there are few works explicitly incorporating this information into fingerprint matching stage, partly due to the difficulty of saving the orientation field in the feature template. In this paper, we propose a novel representation for fingerprints which includes both minutiae and model-based orientation field. Then, fingerprint matching can be done by combining the decisions of the matchers based on the global structure (orientation field) and the local cue (minutiae). We have conducted a set of experiments on large-scale databases and made thorough comparisons with the state-of-the-arts. Extensive experimental results show that combining these local and global discriminative information can largely improve the performance. The proposed system is more robust and accurate than conventional minutiae-based methods, and also better than the previous works which implicitly incorporate the orientation information. In this system, the feature template takes less than 420 bytes, and the feature extraction and matching procedures can be done in about 0.30 s. We also show that the global orientation field is beneficial to the alignment of the fingerprints which are either incomplete or poor-qualitied.

Pearling is an interactive technique for extracting the centerline and varying thickness of (hand-drawn) curves in images. It works in realtime, is capable of tracing complex curve networks, and supports gesture-driven interactive editing to trim or extend the selection.

This paper addresses two major challenges new product development teams face in making a product people want. The first challenge is to frame the design situation based on a real need of a customer. The second, less obvious, challenge is to get everyone on the team in agreement about what that framing is - everyone needs to be on the same page about what it is they're doing. Yet these two challenges are not independent, they are intertwined with each other, connected by the concrete research and sharing activities the teams perform. We introduce a framework to help understand the path of a design team along these two dimensions as well as illustrations of the three most common paths observed among graduate multidisciplinary new product development teams as supported by interviews and survey data. These case studies form the basis of four themes and twelve design principles to help teams navigate the new product development process.