Margery is a self-starter, capable of mentoring test teams on statistical test design processes and delivering designs and test strategy processes and products. Highly skilled at developing adapt methods and guidelines for specific program/project requirements, especially in instances where no closely related precedent exists.Highly skilled at the integration of scientific, statistical, cognitive science/machine learning classifiers, modeling methods and processes, to develop efficient test strategies that yield defensible results. Skilled at evaluating performance for new/developing systems. Ensuring that every system and process efficiently and effectively demonstrates performance against specific technical and operational requirements. Phone: 859-536-4741 Address: 6529 Greeley Ave, Dayton Ohio 45424
In this article, we report the methods and flexible frameworks employed to develop, integrate, an... more In this article, we report the methods and flexible frameworks employed to develop, integrate, and test adaptive Agent-Based Tactics, Techniques, and Procedures (AB-TTPs), in a complex training research environment. A Modeling and Simulation (M&S) environment developed for the Air Force Research Lab 711th/ HPW was used for the foundation for the Not-So-Grand-Challenge (NSGC); the use-case as applied. To do so, we capitalized on the properties of complex adaptive systems and situations. Allowing for context-based modeling and ultimately an agent's ability to independently asses, test and learn new tactics. These capabilities were accomplished, through agent and system use of modularization, decomposition and/or use of the combinatorial capabilities of the agent/s, the system, and or the situation's functional properties; i.e., affordances. The development and use of a Knowledge-to-Model (k2Mod) Environment Abstraction (EA) architecture gave agent's the capacity to gain situation awareness, recognize change in their environment and react and respond appropriately. In fact, the Adaptive Agent Intelligence (AI), i.e., models used were even able to accurately predict their own performance and tune their own parameters. This method also facilitates the speed by which new agent definitions, situation parameters, agent intelligence, and AB-TTPs can be developed and updated by " AI learning on the fly " ; pun intended. In addition, formalizing such a protocol affords the M&S community a process that promotes portability, usability, reusability, and composability for rapid agent-based modeling development and agent intelligence based research in complex environments. As suggested and as expected, this initial work has now proven to further allow for studies on how Adaptive AI-controlled dog-fighting agents with tunable decision making parameters can learn to optimize performance against an equally intelligent adversary. After Phase I and Phase IIa of the NSGC, Israelsen & Ahmed, (2016) and Israelsen, Ahmed, and Bennett (2017) studied the use of Gaussian Process Bayesian Optimization (GPBO) techniques developed to allow an agent to automatically learn global Gaussian Process (GP) surrogate models, which provided statistical performance predictions in both explored and unexplored areas of the parameter space. Allowing a learning engine to sample full-combat simulation parametric values that were most likely to optimize performance while providing highly informative data points. Ultimately improving future predictions. However, standard GPBO methods did not provide a reliable surrogate model for the highly volatile objective functions found in
Presented in this paper are details regarding a Complex Adaptive System-of-Systems, multi-purpose... more Presented in this paper are details regarding a Complex Adaptive System-of-Systems, multi-purpose, agent-in-the-loop behavior modeling framework. A framework meant to facilitate usage of agents in military training and research, that can either be rapidly developed or easily modified. This particular agent-based modeling effort was initiated because, to date, operator and opponent behavior models used in the current threat systems still require a significant amount of time, effort, and expertise to develop. The approach to such a versatile framework is inspired by the theories of emergent structure-functional properties often found in complex adaptive systems, such as the syntax and semantics inherit in a succession of events, artifact-mediated indirect communication, and facilitation of shared meaning within a multi-stakeholder environment. Also of interest are agent/environment interactions and the affordances available to an agent; a theory derived from ecological psychology. This research both supports and is supported by multi-agent simulations of military operations where the agents represent logical realistic models of military operators. A method for rapid design and evaluation of agent-based models will be identified and further discussed via descriptions of two exemplar implementations in distinct large scale complex adaptive multi-stakeholder multi-component systems.
To date, workload has been modeled rather successfully (Laughery, 2000), and situation awareness ... more To date, workload has been modeled rather successfully (Laughery, 2000), and situation awareness has even been modeled by Plott, Endsley, and Strater (2004). However, the interaction between workload and situation awareness has yet to be modeled through task networks with sufficient rigor to aid in the design of systems to support human performance. Because situation awareness supports effective decision making (Stanners and French, 2005), the upper and lower limits where the correlation between workload and situation awareness transitions from positive to negative and begins to negatively impact situation awareness, is sought. An overview of the problem space, ideas for modeling the optimal space between the upper and lower limits, and the interactions between cognitive workload and situation awareness is outlined.
Thank you for your interest in our work and your professional investment in ensuring we all benefit from a healthy level of ethical standards in our reporting and publication practices. We look forward to learning more about any advances you may make based your use of any portion, no matter how minute, of our collective body of work.
To that end, this work may be used in any publication, working paper, and or lecture. However a full citation, inclusive of all the author’s last names, must be placed in each paragraph where ideas you express are/were derived from our own. In addition, the full reference associated with each citation must appear in the document’s reference section.
We view any generation of new ideas that you may find success in that were derived from those ideas expressed within our own publications, as a positive feedback loop that can in turn enrich our future research and development efforts along with that of any of the respective communities of practice who may find inspiration within subsequent publications thereof.
ABSTRACT: To date it seemed that to simulate forces and or other entities in a virtual environment required also simulating the sensory, perception and cognition processes. A data gathering and parsing process that ranges from semi raw sensory information to perceptions, events, and actions to-be-taken in the world, but at the very least the perception, action, and decisions making cognitive/behavioral/ modality processes. To capture this type of information from the world some agents/models observe, orient, decide, and act (OODA) which is considered to be the needed to produce situation aware adaptive agents capable of operating in a Live Virtual Constructive, Distributed Mission Operations (LVC DMO) environment. However, to do so, as mentioned, this type of agent first must extract domain and situation knowledge/information from the environment about the state of affairs in the world in order to sense orient (observe - understand), decide to simulate an action (act in the world). Recently though, in an effort to not only improve military training simulation CGFs but to also establish a capability to rapidly develop models and agents for a multitude of uses to support the warfighter the Air Force Research Lab 711/HPW Warfighter Readiness Research Division RHAS, along with a few partners from industry have taken on and are proving success in use of a fresh approach toward this “Not-So-Grand-Challenge” (NSGC). The methodology entails the creation of a hierarchical knowledge-level structure to facilitate the speed by which new agent definitions can be developed or modified for use at multiple levels of abstraction. Formalizing the output of such an approach into a protocol or interface will afford the modeling and simulation (M&S) communities an agent development related architectural construct which promotes portability, usability, reusability, composeability, and marketability of establishing rapid development of LVC DMO training agents.
What does the Air Force stand to gain? Well, a few examples easily come to mind. The technology the NSGC project can bore might prove useful serving as an automated knowledge elicitation tool by way of post hoc mining of human-in-the-loop performance data looking for the nuances in human-system-syllabi-interactions during training, or as decision aid, or maybe even a scenario complexity scoring indexing tool for use in learning management systems. Envisioning that not only would such methods of abstraction and systems development prove to be fairly generalizable and portable to other domains of interest and inquiry but also be, more rapidly, readily and easily developed, integrated, tested, utilized and reutilized to support the warfighter. Championing new vistas in the “art-of-the-possible” the NSGC team decided to take a fresh look at these problems and decided to changed the situation up, in favor of adaptive, responsive, situated agent behaviors and architectures that utilize modular, reusable, composable, system components predicted to be more easily developed.
ABSTRACT: Stigmergy is a process by which local agents acting on the environment change the environment in such a way as to change another agent’s behavior. This form of indirect communication and local interactions between agents and the environment often leads to the self-organization of system structures. We present an experiment in which robot agents act independently to move objects (blocks) about in their environment, and inspect the outcome of these actions for clusters as an indication of increased organization. Results indicate that the agents did not act stigmergically, regardless of whether their behavior was fixed or random. However, the results indicate that the starting location of the blocks did have a significant effect on the final distributions. The effect initial conditions, types of emergence, and self organization have on the global product of swarm behavior are discussed and future research is proposed.
Decision under uncertainty, time constraints, and increasing complexity
›
Implications
›
Uncertainty
Drivers, types, and some causes
Decision under uncertainty
Trust in Automation
›
Reliance, reliability, and trust
›
Building trust in human-synthetic systems
Modeling the interaction between WL, SA-SU, DE
For the agents SA-SU of human performance
In-turn human trust of agent behavior
Architectures that include trust factors
Finite State Machine Induction
Summary
In the technology intensive environment we live in today, increasingly we place our lives in the hands of automation. While decision and automation aids are introduced to reduce human error and workload automation may actually be adding to workload and errors (Sheridan, 2002). In fact, many problems that revolve around automation can be linked to limitations in the fundamental cognition and perception capabilities of humans (Parasuraman & Riley, 1997). The benefits and risks of several types and levels of automation will be discussed. A way forward towards the mitigation of risks, through the use of adaptive levels of automation, will also be discussed.
Nonmonotonic Extrapolation
Margery J. Doyle
Institute of Cognitive Science
University of Louis... more Nonmonotonic Extrapolation
Margery J. Doyle
Institute of Cognitive Science
University of Louisiana at Lafayette
Abstract: Common everyday tasks like learning how to brake and accelerate in traffic require function learning where a continuous stimulus variable (e.g. speed) must be associated with another continuous variable (e.g. distance). A contrast exists between parametric and nonparametric models of function learning. Under a parametric theory, an nth-order polynomial is optimized during training. In contrast, nonparametric models rely on the association between individual input-output pairs. It is known thus far that nonparametric and mixture models lead to greater flexibly and have had fair success in modeling human performance. However, these models still extrapolate linearly. In light of these claims, L. Bott and E. Heit (2004) trained participants on a nonmonotonic cyclic cosine function and modeled their nonlinear extrapolation. In this talk, I present the results of an attempt to replicate the work of Bott and Heit. The replication failed to support the critical finding of nonlinear extrapolation.
In this article, we report the methods and flexible frameworks employed to develop, integrate, an... more In this article, we report the methods and flexible frameworks employed to develop, integrate, and test adaptive Agent-Based Tactics, Techniques, and Procedures (AB-TTPs), in a complex training research environment. A Modeling and Simulation (M&S) environment developed for the Air Force Research Lab 711th/ HPW was used for the foundation for the Not-So-Grand-Challenge (NSGC); the use-case as applied. To do so, we capitalized on the properties of complex adaptive systems and situations. Allowing for context-based modeling and ultimately an agent's ability to independently asses, test and learn new tactics. These capabilities were accomplished, through agent and system use of modularization, decomposition and/or use of the combinatorial capabilities of the agent/s, the system, and or the situation's functional properties; i.e., affordances. The development and use of a Knowledge-to-Model (k2Mod) Environment Abstraction (EA) architecture gave agent's the capacity to gain situation awareness, recognize change in their environment and react and respond appropriately. In fact, the Adaptive Agent Intelligence (AI), i.e., models used were even able to accurately predict their own performance and tune their own parameters. This method also facilitates the speed by which new agent definitions, situation parameters, agent intelligence, and AB-TTPs can be developed and updated by " AI learning on the fly " ; pun intended. In addition, formalizing such a protocol affords the M&S community a process that promotes portability, usability, reusability, and composability for rapid agent-based modeling development and agent intelligence based research in complex environments. As suggested and as expected, this initial work has now proven to further allow for studies on how Adaptive AI-controlled dog-fighting agents with tunable decision making parameters can learn to optimize performance against an equally intelligent adversary. After Phase I and Phase IIa of the NSGC, Israelsen & Ahmed, (2016) and Israelsen, Ahmed, and Bennett (2017) studied the use of Gaussian Process Bayesian Optimization (GPBO) techniques developed to allow an agent to automatically learn global Gaussian Process (GP) surrogate models, which provided statistical performance predictions in both explored and unexplored areas of the parameter space. Allowing a learning engine to sample full-combat simulation parametric values that were most likely to optimize performance while providing highly informative data points. Ultimately improving future predictions. However, standard GPBO methods did not provide a reliable surrogate model for the highly volatile objective functions found in
Presented in this paper are details regarding a Complex Adaptive System-of-Systems, multi-purpose... more Presented in this paper are details regarding a Complex Adaptive System-of-Systems, multi-purpose, agent-in-the-loop behavior modeling framework. A framework meant to facilitate usage of agents in military training and research, that can either be rapidly developed or easily modified. This particular agent-based modeling effort was initiated because, to date, operator and opponent behavior models used in the current threat systems still require a significant amount of time, effort, and expertise to develop. The approach to such a versatile framework is inspired by the theories of emergent structure-functional properties often found in complex adaptive systems, such as the syntax and semantics inherit in a succession of events, artifact-mediated indirect communication, and facilitation of shared meaning within a multi-stakeholder environment. Also of interest are agent/environment interactions and the affordances available to an agent; a theory derived from ecological psychology. This research both supports and is supported by multi-agent simulations of military operations where the agents represent logical realistic models of military operators. A method for rapid design and evaluation of agent-based models will be identified and further discussed via descriptions of two exemplar implementations in distinct large scale complex adaptive multi-stakeholder multi-component systems.
To date, workload has been modeled rather successfully (Laughery, 2000), and situation awareness ... more To date, workload has been modeled rather successfully (Laughery, 2000), and situation awareness has even been modeled by Plott, Endsley, and Strater (2004). However, the interaction between workload and situation awareness has yet to be modeled through task networks with sufficient rigor to aid in the design of systems to support human performance. Because situation awareness supports effective decision making (Stanners and French, 2005), the upper and lower limits where the correlation between workload and situation awareness transitions from positive to negative and begins to negatively impact situation awareness, is sought. An overview of the problem space, ideas for modeling the optimal space between the upper and lower limits, and the interactions between cognitive workload and situation awareness is outlined.
Thank you for your interest in our work and your professional investment in ensuring we all benefit from a healthy level of ethical standards in our reporting and publication practices. We look forward to learning more about any advances you may make based your use of any portion, no matter how minute, of our collective body of work.
To that end, this work may be used in any publication, working paper, and or lecture. However a full citation, inclusive of all the author’s last names, must be placed in each paragraph where ideas you express are/were derived from our own. In addition, the full reference associated with each citation must appear in the document’s reference section.
We view any generation of new ideas that you may find success in that were derived from those ideas expressed within our own publications, as a positive feedback loop that can in turn enrich our future research and development efforts along with that of any of the respective communities of practice who may find inspiration within subsequent publications thereof.
ABSTRACT: To date it seemed that to simulate forces and or other entities in a virtual environment required also simulating the sensory, perception and cognition processes. A data gathering and parsing process that ranges from semi raw sensory information to perceptions, events, and actions to-be-taken in the world, but at the very least the perception, action, and decisions making cognitive/behavioral/ modality processes. To capture this type of information from the world some agents/models observe, orient, decide, and act (OODA) which is considered to be the needed to produce situation aware adaptive agents capable of operating in a Live Virtual Constructive, Distributed Mission Operations (LVC DMO) environment. However, to do so, as mentioned, this type of agent first must extract domain and situation knowledge/information from the environment about the state of affairs in the world in order to sense orient (observe - understand), decide to simulate an action (act in the world). Recently though, in an effort to not only improve military training simulation CGFs but to also establish a capability to rapidly develop models and agents for a multitude of uses to support the warfighter the Air Force Research Lab 711/HPW Warfighter Readiness Research Division RHAS, along with a few partners from industry have taken on and are proving success in use of a fresh approach toward this “Not-So-Grand-Challenge” (NSGC). The methodology entails the creation of a hierarchical knowledge-level structure to facilitate the speed by which new agent definitions can be developed or modified for use at multiple levels of abstraction. Formalizing the output of such an approach into a protocol or interface will afford the modeling and simulation (M&S) communities an agent development related architectural construct which promotes portability, usability, reusability, composeability, and marketability of establishing rapid development of LVC DMO training agents.
What does the Air Force stand to gain? Well, a few examples easily come to mind. The technology the NSGC project can bore might prove useful serving as an automated knowledge elicitation tool by way of post hoc mining of human-in-the-loop performance data looking for the nuances in human-system-syllabi-interactions during training, or as decision aid, or maybe even a scenario complexity scoring indexing tool for use in learning management systems. Envisioning that not only would such methods of abstraction and systems development prove to be fairly generalizable and portable to other domains of interest and inquiry but also be, more rapidly, readily and easily developed, integrated, tested, utilized and reutilized to support the warfighter. Championing new vistas in the “art-of-the-possible” the NSGC team decided to take a fresh look at these problems and decided to changed the situation up, in favor of adaptive, responsive, situated agent behaviors and architectures that utilize modular, reusable, composable, system components predicted to be more easily developed.
ABSTRACT: Stigmergy is a process by which local agents acting on the environment change the environment in such a way as to change another agent’s behavior. This form of indirect communication and local interactions between agents and the environment often leads to the self-organization of system structures. We present an experiment in which robot agents act independently to move objects (blocks) about in their environment, and inspect the outcome of these actions for clusters as an indication of increased organization. Results indicate that the agents did not act stigmergically, regardless of whether their behavior was fixed or random. However, the results indicate that the starting location of the blocks did have a significant effect on the final distributions. The effect initial conditions, types of emergence, and self organization have on the global product of swarm behavior are discussed and future research is proposed.
Decision under uncertainty, time constraints, and increasing complexity
›
Implications
›
Uncertainty
Drivers, types, and some causes
Decision under uncertainty
Trust in Automation
›
Reliance, reliability, and trust
›
Building trust in human-synthetic systems
Modeling the interaction between WL, SA-SU, DE
For the agents SA-SU of human performance
In-turn human trust of agent behavior
Architectures that include trust factors
Finite State Machine Induction
Summary
In the technology intensive environment we live in today, increasingly we place our lives in the hands of automation. While decision and automation aids are introduced to reduce human error and workload automation may actually be adding to workload and errors (Sheridan, 2002). In fact, many problems that revolve around automation can be linked to limitations in the fundamental cognition and perception capabilities of humans (Parasuraman & Riley, 1997). The benefits and risks of several types and levels of automation will be discussed. A way forward towards the mitigation of risks, through the use of adaptive levels of automation, will also be discussed.
Nonmonotonic Extrapolation
Margery J. Doyle
Institute of Cognitive Science
University of Louis... more Nonmonotonic Extrapolation
Margery J. Doyle
Institute of Cognitive Science
University of Louisiana at Lafayette
Abstract: Common everyday tasks like learning how to brake and accelerate in traffic require function learning where a continuous stimulus variable (e.g. speed) must be associated with another continuous variable (e.g. distance). A contrast exists between parametric and nonparametric models of function learning. Under a parametric theory, an nth-order polynomial is optimized during training. In contrast, nonparametric models rely on the association between individual input-output pairs. It is known thus far that nonparametric and mixture models lead to greater flexibly and have had fair success in modeling human performance. However, these models still extrapolate linearly. In light of these claims, L. Bott and E. Heit (2004) trained participants on a nonmonotonic cyclic cosine function and modeled their nonlinear extrapolation. In this talk, I present the results of an attempt to replicate the work of Bott and Heit. The replication failed to support the critical finding of nonlinear extrapolation.
Uploads
Thank you for your interest in our work and your professional investment in ensuring we all benefit from a healthy level of ethical standards in our reporting and publication practices. We look forward to learning more about any advances you may make based your use of any portion, no matter how minute, of our collective body of work.
To that end, this work may be used in any publication, working paper, and or lecture. However a full citation, inclusive of all the author’s last names, must be placed in each paragraph where ideas you express are/were derived from our own. In addition, the full reference associated with each citation must appear in the document’s reference section.
We view any generation of new ideas that you may find success in that were derived from those ideas expressed within our own publications, as a positive feedback loop that can in turn enrich our future research and development efforts along with that of any of the respective communities of practice who may find inspiration within subsequent publications thereof.
ABSTRACT: To date it seemed that to simulate forces and or other entities in a virtual environment required also simulating the sensory, perception and cognition processes. A data gathering and parsing process that ranges from semi raw sensory information to perceptions, events, and actions to-be-taken in the world, but at the very least the perception, action, and decisions making cognitive/behavioral/ modality processes. To capture this type of information from the world some agents/models observe, orient, decide, and act (OODA) which is considered to be the needed to produce situation aware adaptive agents capable of operating in a Live Virtual Constructive, Distributed Mission Operations (LVC DMO) environment. However, to do so, as mentioned, this type of agent first must extract domain and situation knowledge/information from the environment about the state of affairs in the world in order to sense orient (observe - understand), decide to simulate an action (act in the world). Recently though, in an effort to not only improve military training simulation CGFs but to also establish a capability to rapidly develop models and agents for a multitude of uses to support the warfighter the Air Force Research Lab 711/HPW Warfighter Readiness Research Division RHAS, along with a few partners from industry have taken on and are proving success in use of a fresh approach toward this “Not-So-Grand-Challenge” (NSGC). The methodology entails the creation of a hierarchical knowledge-level structure to facilitate the speed by which new agent definitions can be developed or modified for use at multiple levels of abstraction. Formalizing the output of such an approach into a protocol or interface will afford the modeling and simulation (M&S) communities an agent development related architectural construct which promotes portability, usability, reusability, composeability, and marketability of establishing rapid development of LVC DMO training agents.
What does the Air Force stand to gain? Well, a few examples easily come to mind. The technology the NSGC project can bore might prove useful serving as an automated knowledge elicitation tool by way of post hoc mining of human-in-the-loop performance data looking for the nuances in human-system-syllabi-interactions during training, or as decision aid, or maybe even a scenario complexity scoring indexing tool for use in learning management systems. Envisioning that not only would such methods of abstraction and systems development prove to be fairly generalizable and portable to other domains of interest and inquiry but also be, more rapidly, readily and easily developed, integrated, tested, utilized and reutilized to support the warfighter. Championing new vistas in the “art-of-the-possible” the NSGC team decided to take a fresh look at these problems and decided to changed the situation up, in favor of adaptive, responsive, situated agent behaviors and architectures that utilize modular, reusable, composable, system components predicted to be more easily developed.
ABSTRACT: In order to properly simulate behavior based representations of real red and/or blue force operators through the use of models/agents acting as computer generated forces, it is necessary that the behavioral model receive information about the world including who exists, where they are, and relevant events. This includes not only what the actual hardware platform would receive or be capable of sensing, but also the information a human pilot could capitalize upon; information often obtained through communication channels or via the pilot’s own sensory, perception, knowledge, and/or decision making processes. For example, a pilot receives an air picture from the Air Battle Manager, which is derived via the ABM’s assessment of tactical information. In live and virtual exercises, any additional information a pilot might receive is limited to what the systems on his/her aircraft can sense; limitations not realized by a constructive model connected to the training network. Meaning, a behavior model of an aircraft with a maximum actual radar detection range of 15 miles would, without constraints representative of actual system capabilities, may detect entities outside of its normal range of capability; greatly diminishing external validity, realism, and believability. As such, we are developing a proof-of-concept for a Fighter Combat-Tactical Awareness Capability. This approach involves a model-to-Distributed Interactive Simulation-Environment Abstraction-Tactical Observer Agent to accommodate realistic information sharing with model/system function components. The EA–TOA translates low-level Distributed Interactive Simulation information into objects and then uses filters, based on actual maneuver recognition, hardware and human-in-the-loop oriented capability and limitation parameters, to create a Tactical Awareness Capability; limiting the information available to a model to the semantic/knowledge level of analysis. Herein, the genesis of FC-TAC, realizable through a multi-step data and EA process, syntactic and semantic validation, the application’s generalizability, and an actual proof-of-concept currently in use will be discussed.
Margery J. Doyle retains full Intellectual Property and Copyrights © 2014 for these works.
ABSTRACT: Stigmergy is a process by which local agents acting on the environment change the environment in such a way as to change another agent’s behavior. This form of indirect communication and local interactions between agents and the environment often leads to the self-organization of system structures. We present an experiment in which robot agents act independently to move objects (blocks) about in their environment, and inspect the outcome of these actions for clusters as an indication of increased organization. Results indicate that the agents did not act stigmergically, regardless of whether their behavior was fixed or random. However, the results indicate that the starting location of the blocks did have a significant effect on the final distributions. The effect initial conditions, types of emergence, and self organization have on the global product of swarm behavior are discussed and future research is proposed.
Overview
Decision under uncertainty, time constraints, and increasing complexity
›
Implications
›
Uncertainty
Drivers, types, and some causes
Decision under uncertainty
Trust in Automation
›
Reliance, reliability, and trust
›
Building trust in human-synthetic systems
Modeling the interaction between WL, SA-SU, DE
For the agents SA-SU of human performance
In-turn human trust of agent behavior
Architectures that include trust factors
Finite State Machine Induction
Summary
In the technology intensive environment we live in today, increasingly we place our lives in the hands of automation. While decision and automation aids are introduced to reduce human error and workload automation may actually be adding to workload and errors (Sheridan, 2002). In fact, many problems that revolve around automation can be linked to limitations in the fundamental cognition and perception capabilities of humans (Parasuraman & Riley, 1997). The benefits and risks of several types and levels of automation will be discussed. A way forward towards the mitigation of risks, through the use of adaptive levels of automation, will also be discussed.
Margery J. Doyle
Institute of Cognitive Science
University of Louisiana at Lafayette
Abstract: Common everyday tasks like learning how to brake and accelerate in traffic require function learning where a continuous stimulus variable (e.g. speed) must be associated with another continuous variable (e.g. distance). A contrast exists between parametric and nonparametric models of function learning. Under a parametric theory, an nth-order polynomial is optimized during training. In contrast, nonparametric models rely on the association between individual input-output pairs. It is known thus far that nonparametric and mixture models lead to greater flexibly and have had fair success in modeling human performance. However, these models still extrapolate linearly. In light of these claims, L. Bott and E. Heit (2004) trained participants on a nonmonotonic cyclic cosine function and modeled their nonlinear extrapolation. In this talk, I present the results of an attempt to replicate the work of Bott and Heit. The replication failed to support the critical finding of nonlinear extrapolation.
Thank you for your interest in our work and your professional investment in ensuring we all benefit from a healthy level of ethical standards in our reporting and publication practices. We look forward to learning more about any advances you may make based your use of any portion, no matter how minute, of our collective body of work.
To that end, this work may be used in any publication, working paper, and or lecture. However a full citation, inclusive of all the author’s last names, must be placed in each paragraph where ideas you express are/were derived from our own. In addition, the full reference associated with each citation must appear in the document’s reference section.
We view any generation of new ideas that you may find success in that were derived from those ideas expressed within our own publications, as a positive feedback loop that can in turn enrich our future research and development efforts along with that of any of the respective communities of practice who may find inspiration within subsequent publications thereof.
ABSTRACT: To date it seemed that to simulate forces and or other entities in a virtual environment required also simulating the sensory, perception and cognition processes. A data gathering and parsing process that ranges from semi raw sensory information to perceptions, events, and actions to-be-taken in the world, but at the very least the perception, action, and decisions making cognitive/behavioral/ modality processes. To capture this type of information from the world some agents/models observe, orient, decide, and act (OODA) which is considered to be the needed to produce situation aware adaptive agents capable of operating in a Live Virtual Constructive, Distributed Mission Operations (LVC DMO) environment. However, to do so, as mentioned, this type of agent first must extract domain and situation knowledge/information from the environment about the state of affairs in the world in order to sense orient (observe - understand), decide to simulate an action (act in the world). Recently though, in an effort to not only improve military training simulation CGFs but to also establish a capability to rapidly develop models and agents for a multitude of uses to support the warfighter the Air Force Research Lab 711/HPW Warfighter Readiness Research Division RHAS, along with a few partners from industry have taken on and are proving success in use of a fresh approach toward this “Not-So-Grand-Challenge” (NSGC). The methodology entails the creation of a hierarchical knowledge-level structure to facilitate the speed by which new agent definitions can be developed or modified for use at multiple levels of abstraction. Formalizing the output of such an approach into a protocol or interface will afford the modeling and simulation (M&S) communities an agent development related architectural construct which promotes portability, usability, reusability, composeability, and marketability of establishing rapid development of LVC DMO training agents.
What does the Air Force stand to gain? Well, a few examples easily come to mind. The technology the NSGC project can bore might prove useful serving as an automated knowledge elicitation tool by way of post hoc mining of human-in-the-loop performance data looking for the nuances in human-system-syllabi-interactions during training, or as decision aid, or maybe even a scenario complexity scoring indexing tool for use in learning management systems. Envisioning that not only would such methods of abstraction and systems development prove to be fairly generalizable and portable to other domains of interest and inquiry but also be, more rapidly, readily and easily developed, integrated, tested, utilized and reutilized to support the warfighter. Championing new vistas in the “art-of-the-possible” the NSGC team decided to take a fresh look at these problems and decided to changed the situation up, in favor of adaptive, responsive, situated agent behaviors and architectures that utilize modular, reusable, composable, system components predicted to be more easily developed.
ABSTRACT: In order to properly simulate behavior based representations of real red and/or blue force operators through the use of models/agents acting as computer generated forces, it is necessary that the behavioral model receive information about the world including who exists, where they are, and relevant events. This includes not only what the actual hardware platform would receive or be capable of sensing, but also the information a human pilot could capitalize upon; information often obtained through communication channels or via the pilot’s own sensory, perception, knowledge, and/or decision making processes. For example, a pilot receives an air picture from the Air Battle Manager, which is derived via the ABM’s assessment of tactical information. In live and virtual exercises, any additional information a pilot might receive is limited to what the systems on his/her aircraft can sense; limitations not realized by a constructive model connected to the training network. Meaning, a behavior model of an aircraft with a maximum actual radar detection range of 15 miles would, without constraints representative of actual system capabilities, may detect entities outside of its normal range of capability; greatly diminishing external validity, realism, and believability. As such, we are developing a proof-of-concept for a Fighter Combat-Tactical Awareness Capability. This approach involves a model-to-Distributed Interactive Simulation-Environment Abstraction-Tactical Observer Agent to accommodate realistic information sharing with model/system function components. The EA–TOA translates low-level Distributed Interactive Simulation information into objects and then uses filters, based on actual maneuver recognition, hardware and human-in-the-loop oriented capability and limitation parameters, to create a Tactical Awareness Capability; limiting the information available to a model to the semantic/knowledge level of analysis. Herein, the genesis of FC-TAC, realizable through a multi-step data and EA process, syntactic and semantic validation, the application’s generalizability, and an actual proof-of-concept currently in use will be discussed.
Margery J. Doyle retains full Intellectual Property and Copyrights © 2014 for these works.
ABSTRACT: Stigmergy is a process by which local agents acting on the environment change the environment in such a way as to change another agent’s behavior. This form of indirect communication and local interactions between agents and the environment often leads to the self-organization of system structures. We present an experiment in which robot agents act independently to move objects (blocks) about in their environment, and inspect the outcome of these actions for clusters as an indication of increased organization. Results indicate that the agents did not act stigmergically, regardless of whether their behavior was fixed or random. However, the results indicate that the starting location of the blocks did have a significant effect on the final distributions. The effect initial conditions, types of emergence, and self organization have on the global product of swarm behavior are discussed and future research is proposed.
Overview
Decision under uncertainty, time constraints, and increasing complexity
›
Implications
›
Uncertainty
Drivers, types, and some causes
Decision under uncertainty
Trust in Automation
›
Reliance, reliability, and trust
›
Building trust in human-synthetic systems
Modeling the interaction between WL, SA-SU, DE
For the agents SA-SU of human performance
In-turn human trust of agent behavior
Architectures that include trust factors
Finite State Machine Induction
Summary
In the technology intensive environment we live in today, increasingly we place our lives in the hands of automation. While decision and automation aids are introduced to reduce human error and workload automation may actually be adding to workload and errors (Sheridan, 2002). In fact, many problems that revolve around automation can be linked to limitations in the fundamental cognition and perception capabilities of humans (Parasuraman & Riley, 1997). The benefits and risks of several types and levels of automation will be discussed. A way forward towards the mitigation of risks, through the use of adaptive levels of automation, will also be discussed.
Margery J. Doyle
Institute of Cognitive Science
University of Louisiana at Lafayette
Abstract: Common everyday tasks like learning how to brake and accelerate in traffic require function learning where a continuous stimulus variable (e.g. speed) must be associated with another continuous variable (e.g. distance). A contrast exists between parametric and nonparametric models of function learning. Under a parametric theory, an nth-order polynomial is optimized during training. In contrast, nonparametric models rely on the association between individual input-output pairs. It is known thus far that nonparametric and mixture models lead to greater flexibly and have had fair success in modeling human performance. However, these models still extrapolate linearly. In light of these claims, L. Bott and E. Heit (2004) trained participants on a nonmonotonic cyclic cosine function and modeled their nonlinear extrapolation. In this talk, I present the results of an attempt to replicate the work of Bott and Heit. The replication failed to support the critical finding of nonlinear extrapolation.