Peter Weyhrauch

System for authoring highly interactive, personality-rich interactive characters

We describe an innovative system for authoring expressive, fully autonomous interactive characters. The focus of our work is creating a system to allow rich authoring that captures as much of the artistic intent of the author in procedural form as we can, and that provides automatic support for expressive execution of that content. The system is composed of two parts: (1)a programming language with unusual language features including concurrency, reflection, backtracking, continuously monitored expressions, and a model of emotion, that was created for the expression of interactive self-animating characters; and (2) a motion synthesis system that combines hand-animated motion data with artistically authored procedures for generalizing the motion while preserving the artistic intent. This system has been used to create over a dozen interactive characters, which have been shown at juried venues, as well as being deployed commercially. We describe how artistic qualities important to interactive characters are encoded and supported using this system, and demonstrate the system with an implemented interactive character.

Prevention of Surgical Skill Decay

The U.S. military medical community spends a great deal of time and resources training its personnel to provide them with the knowledge and skills necessary to perform life-saving tasks, both on the battlefield and at home. However, personnel may fail to retain specialized knowledge and skills if they are not applied during the typical periods of nonuse within the military deployment cycle, and retention of critical knowledge and skills is crucial to the successful care of warfighters. For example, we researched the skill and knowledge loss associated with specialized surgical skills such as those required to perform laparoscopic surgery (LS) procedures. These skills are subject to decay when military surgeons perform combat casualty care during their deployment instead of LS. This article describes our preliminary research identifying critical LS skills, as well as their acquisition and decay rates. It introduces models that identify critical skills related to laparoscopy, and proposes objective metrics for measuring these critical skills. This research will provide insight into best practices for (1) training skills that are durable and resistant to skill decay, (2) assessing these skills over time, and (3) introducing effective refresher training at appropriate intervals to maintain skill proficiency.

Modeling ground soldier situational awareness for constructive simulation with rules

The behavior models that control simulated warfighters in most modeling and simulation (M&S) efforts are fairly simple, relying predominantly on behavior scripting and simple rules to produce actions. As a result, the simulated entities do not reflect critical situational awareness factors used by Ground Soldiers or allow for the modeling of devices that influence situational awareness, such as user defined operating pictures (UDOPs). This paper describes our approach to this challenge, providing 1) a rule-based method for modeling Ground Soldier situational awareness and devices that influence situational awareness and 2) a user friendly graphical authoring tool for creating these rules. We present a requirements analysis of this modeling task and discuss and provide examples of how our method may be employed for modeling Soldier perception and inferences as well as devices that affect situational awareness.

Design of Training Protocol for Perceptual and Technical Skills in a Minimally Invasive Surgery

This study investigates how the technical and perceptual skills in laparoscopic surgery, typically acquired separately in the initial learning phases, can be trained together. A task analysis and cognitive task analysis were conducted using a cholecystectomy procedure and a fundoplication procedure. An experiment was conducted to examine the interaction of technical and perceptual skill learning. Subjects were divided into three groups based on order of skills training: 1) technical-perceptual-combined skills training order, 2) perceptual-technical-combined skills training order, and 3) combined skills training. After the training sessions, performance was evaluated using the combined skill. Preliminary results indicate that performance of the group trained in the combined skills condition performed equally quickly as those who trained the technical and perceptual skills separately first. In addition, the number of technical errors and perceptual errors committed were lower. This suggests that surgical skills training may be more efficient if perceptual learning is combined with motor skills during the initial phases of training. This has implications for the design of surgical training simulators and surgical education in general.© 2012 ASME

Towards an architecture for collaborative human/AI control of interactive characters

Interactive characters - widely used for entertainment, education, and training - may be controlled by human operators or software agents. Human operators are extremely capable in supporting this interaction, but the cost per interaction is high. Believable agents are software controlled characters that attempt natural and engaging interaction. Believable agents are inexpensive to operate, but they cannot currently support a full range of interaction. To combine the strengths of human operators and believable agents, this paper presents steps toward an architecture for collaborative human/AI control of interactive characters. A human operator monitors the interactions of users with a group of believable agents and acts to intervene and improve interaction. We identify challenges in constructing this architecture and propose an architecture design to address these challenges. We discuss technologies that enable the operator to monitor many believable agents at once and act to intervene quickly and on many levels of granularity. To increase speed and usability, we employ principles of narrative structure in the design of our architectural components.

Development of Virtual Simulations for Medical Team Training: An Evaluation of Key Features

Medical team training (MTT) conducted in a virtual environment fosters growth in cognitive, technical, and clinical aptitudes while offering advantages of flexibility, cost, and ease of scheduling over traditional high- fidelity simulations. Growing technology facilitates innovations to improve the ability to emulate roles, rules, resources, and fidelity. Our objective was to evaluate elements of key features that inform technical specifications for virtual simulations. A narrative review included 27 articles as relevant to elaborate on five key features identified as critical to development of virtual environments for MTT: automated assessment, task fidelity, interface modality, virtual teammates, and adaptability. Designers continue to improve the technology of virtual reality to create better and more enhanced training modules. We must better understand how variances in simulation features impacts performance outcomes and learned behavior. Future research can more deeply examine features beyond the five reviewed here to guide development of effective, cost-efficient virtual simulations for MTT.

Thematic Issues in Analysis and Visualization of Emergency Department Patient Flow

Emergency Department (ED) congestion is a significant problem affecting clinical outcomes, patient satisfaction and hospital costs. Identifying and resolving bottlenecks in the flow of patients from the ED to eventual admission or discharge has the potential to reduce wait times, improve care for individual patients, and increase the volume of patients treated at the hospital over time. Our objective was to review methods commonly used to measure, analyze, and visualize patient flow, characterize drawbacks to these methods, and identify areas in which analysis and visualization can be improved to make bottlenecks easier to identify and resolve. Sixty-five articles obtained from PubMed and Google Scholar searches were reviewed to identify: (1) variables used to measure ED throughput; (2) downstream effects of ED congestion; (3) factors contributing to ED congestion; (4) techniques used to predict or respond to ED congestion; and (5) tools used to visualize data on ED throughput. Hospital resource availability, patient demographics, and environmental factors have all been used to predict contributors to ED congestion. Unfortunately, the hospital practices most critical to ED congestion are unlikely to change as they involve increasing the number of beds and providers or modifying protocols with EMS, insurance, and other care facilities. Therefore, interventions addressing optimization of ED resource allocation and visualization of ED data are the best avenue to yield more efficient ED operation.

Modeling and Comparing Nurse and Physician Trauma Assessment Skills

Medical providers must master a large number of complicated tasks to deliver quality care and minimize unwanted clinical outcomes. In order to optimally train these tasks, medical training systems would benefit from models of skill that enable objective assessment of proficiency and define important declarative knowledge, cognitive states, and decision-making rules that are necessary for effective learning and performance. This article describes the Methodology for Annotated Skill Trees (MAST), a skill-modeling framework that facilitates the creation of descriptive and rule-based content that supports skill acquisition. This framework is used to generate models of trauma assessment skills from two existing curricula: Advanced Trauma Life Support (ATLS) and the Trauma Nurse Core Course (TNCC). Key differences between these curriculas teaching methods for the same procedure and skill are highlighted through the use of the model framework. The framework comparison provides insight into the underlying teaching approach and highlights the fact that some skills are not represented in medical education materials.

Hybrid Modeling of Cyber Adversary Behavior

Cyber adversaries continue to become more proficient and sophisticated, increasing the vulnerability of the network systems that pervade all aspects of our lives. While there are many approaches to modeling network behavior and identifying anomalous and potentially malicious traffic, most of these approaches detect attacks once they have already occurred, enabling reaction only after the damage has been done. In traditional security studies, mitigating attacks has been a focus of many research and planning efforts, leading to a rich field of adversarial modeling to represent and predict what an adversary might do. In this paper, we present an analogous approach to modeling cyber adversaries to gain a deeper understanding of the behavioral dynamics underlying cyber attacks and enable predictive analytics and proactive defensive planning. We present a hybrid modeling approach that combines aspects of cognitive modeling, decision-theory, and reactive planning to capture different facets of adversary decision making and behavior.

A Flexible Framework for the Creation of Narrative-Centered Tools

To better support the creation of narrative-centered tools, developers need a flexible framework to integrate, catalog, select, and reuse narrative models. Computational models of narrative enable the creation of software tools to aid narrative processing, analysis, and generation. Narrative-centered tools explicitly or implicitly embody one or more models of narrative by their definition. However, narrative model creation is often expensive and difficult with no guaranteed benefit to the end system. This paper describes our preliminary approach towards creating the SONNET narrative framework, a flexible framework to integrate, catalog, select, and reuse narrative models, thereby lowering development costs and improving benefits from each model. The framework includes a lightweight ontology language for the definition of key terms and interrelationships among them. The framework specifies model metadata to allow developers to discover and understand models more readily. We discuss the structure of this framework and ongoing development incorporating narrative models.