J. Peter Kincaid

Linguistic cues and memory for synthetic and natural speech.

Past research has demonstrated that there are cognitive processing costs associated with comprehension of speech generated by text-to-speech synthesizers, relative to comprehension of natural speech. This finding has important performance implications for the many applications that use such systems. The purpose of this study was to ascertain whether certain characteristics of synthetic speech slow on-line, real-time cognitive processing. Whereas past research has focused on the phonemic acoustic structure of synthetic speech, we manipulated prosodic, syntactic, and semantic cues in a task requiring participants to recall sentences spoken either by a human or by one of two speech synthesizers. The findings were interpreted to suggest that inappropriate prosodic modeling in synthetic speech was the major source of a performance differential between natural and synthetic speech. Prosodic cues, along with others, guide the parsing of speech and provide redundancy. When these cues are absent or inaccurate, the additional burden placed on working memory may exceed its capacity, particularly in time-limited, demanding tasks. Actual or potential applications of this research include improvement of text-to-speech output systems in warning systems, feedback devices in aerospace vehicles, educational and training modules, aids for the handicapped, consumer products, and technologies designed to increase the functional independence of older adults.

Simulation techniques for training emergency response

The Institute for Simulation and Training (IST), University of Central Florida, in conjunction with the US Army and the Orange County (Florida) Fire Rescue Department, have developed and fielded a series of simulation techniques for conducting disaster exercises and training public safety personnel to respond to disasters. Simulator training offers several important advantages, including cost and safety. We are also gathering persuasive evidence that training effectiveness is substantially improved by the use of simulation as compared with traditional field exercises. Our programme is now in its seventh year and has concentrated on training for two main skill domains: emergency management incident command, and emergency medical care performed in the field. We provide a programmatic overview, including descriptions and illustrations of the following projects: Project Plowshares, sponsored by the US Army, involved conversion of a war game simulation to support civilian emergency management exercises; the Combat Trauma Patient Simulator, a project funded by the US Army and the manufacturer (METI), uses a highly realistic medical simulator connected to the internet to model treatment of battlefield casualties; crisis management field exercises have been supported by the use of simulation techniques and technology, in conjunction with the US and Swedish military and local public safety agencies, particularly the Orange County Fire Rescue Department.

A framework for web-based dissemination of models and lessons learned from emergency-response exercises and operations

Emerging web-based technologies enable information sharing within and among rescue agencies and organisations. This development can facilitate education, training and development through effective distribution of documentation, models and lessons learned from training exercises and rescue operations. We discuss central aspects of this enterprise such as the need for domain-specific data models, methodologies for detailed exercise documentation, and the design of presentation and visualisation tools. Based on this investigation, we present a framework that supports modelling, documentation and visualisation of rescue operations for web-based distribution. We briefly describe the implementation of the components of the framework. Finally, we discuss how to apply the methodology and framework to training of first responders and university-level education of emergency managers and officials.

Electronic Authoring and Delivery of Technical Information.

Current developments in automating the processes to author technical information and deliver it using microcomputers are described in this article. The Department of Defense (DoD) has directed that the entire logistics support system (including technical information now contained in printed manuals) be computer-based for new major systems entering production during the 1990s. Addressed are issues relating to authoring efficiency, information access, user acceptance, and screen formats. The article also reviews several ongoing projects and discusses implications for making the transition from paper to computer-based technical manuals.

Computer Aided Authoring and Editing.

Using computers to author and edit text is now feasible and economical. This article describes two elements of a computer based publishing system which aid authors by automatically generating certain kinds of training materials, and in writing easily understood text Authoring routines automatically format and compose text and merge stored graphics with the text. Highly redundant training materials require only a small data base to produce a sizeable workbook. Editing routines aid authors in writing readable text by: 1) flagging uncommon words and long sentences; 2) suggesting substitutes for awkward or difficult words and phrases; and 3) indicating the readability grade level. These routines are operating as elements of a prototype publishing system in the U.S. Navys Training Analysis and Evaluation Group minicomputer. They are being developed for use in the U.S. Navys computer based publishing system.

OUPS: A Combined Approach Using SMOTE and Propensity Score Matching

Building accurate classifiers is difficult when using data that is skewed or imbalanced which is typical of real world data sets. Two popular approaches that have been applied for improving classification accuracy and statistical comparisons of imbalanced data sets are: synthetic minority over-sampling technique (SMOTE) and propensity score matching (PSM). A novel sampling approach is introduced referred to as over-sampling using propensity scores (OUPS) that blends the two and is simple and easy to perform resulting in improvement in accuracy and sensitivity over both SMOTE and PSM. The performance of our proposed approach is assessed using a simulation experiment and several performance metrics are shown where this approach fares and falls in comparison to the others.

Field of View Effects on Pilot Performance in Flight

For flight training, head-worn displays represent low-cost, wide field of regard, deployable systems when compared to traditional simulation facilities. However, current head-worn systems provide limited effective fields of view. Wide field of view alternatives promise to increase transfer of training effectiveness through enhanced situation awareness. To test this proposition, this experiment manipulated the pilots effective field of view and examined subsequent flight performance, which was measured primarily by runway alignment error and vertical track error. Results indicated a significant and quantifiable change in visual scan pattern, head movement, and flight control performance as the effective field of view was sequentially decreased. As field of view decreased, the average visual scan pattern changed to focus less out the window and more on the instruments inside the cockpit. The head range of movement significantly increased below an 80° horizontal × 54° vertical effective field of view as well as significantly decreasing runway alignment and vertical track performance, which occurred below 120° horizontal × 81° vertical effective field of view.

Emergency Vehicles that Become Accident Statistics: Understanding and Limiting Accidents Involving Emergency Vehicles

Emergency vehicle accidents represent a nationwide problem for those involved in emergency services. These accidents result in tremendous monetary damage to vehicle equipment, as well as being costly in terms of personal injury, morale, and the public image of associated fire departments. As a preliminary investigation, researchers at a medium-sized southeastern university worked in conjunction with a local fire rescue department to explore the factors related to emergency vehicle accidents. The reported effort represents a field study where data were gathered through literature reviews, source documents, interviews, and naturalistic observation. Findings indicate a lack of mandated, standardized training and evaluation procedures, multiple stress factors within the operating environment, and little on-the-job driving experience as possible reasons for this high accident rate. This paper concludes with several recommendations in three major areas: policy changes, training, and public awareness.

Computer Aids for Authoring Technical Text Written in Controlled English

This paper describes the development and validation of software to automate the authoring of training materials written in controlled English, such as Simplified English (SE). In SE, writers must adhere to many writing rules. While such materials are easy to read, they are very difficult to write. For example, use of short names for equipment must be consistent; also use of synonyms is not allowed. The software provides feedback regarding adherence to stipulated vocabulary and writing rules. Algorithms contained in the software include sentence parsing routines to verify that words are used according to their defined part of speech.

The effect of visual display on performance in mixed reality simulation

Abstract The United States Army continues to develop new and effective ways to use simulation for training. One example is the Non-Rated Crew Member Manned Module (NCM3), a simulator designed to train helicopter crewmembers in critical, high-risk tasks such as aerial door gunnery. The goal of this study was to evaluate visual display’s effect on performance in mixed reality aerial door gunnery. Two discrete groups of subjects, expert and novice, were employed in this study. Participants were randomly assigned to one of the two visual display treatments and executed three training scenarios in the NCM3. Independent variables were visual display, immersive tendency and Simulator Sickness Questionnaire scores. Dependent variables included performance, presence and simulator sickness change scores. The results of the study indicated no main effect of visual display on performance for the expert population while a main effect of visual display on performance was discovered for the novice population. Both visual treatment groups experienced the same degree of presence and simulator sickness. No relationship between an individual’s immersive tendency and their performance nor level of presence was found. Results of this study both support and challenge the commonly held notion that higher immersive simulation leads to better performance.