Research Showcase

Sensor Fusion for Context Understanding

Huadong Wu et al. — Mon, 13 May 2013 13:40:24 PDT

To answer the challenge of context-understanding for HCI, we propose and test experimentally a top-down sensor fusion approach. We seek to systematize the sensing process in two steps: first, decompose relevant context information in such a way that it can be described in a model of discrete facts and quantitative measurements; second, we build a generalizable sensor fusion architecture to deal with highly distributed sensors in a dynamic configuration to collect, fuse and populate our context information model. This paper describes our information model, system architecture, and preliminary experimental results.

Driving in Traffic: Short-Range Sensing for Urban Collision Avoidance

Chuck Thorpe et al. — Mon, 13 May 2013 13:40:20 PDT

Intelligent vehicles are beginning to appear on the market, but so far their sensing and warning functions only work on the open road. Functions such as runoff-road warning or adaptive cruise control are designed for the uncluttered environments of open highways. We are working on the much more difficult problem of sensing and driver interfaces for driving in urban areas. We need to sense cars and pedestrians and curbs and fire plugs and bicycles and lamp posts; we need to predict the paths of our own vehicle and of other moving objects; and we need to decide when to issue alerts or warnings to both the driver of our own vehicle and (potentially) to nearby pedestrians. No single sensor is currently able to detect and track all relevant objects. We are working with radar, ladar, stereo vision, and a novel light-stripe range sensor. We have installed a subset of these sensors on a city bus, driving through the streets of Pittsburgh on its normal runs. We are using different kinds of data fusion for different subsets of sensors, plus a coordinating framework for mapping objects at an abstract level.

Dependable Perception for Robots

Chuck Thorpe et al. — Mon, 13 May 2013 13:40:17 PDT

The weakest link in many mobile robots is perception. In order to build robots that are reliable and dependable and safe, we need to build robots that can see. Perception is becoming a solved problem for certain constrained environments. But for robots working outdoors, and at high speeds, and in close proximity to people, perception is still incomplete. Our robots need to see objects; to detect motion; and to detect which of those objects are people. In the current state of the art, this requires multiple sensors and multiple means of interpretation. This paper illustrates those principles in the context of the CMU Navlab Group's work on vehicle safety for busses and passenger cars.

Benefits Estimation of Sensor-Friendly Vehicle and Roadway Cooperative Safety Systems

James A. Misener et al. — Mon, 13 May 2013 13:40:13 PDT

An analysis was performed to estimate the potential national costs and benefits of cooperative vehicle and roadway measures to enhance the effectiveness of driver assistance systems. These cooperative measures ?query-response communication system, light emitting diode brake light messaging, radar cross section paint striping modifications, fluorescent paint for lane and other marking applications, passive amplifiers on license plates, spatial tetrahedral arrays of reflectors, and in-vehicle corner cubes ?are briefly described, along with assumptions that were made regarding performance. For the example lane departure case, the incremental nationwide effectiveness over an autonomous collision avoidance system is estimated and monetized. This was generally determined with respect to annual crash reduction savings, although the technique used allows other mobility benefits to be considered. The marginal benefits of providing each sensor friendly technology were then calculated and aggregated across the various IVI services so that a total marginal benefit was determined for each technology. Complementing this, a method has been established to estimate the magnitude of at- and near-intersection LVNM crashes for these technologies. Together, these methods illustrate national benefits across all crash types (the three-step process) and a more focused means to estimate benefits for a particular crash type (rear end collisions at or near intersections) ?and provide a composite approach to the problem.

Robust Monocular Visual Odometry for a Ground Vehicle in Undulating Terrain

Ji Zhang et al. — Mon, 13 May 2013 13:40:08 PDT

Here we present a robust method for monocular visual odometry capable of accurate position estimation even when operating in undulating terrain. Our algorithm uses a steering model to separately recover rotation and translation. Robot 3DOF orientation is recovered by minimizing image projection error, while, robot translation is recovered by solving an NP-hard optimization problem through an approximation. The decoupled estimation ensures a low computational cost. The proposed method handles undulating terrain by approximating ground patches as locally ﬂat but not necessarily level, and recovers the inclination angle of the local ground in motion estimation. Also, it can automatically detect when the assumption is violated by analysis of the residuals. If the imaged terrain cannot be sufﬁciently approximated by locally ﬂat patches, wheel odometry is used to provide robust estimation. Our ﬁeld experiments show a mean relative error of less than 1%.

Automated Crop Yield Estimation for Apple Orchards

Qi Wang et al. — Mon, 13 May 2013 13:40:05 PDT

Crop yield estimation is an important task in apple orchard management. The current manual sampling-based yield estimation is time-consuming, labor-intensive and inaccurate. To deal with this challenge, we developed a computer vision-based system for automated, rapid and accurate yield estimation. The system uses a two-camera stereo rig for image acquisition. It works at nighttime with controlled artificial lighting to reduce the variance of natural illumination. An autonomous orchard vehicle is used as the support platform for automated data collection. The system scans both sides of each tree row in orchards. A computer vision algorithm detects and registers apples from acquired sequential images, and then generates apple counts as crop yield estimation. We deployed the yield estimation system in Washington state in September, 2011. The results show that the system works well with both red and green apples in the tall-spindle planting system. The crop yield estimation errors are -3.2% for a red apple block with about 480 trees, and 1.2% for a green apple block with about 670 trees.

Global Pose Estimation with Limited GPS and Long Range Visual Odometry

Joem Rehder et al. — Mon, 13 May 2013 13:40:02 PDT

Here we present an approach to estimate the global pose of a vehicle in the face of two distinct problems; first, when using stereo visual odometry for relative motion estimation, a lack of features at close range causes a bias in the motion estimate. The other challenge is localizing in the global coordinate frame using very infrequent GPS measurements. Solving these problems we demonstrate a method to estimate and correct for the bias in visual odometry and a sensor fusion algorithm capable of exploiting sparse global measurements. Our graph-based state estimation framework is capable of inferring global orientation using a unified representation of local and global measurements and recovers from inaccurate initial estimates of the state, as intermittently available GPS information may delay the observability of the entire state. We also demonstrate a reduction of the complexity of the problem to achieve real-time throughput. In our experiments, we show in an outdoor dataset with distant features where our bias corrected visual odometry solution makes a fivefold improvement in the accuracy of the estimated translation compared to a standard approach. For a traverse of 2km we demonstrate the capabilities of our graph-based state estimation approach to successfully infer global orientation with as few as 6 GPS measurements and with two-fold improvement in mean position error using the corrected visual odometry.

Tactile Sensing by the Sole of the Foot Part II: Calibration and Real-time Processing

Abhinav Kalamdani et al. — Fri, 10 May 2013 14:03:07 PDT

This paper introduces prototype experimental apparatus and the calibration and real-time signal processing required to investigate stability in standing, walking and running of humanoid robots using pressure sensing at the sole-of-the-foot contact. The system can provide very good spatial or temporal resolution and these can be traded off against each other dynamically to accommodate the instantaneous requirement, for example, sparsely sampling the whole sole during static balancing vs. densely sampling the impact region during walking or running. Dynamic variation in sampling policy during different phases of the gait is foreseen so as to optimise utilisation of the total sampling bandwidth available. Periodic signals like walking and running would be sampled repetitively, achieving by accumulation both high spatial and high temporal resolution.

Nonrigid Structure from Motion in Trajectory Space

Ijaz Akhter et al. — Fri, 10 May 2013 13:40:29 PDT

Existing approaches to nonrigid structure from motion assume that the instantaneous 3D shape of a deforming object is a linear combination of basis shapes, which have to be estimated anew for each video sequence. In contrast, we propose that the evolving 3D structure be described by a linear combination of basis trajectories. The principal advantage of this approach is that we do not need to estimate any basis vectors during computation. We show that generic bases over trajectories, such as the Discrete Cosine Transform (DCT) basis, can be used to compactly describe most real motions. This results in a significant reduction in unknowns, and corresponding stability in estimation. We report empirical performance, quantitatively using motion capture data, and qualitatively on several video sequences exhibiting nonrigid motions including piece-wise rigid motion, partially nonrigid motion (such as a facial expression), and highly nonrigid motion (such as a person dancing).

Linear Motion Estimation for Systems of Articulated Planes

Ankur Datta et al. — Fri, 10 May 2013 13:40:24 PDT

In this paper, we describe the explicit application of articulation constraints for estimating the motion of a system of planes. We relate articulations to the relative homography between planes and show that for affine cameras, these articulations translate into linear equality constraints on a linear least squares system, yielding accurate and numerically stable estimates of motion. The global nature of motion estimation allows us to handle areas where there is limited texture information and areas that leave the field of view. Our results demonstrate the accuracy of the algorithm in a variety of cases such as human body tracking, motion estimation of rigid, piecewise planar scenes and motion estimation of triangulated meshes.

An Overview of the Northern Influences on Tolkien's Works

Gloriana St. Clair — Wed, 08 May 2013 13:32:30 PDT

J.R.R.Tolkien studied the Old Norse literature and mythology thoroughly. While knowing Northern literature does not provide a key to unlock the meanings of his major works, his characters, creatures, implements, customs, incidents, and themes do have antecedents in the Eddas and sagas. This paper assesses the extent and impact of those antecedents.

Volsunga Saga and Narn: Some Analogies

Gloriana St. Clair — Wed, 08 May 2013 13:32:29 PDT

"Narn", one of the works in the Unfinished Tales, has many parallels with the thirteenth-century Old Norse Volsunga Saga, which Tolkien read and studied, This paper will assess comparisons between the heroes, women, dragons, plots, and tokens for their contribution to understanding Tolkien's relationship to his sources, and will note Tolkien's craft in source-assimilation.

Tolkien as Reviser: A Case Study

Gloriana St. Clair — Wed, 08 May 2013 13:32:27 PDT

The publication of drafts of The Lord of the Rings allows scholars to assess Tolkien as a reviser. A comparison of the early presentations of Gondor in The History of "The Lord of the Rings" , with the finished scenes indicates the nature and direction of Tolkien's changes. This paper will discuss how the process of revision contributed to the overall effect of the work.

Learning Opportunity Costs in Multi-Robot Market Based Planners

Jeff Schneider et al. — Wed, 01 May 2013 13:18:49 PDT

Direct human control of multi-robot systems is limited by the cognitive ability of humans to coordinate numerous interacting components. In remote environments, such as those encountered during planetary or ocean exploration, a further limit is imposed by communication bandwidth and delay. Market based planning can give humans a higher-level interface to multi-robot systems in these scenarios. Operators provide high level tasks and attach a reward to the achievement of each task. The robots then trade these tasks through a market based mechanism. The challenge for the system designer is to create bidding algorithms for the robots that yield high overall system performance. Opportunity cost provides a nice basis for such bidding algorithms since it encapsulates all the costs and benefits we are interested in. Unfortunately, computing it can be difficult. We propose a method of learning opportunity costs in market based planners. We provide analytic results in simplified scenarios and empirical results on our FIRE simulator, which focuses on exploration of Mars by multiple, heterogeneous rovers.

Automatic Construction of Active Appearance Models as an Image Coding Problem

Simon Baker et al. — Wed, 01 May 2013 13:18:45 PDT

The automatic construction of Active Appearance Models (AAMs) is usually posed as finding the location of the base mesh vertices in the input training images. In this paper, we re-pose the problem as an energy-minimizing image coding problem and propose an efficient gradient-descent algorithm to solve it.

Semantic-based Biomedical Image Indexing and Retrieval

Yanxi Liu et al. — Wed, 01 May 2013 13:18:42 PDT

This paper summarizes our work and our understanding on volumetric pathological neuroimage retrieval under the framework of classification-driven feature selection. The main effort concerns off-line image feature space reduction for improved image indexing feature discriminating power as well as reduced computational cost during on-line pathological neuroimage retrieval.

Reconstructing 3D Human Pose from 2D Image Landmarks

Varun Ramakrishna et al. — Wed, 01 May 2013 13:18:37 PDT

Reconstructing an arbitrary configuration of 3D points from their projection in an image is an ill-posed problem. When the points hold semantic meaning, such as anatomical landmarks on a body, human observers can often infer a plausible 3D configuration, drawing on extensive visual memory. We present an activity-independent method to recover the 3D configuration of a human figure from 2D locations of anatomical landmarks in a single image, leveraging a large motion capture corpus as a proxy for visual memory. Our method solves for anthropometrically regular body pose and explicitly estimates the camera via a matching pursuit algorithm operating on the image projections. Anthropometric regularity (i.e., that limbs obey known proportions) is a highly informative prior, but directly applying such constraints is intractable. Instead, we enforce a necessary condition on the sum of squared limb-lengths that can be solved for in closed form to discourage implausible configurations in 3D. We evaluate performance on a wide variety of human poses captured from different viewpoints and show generalization to novel 3D configurations and robustness to missing data.

3D Reconstruction of a Moving Point from a Series of 2D Projections

Hyun Soo Park et al. — Wed, 01 May 2013 13:18:33 PDT

This paper presents a linear solution for reconstructing the 3D trajectory of a moving point from its correspondence in a collection of 2D perspective images, given the 3D spatial pose and time of capture of the cameras that produced each image. Triangulation-based solutions do not apply, as multiple views of the point may not exist at each instant in time. A geometric analysis of the problem is presented and a criterion, called reconstructibility, is defined to precisely characterize the cases when reconstruction is possible, and how accurate it can be. We apply the linear reconstruction algorithm to reconstruct the time evolving 3D structure of several real-world scenes, given a collection of non-coincidental 2D images.

Dynamic Seethroughs: Synthesizing Hidden Views of Moving Objects

Peter Barnum et al. — Wed, 01 May 2013 13:18:29 PDT

This paper presents a method to create an illusion of seeing moving objects through occluding surfaces in a video. This illusion is achieved by transferring information from a camera viewing the occluded area. In typical view interpolation approaches for 3D scenes, some form of correspondence across views is required. For occluded areas, establishing direct correspondence is impossible as information is missing in one of the views. Instead, we use a 2D projective invariant to capture information about occluded objects (which may be moving). Since invariants are quantities that do not change across views, a visually compelling rendering of hidden areas is achieved without the need for explicit correspondences. A piece-wise planar model of the scene allows the entire rendering process to take place without any 3D reconstruction, while still producing visual parallax. Because of the simplicity and robustness of the 2D invariant, we are able to transfer both static backgrounds and moving objects in real time. A complete working system has been implemented that runs live at 5Hz. Applications for this technology include the ability to look through corners at tight intersections for automobile safety, concurrent visualization of a surveillance camera network, and monitoring systems for patients/elderly/children.

Modeling the Product Manifold of Posture and Motion

Ankur Datta et al. — Wed, 01 May 2013 13:18:25 PDT

Long-term human motion is composed of an ensemble of different activities with varying complexity. This makes it challenging to develop models to accurately estimate human motion. In this paper, we exploit the dependencies that exist between posture and motion for long-term human motion estimation. We propose to model the nonlinear motion manifold as a collection of local linear models, noting that given a particular posture, the variation in motion for that posture can be well-approximated by a linear model. A collection of local linear models is easy to fit and also has the expressiveness to encode several activities in any arbitrary order. Furthermore, to account for the varying complexity of different activities, each local linear model can have a different dimensionality. A collection of local linear models, thus, avoids the limitation of global models that require a uniform dimensionality for the latent motion manifold. This model allows us to linearly regularize motion estimation algorithms over the nonlinear human motion manifold. Our results demonstrate that a collection of local linear models provides an effective representation for the motion manifold when compared to other global models such as the bilinear model and the Principal Component Analysis.