handbook of multisensor data fusion phần 9 pdf

©2001 CRC Press LLC estimation. The Level 1 process results in an evolving database that contains estimates of the position, velocity, attributes, and identities of physically constrained entities (e.g., targets and emitters). Subse- quently, automated reasoning methods are applied in an attempt to perform automated situation assess- ment and threat assessment. These automated reasoning methods are drawn from the discipline of artificial intelligence. Ultimately, the results of this dynamic process are displayed for a human user or analyst (via a human- computer interface (HCI) function). Note that this description of the data fusion process has been greatly simplified for conceptual purposes. Actual data fusion processing is much more complicated and involves an interleaving of the Level 1 through Level 3 (and Level 4) processes. Nevertheless, this basic orientation is often used in developing data fusion systems: the sensors are viewed as the information source and the human is viewed as the information user or sink. In one sense, the rich information from the sensors (e.g., the radio frequency time series and imagery) is compressed for display on a small, two-dimensional computer screen. Bram Ferran, the vice president of research and development at Disney Imagineering Company, recently pointed out to a government agency that this approach is a problem for the intelligence community. Ferran 8 argues that the broadband sensor data are funneled through a very narrow channel (i.e., the computer screen on a typical workstation) to be processed by a broadband human analyst. In his view, the HCI becomes a bottleneck or very narrow filter that prohibits the analyst from using his extensive pattern recognition and analytical capability. Ferran suggests that the computer bottleneck effectively defeats one million years of evolution that have made humans excellent data gatherers and processors. Interestingly, Clifford Stoll 9,10 makes a similar argument about personal computers and the multimedia misnomer. Researchers in the data fusion community have not ignored this problem. Waltz and Llinas 3 noted that the overall effectiveness of a data fusion system (from sensing to decisions) is affected by the efficacy of the HCI. Llinas and his colleagues 11 investigated the effects of human trust in aided adversarial decision support systems, and Hall and Llinas 12 identified the HCI area as a key research need for data fusion. Indeed, in the past decade, numerous efforts have been made to design visual environments, special displays, HCI toolkits, and multimedia concepts to improve the information display and analysis process. Examples can be found in the papers by Neal and Shapiro, 13 Morgan and Nauda, 14 Nelson, 15 Marchak and Whitney, 16 Pagel, 17 Clifton, 18 Hall and Wise, 19 Kerr et al., 20 Brendle, 21 and Steele, Marzen, and Corona. 22 A particularly interesting antisubmarine warfare (ASW) experiment was reported by Wohl et al. 23 Wohl and his colleagues developed some simple tools to assist ASW analysts in interpreting sensor data. The tools were designed to overcome known limitations in human decision making and perception. Although very basic, the support tools provided a significant increase in the effectiveness of the ASW analysis. The experiment suggested that cognitive-based tools might provide the basis for significant improvements in the effectiveness of a data fusion system. FIGURE 19.1 Joint directors of laboratories (JDL) data fusion process model. Sources Human Computer Interaction DATA FUSION DOMAIN Source Pre-Processing Level One Object Refinement Level Two Situation Refinement Level Three Threat Refinement Level Four Process Refinement Database Management System Support Database Fusion Database ©2001 CRC Press LLC 20 Assessing the Performance of Multisensor Fusion Processes 20.1 Introduction 20.2 Test and Evaluation of the Data Fusion Process Establishing the Context for Evaluation • T&E Philosophies • T&E Criteria • Approach to T&E • The T&E Process — A Summary 20.3 Tools for Evaluation: Testbeds, Simulations, and Standard Data Sets 20.4 Relating Fusion Performance to Military Effectiveness — Measures of Merit 20.5 Summary References 20.1 Introduction In recent years, numerous prototypical systems have been developed for multisensor data fusion. A paper by Hall, Linn, and Llinas 1 describes over 50 such systems developed for DoD applications even some 10 years ago. Such systems have become ever more sophisticated. Indeed, many of the prototypical systems summarized by Hall, Linn, and Llinas 1 utilize advanced identification techniques such as knowledge- based or expert systems, Dempster-Shafer interface techniques, adaptive neural networks, and sophisti- cated tracking algorithms. While much research is being performed to develop and apply new algorithms and techniques, much less work has been performed to formalize the techniques for determining how well such methods work or to compare alternative methods against a common problem. The issues of system performance and system effectiveness are keys to establishing, first, how well an algorithm, technique, or collection of techniques performs in a technical sense and, second, the extent to which these techniques, as part of a system, contribute to the probability of success when that system is employed on an operational mission. An important point to remember in considering the evaluation of data fusion processes is that those processes are either a component of a system (if they were designed-in at the beginning) or they are enhancements to a system (if they have been incorporated with the intention of performance enhance- ment). Said otherwise, it is not usual that the data fusion processes are “the” system under test; data fusion processes are said to be designed into systems rather than being systems in their own right. What is important to understand in this sense is that the data fusion processes contribute a marginal or piecewise James Llinas State University of New York ©2001 CRC Press LLC 21 Dirty Secrets in Multisensor Data Fusion* 21.1 Introduction 21.2 The JDL Data Fusion Process Model 21.3 Current Practices and Limitations in Data Fusion Level 1: Object Refinement • Level 2: Situation Refinement • Level 3: Threat Refinement • Level 4: Process Refinement • Human-Computer Interface (HCI) • Database Management 21.4 Research Needs Data Sources • Source Preprocessing • Level 1: Object Refinement • Level 2: Situation Refinement and Level 3: Threat Refinement • Human-Computer Interface (HCI) • Database Management • Level 4: Processing • Infrastructure Needs 21.5 Pitfalls in Data Fusion 21.6 Summary References 21.1 Introduction Over the past two decades, an enormous amount of Department of Defense (DoD) funding has been applied to the problem of data fusion systems, and a large number of prototype systems have been implemented. 1 The data fusion community has developed a data fusion process model, 2 a data fusion lexicon, 3 and engineering guidelines for system development. 4 Although a significant amount of progress has been made, 5,6 much work remains to be done. Hall and Garga, 7 for example, identified a number of pitfalls or problem areas in implementing data fusion systems. Hall and Llinas 8 described some short- comings in the use of data fusion systems to support individual soldiers, and M. J. Hall, S. A. Hall, and Ta te 9 addressed issues related to the effectiveness of human-computer interfaces for data fusion systems. This chapter summarizes recent progress in multisensor data fusion research and identifies areas in which additional research is needed. In addition, it describes some issues — or dirty secrets — in the current practice of data fusion systems. *This chapter is based on a paper by David L. Hall and Alan N. Steinberg, Dirty secrets of multisensor data fusion, Proceedings of the 2000 MSS National Symposium on Sensor Data Fusion, Vol. 1, pp. 1–16, June 2000, San Antonio, TX. David L. Hall The Pennsylvania State University Alan N. Steinberg Utah State University IV Sample Applications 22 A Survey of Multisensor Data Fusion Systems Mary L. Nichols Introduction • Recent Survey of Data Fusion Activities • Assessment of System Capabilities 23 Data Fusion for Developing Predictive Diagnostics for Electromechanical Systems Carl S. Byington and Amulya K. Garga Introduction • Aspects of a CBM System • The Diagnosis Problem • Multisensor Fusion Toolkit • Application Examples • Concluding Remarks 24 Information Technology for NASA in the 21st Century Robert J. Hansen, Daniel Cooke, Kenneth Ford and Steven Zornetzer Introduction • NASA Applications • Critical Research Investment Areas for NASA • High- Performance Computing and Networking • Conclusions 25 Data Fusion for a Distributed Ground-Based Sensing System Richard R. Brooks Introduction • Problem Domain • Existing Systems • Prototype Sensors for SenseIT • Software Architecture • Declarative Language Front-End • Subscriptions • Mobile Code • Diffusion Network Routing • Collaborative Signal Processing • Information Security • Summary 26 An Evaluation Methodology for Fusion Processes Based on Information Needs Hans Keithley Introduction • Information Needs • Key Concept • Evaluation Methodology ©2001 CRC Press LLC ©2001 CRC Press LLC 22 A Survey of Multisensor Data Fusion Systems 22.1 Introduction 22.2 Recent Survey of Data Fusion Activities 22.3 Assessment of System Capabilities References 22.1 Introduction During the past two decades, extensive research and development on multisensor data fusion has been performed for the Department of Defense (DoD). By the early 1990s, an extensive set of fusion systems had been reported for a variety of applications ranging from automated target recognition (ATR) and identification-friend-foe-neutral (IFFN) systems to systems for battlefield surveillance. Hall, Linn, and Llinas 1 provided a description of 54 such systems and an analysis of the types of fusion processing, the applications, the algorithms, and the level of maturity of the reported systems. Subsequent to that survey, Llinas and Antony 2 described 13 data fusion systems that performed automated reasoning (e.g., for situation assessment) using the blackboard reasoning architecture. By the mid-1990s, extensive commer- cial off-the-shelf (COTS) software was becoming available for different data fusion techniques and for decision support. Hall and Linn 3 described a survey of COTS software for data fusion and Buede 4,5 performed surveys and analyses of COTS software for decision support. This chapter presents a new survey of data fusion systems for DoD applications. The survey was part of an extensive effort to identify and assess DoD fusion systems and activities. This chapter summarizes 79 systems and provides an assessment of the types of fusion processing performed and their operational status. 22.2 Recent Survey of Data Fusion Activities A survey of DoD operational, prototype, and planned data fusion activities was performed in 1999–2000. The data fusion activities that were surveyed had disparate missions and provided a broad range of fusion capabilities. They represented all military services. The survey emphasized the level of fusion provided (according to the JDL model described in Chapter 2 of this book) and the capability to fuse different types of intelligence data. A summary of the survey results is provided here. In the survey, a data fusion system was considered to be more than a mathematical algorithm used to automatically achieve the levels of data fusion described in Chapter 2. In military applications, data fusion is frequently accomplished by a combination of the mathematical algorithms (or “fusion engines”) and Mary L. Nichols The Aerospace Corporation ©2001 CRC Press LLC 23 Data Fusion for Developing Predictive Diagnostics for Electromechanical Systems 23.1 Introduction Condition-Based Maintenance Motivation 23.2 Aspects of a CBM System 23.3 The Diagnosis Problem Feature-Level Fusion • Decision-Level Fusion • Model- Based Development 23.4 Multisensor Fusion Toolkit 23.5 Application Examples Mechanical Power Transmission • Fluid Systems • Electrochemical Systems 23.6 Concluding Remarks Acknowledgments References 23.1 Introduction Condition-based maintenance (CBM) is a philosophy of performing maintenance on a machine or system only when there is objective evidence of need or impending failure. By contrast, time-based or use-based maintenance involves performing periodic maintenance after specified periods of time or hours of operation. CBM has the potential to decrease life-cycle maintenance costs (by reducing unnecessary maintenance actions), increase operational readiness, and improve safety. Implementation of condition-based maintenance involves predictive diagnostics (i.e., diagnosing the current state or health of a machine and predicting time to failure based on an assumed model of anticipated use). CBM and predictive diagnostics depend on multisensor data — such as vibration, temperature, pressure, and presence of oil debris — which must be effectively fused to determine machinery health. Indeed, Hansen et al. suggested that predictive diagnostics involves many of the same functions and challenges demonstrated in more traditional Department of Defense (DoD) applications of data fusion (e.g., signal processing, pattern recognition, estimation, and automated reasoning). 1 This chapter demonstrates the potential for technology transfer from the study of CBM to DoD fusion applications. Carl S. Byington The Pennsylvania State University Amulya K. Garga The Pennsylvania State University . Applications 22 A Survey of Multisensor Data Fusion Systems Mary L. Nichols Introduction • Recent Survey of Data Fusion Activities • Assessment of System Capabilities 23 Data Fusion for Developing. development on multisensor data fusion has been performed for the Department of Defense (DoD). By the early 199 0s, an extensive set of fusion systems had been reported for a variety of applications. prototype, and planned data fusion activities was performed in 199 9–2000. The data fusion activities that were surveyed had disparate missions and provided a broad range of fusion capabilities.