Search results

Now showing 1 - 10 of 191
  • Publication
    Anthropometry for Computer Graphics Human Figures
    (1989) Grosso, Marc R; Quach, Richard D.; Otani, Ernest; Zhao, Jianmin; Wei, Susanna; Ho, Pei-Hwa; Lu, Jiahe; Badler, Norman I
    Anthropometry as it applies to Computer Graphics is examined in this report which documents the Anthropometry work done in the Computer Graphics Research Laboratory at the University of Pennsylvania from 1986 to 1988. A detailed description of the basis for this work is given along with examples of the variability of computer graphics human figures resulting from this work. Also discussed is the unique and versatile user interface developed to allow easy manipulation of the data used to describe the anthropometric parameters required to define human figure models. The many appendices contain the specifics of our models as well as much of the data used to define the models.
  • Publication
    A GPU-Based Approach for Real-Time Haptic Rendering of 3D Fluids
    (2008-12-01) Yang, Meng; Safonova, Alla; Kuchenbecker, Katherine J; Zhou, Zehua
    Real-time haptic rendering of three-dimensional fluid flow will improve the interactivity and realism of video games and surgical simulators, but it remains a challenging undertaking due to its high computational cost. In this work we propose an innovative GPUbased approach that enables real-time haptic rendering of highresolution 3D Navier-Stokes fluids. We show that moving the vast majority of the computation to the GPU allows for the simulation of touchable fluids at resolutions and frame rates that are significantly higher than any other recent real-time methods without a need for pre-computations [Baxter and Lin 2004; Mora and Lee 2008; Dobashi et al. 2006].
  • Publication
    Learning Determinantal Point Processes
    (2011-07-01) Taskar, Ben; Kulesza, Alex
    Determinantal point processes (DPPs), which arise in random matrix theory and quantum physics, are natural models for subset selection problems where diversity is preferred. Among many remarkable properties, DPPs other tractable algorithms for exact inference, including computing marginal probabilities and sampling; how- ever, an important open question has been how to learn a DPP from labeled training data. In this paper we propose a natural feature-based parameterization of conditional DPPs, and show how it leads to a convex and efficient learning formulation. We analyze the relationship between our model and binary Markov random fields with repulsive potentials, which are qualitatively similar but computationally intractable. Finally, we apply our approach to the task of extractive summarization, where the goal is to choose a small subset of sentences conveying the most important information from a set of documents. In this task there is a fundamental tradeoff between sentences that are highly relevant to the collection as a whole, and sentences that are diverse and not repetitive. Our parameterization allows us to naturally balance these two characteristics. We evaluate our system on data from the DUC 2003/04 multi- document summarization task, achieving state-of-the-art results.
  • Publication
    Efforts in Preparation for Jack Validation
    (1997-12-01) Badler, Norman I; Azuola, Francisco; Huh, Suejung; Ho, Pei-Hwa; Kokkevis, Evangelos; Ting, Bond-Jay
    This document presents a detailed record of the methodologies, assumptions, limitations, and references used in creating the human figure model in Jack, a program that displays and manipulates articulated geometric figures. This report reflects current efforts to develop and refine Jack software to enable its validation and verification as a tool for performing human engineering analysis. These efforts include human figure model improvements, statistical anthropometric data processing methods, enhanced human figure model construction and measuring methods, and automated accomodation analysis. This report discusses basic details of building human models, model anthropometry, scaling, Jack anthropometry-based human models, statistical data processing, figure generation tools, anthropometric errors, inverse dynamics, smooth skin implementation, guidelines used in estimating landmark locations on the model, and recommendations for validating and verifying the Jack human figure model.
  • Publication
  • Publication
    Design Concepts for Automating Maintenance Instructions
    (2000-02-01) Badler, Norman I; Erignac, Charles A; McDonald, Patrick Vincent; Sanchez, Edgar; Boyle, Edward S
    This research task was performed under the Technology for Readiness and Sustainment (TRS) contract (F33615-99-D-6001) for the Air Force Research Laboratory (AFRL), Sustainment Logistics Branch (HESS) at Wright-Patterson AFB, OH. The period of performance spanned one year starting 29 January 1999. The objective of this task was to develop and demonstrate a framework that can support the automated validation and verification of aircraft maintenance Technical Orders (TOs). The research team examined all stages ofTO generation to determine which tasks most warranted further research. From that investigation, validation and verification of appropriate, safe, and correct procedure steps emerged as the primary research target. This process would be based on available computer-aided design (CAD) data, procedure step ordering from existing sources, and human models. This determination was based on which tasks could yield the greatest impact on the authoring process and offer the greatest potential economic benefits. The team then developed a research roadmap and outlined specific technologies to be addressed in possible subsequent Air Force research tasks. To focus on the potential technology integration of the validation and verification component into existing or future TO generation procedures, we defined a demonstration scenario. Using the Front Uplock Hook assembly from an F/A-18 as the subject, we examined task procedure steps and failures that could be exposed by automated validation tools. These included hazards to personnel, damage to equipment, and incorrect disassembly order. Using the Parameterized Action Representation (PAR) developed on previous projects for actions and equipment behaviors, we characterized procedure steps and their positive and negative consequences. Finally, we illustrated a hypothetical user interface extension to a typical Interactive Electronic Technical Manual (IETM) authoring system to demonstrate how this process might appear to the TO author.
  • Publication
    Real-Time Evacuation Simulation in Mine Interior Model of Smoke and Action
    (2010-01-01) Huang, Pengfei; Kider, Joseph T; Sunshine-Hill, Ben; McCaffrey, Jonathan B.; Rios, Desiree Velazquez; Badler, Norman I; Kang, Jinsheng
    Virtual human crowd models have been used in the simulation of building and urban evacuation, but have not yet applied to underground coal mine operations and escape situations with emphasis on smoke, fires and physiological behaviors. We explore this through a real-time simulation model, MIMOSA (Mine Interior Model Of Smoke and Action), which integrates an underground coal mine virtual environment, a fire and smoke propagation model, and a human physiology and behavior model. Each individual agent has a set of physiological parameters as variables of time and environment, simulating a miner’s physiological condition during normal operations as well as during emergencies due to fire and smoke. To obtain appropriate agent navigation in the mine environment, we have extended the HiDAC framework (High- Density Autonomous Crowds) navigation from a grid-based cell-portal graph to a geometrybased portal path and integrated a novel cellportal and shortest path visibility algorithm.
  • Publication
    A permutation-augmented sampler for DP mixture models
    (2007-06-01) Taskar, Ben; Liang, Percy; Jordan, Michael
    We introduce a new inference algorithm for Dirichlet process mixture models. While Gibbs sampling and variational methods focus on local moves, the new algorithm makes more global moves. This is done by introducing a permutation of the data points as an auxiliary variable. The algorithm is a blocked sampler which alternates between sampling the clustering and sampling the permutation. The key to the efficiency of this approach is that it is possible to use dynamic programming to consider all exponentially many clusterings consistent with a given permutation. We also show that random projections can be used to effectively sample the permutation. The result is a stochastic hill-climbing algorithm that yields burn-in times significantly smaller than those of collapsed Gibbs sampling.
  • Publication
    Pedestrian Anomaly Detection Using Context-Sensitive Crowd Simulation
    (2012-11-01) Boatright, Cory D; Kapadia, Mubbasir; Shapira, Jennie M; Badler, Norman I
    Detecting anomalies in crowd movement is an area of considerable interest for surveillance and security applications. The question we address is: What constitutes an anomalous steering choice for an individual in the group? Deviation from “normal” behavior may be defined as a subject making a steering decision the observer would not, provided the same circumstances. Since the number of possible spatial and movement configurations is huge and human steering behavior is adaptive in nature, we adopt a context-sensitive approach to assess individuals rather than assume population-wide homogeneity. When presented with spatial trajectories from processed surveillance data, our system creates a shadow simulation. The simulation then establishes the current, local context for each agent and computes a predicted steering behavior against which the person’s actual motion can be statistically compared. We demonstrate the efficacy of our technique with preliminary results using real-world tracking data from the Edinburgh Pedestrian Dataset.
  • Publication
    Sparsity in Dependency Grammar Induction
    (2010-07-01) Taskar, Ben; Pereira, Fernando CN; Graca, Joao V; Gillenwater, Jennifer; Ganchev, Kuzman
    A strong inductive bias is essential in unsupervised grammar induction. We explore a particular sparsity bias in dependency grammars that encourages a small number of unique dependency types. Specifically, we investigate sparsity-inducing penalties on the posterior distributions of parent-child POS tag pairs in the posterior regularization (PR) framework of Graça et al. (2007). In experiments with 12 languages, we achieve substantial gains over the standard expectation maximization (EM) baseline, with average improvement in attachment accuracy of 6.3%. Further, our method outperforms models based on a standard Bayesian sparsity-inducing prior by an average of 4.9%. On English in particular, we show that our approach improves on several other state-of-the-art techniques.