Air Midas


Published on

Paper I prepared for one of my classes at ERAU WWOnline

  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Air Midas

  1. 1. Running Head: THE DEFINING CHARACTERS OF AIR MIDAS: AIR MAN-MACHINE INTEGRATION DESIGN AND ANALYSIS SYSTEM The Defining Characters of Air MIDAS: Air Man-Machine Integration Design and Analysis System Mersie A. Melke Embry-Riddle Aeronautical University Daytona Beach, Florida Department of Distance Learning Instructor: Susan Bailey-Schmidt January 29, 2015 Defining Characters Air MIDAS 1
  2. 2. Abstract Humans have been using tools to enhance their outputs. This need to do better than the past drove humans to the use of more complex machinery capable of achieving things impossible to humans alone. However, as machines became more powerful and helpful to humans, the relationship between the two entities began to have an agenda of its own. Consequently, studying about man-machine relations is inevitable in current times. One tool that helps in doing such studies is the Air MIDAS software. This paper shows the defining characters, the requirements and possible outputs of this software. In addition, this paper addresses the background on the topic of man-machine performance modeling as related to the Air MIDAS software. Defining Characters Air MIDAS 2
  3. 3. Introduction Systems that function because of inputs from both humans and machines are results of contemporary times. Both humans and machines bring a set of characters and behaviors in the way they react, both to each other and to the work environment. Therefore, in addition to having a comprehensive knowledge of the technical side of machines, having an understanding of humans and their reaction to machine augmented systems has become an issue that needs addressing. This is because enhancing the human-machine relationship is one of the ways for optimizing output of a prospective system or a system with previous experience, besides developing machines with capacity superior than their predecessors. In studying the relationship between man and machine mentioned above, two methods are of relevant value until now. One method involves humans as part of an experiment while evaluating the characteristic of both the machine and the humans in the system. This method is termed as the Human In The Loop (HITL) simulation. HITL simulations calls for integration of independent elements (technologies, service providers and service receivers), the logistics of whose coordination places a complexity and expense burden on the simulation (Corker, Gore, 2001). Another type of simulation is the Human Out Of The Loop (HOOTL) type of simulation. HOOTL simulation is one that uses computer models of human performance as the human agent interacts with new and/or simulated technologies and procedures. These simulation tools are computer based simulation processes where human characteristics taken from years of research from respective fields, embedded within a computer-generated representation of humans’ operating environment, create a mockup of the real world. HOOTL simulation is an alternative methodology to the expensive HITL simulations. One reason is the ability to use HOOTL simulations at an earlier process in the development of a product, system or technology (Corker, Gore, 2001). Defining Characters Air MIDAS 3
  4. 4. Both methods are collectively termed as human performance modeling techniques. Development of these methods is with the preconceived idea of deriving or anticipating possible outcomes from a machine or machine augmented system that involves humans. As a result, with having such kind of information, designers would be able to enhance their prospective ideas to develop an optimal design. When thoughts of upgrading already functioning systems arises, HITL and HOOTL methods can be employed to derive needed information about the human-machine interface. Background of Human Performance Modeling Techniques Human performance modeling debuted over 50 years ago with quasi-linear and manual control models. Human performance modeling in these times related to modeling human tracking behavior in a closed-loop person-machine system (Gore, 1999). The reason for the term quasi-linear stems from an engineer’s assumption of a linear transfer function model for the operator’s control behavior in perceiving an error and translating this error to a response (Gore, 1999). In the aviation environment, human performance modeling initiated in the 1950s (Foyle, Hooey, 2008). A person by the name Duane McRuer was interested in aircraft handling qualities. Consequently, McRuer pioneered recasting the dynamics of flight traditionally expressed in partial differential equations into control engineering transfer function terms (Foyle, Hooey, 2008). The former is a relationship between several flight variables that are interdependent on one another. Consequently, an outcome of a system comprised of these variables will consider the changes of the variables in terms of one another. The latter entails a workflow kind of relationship that has a certain set of input, a certain set of analysis and a corresponding out put. However, McRuer came to a challenge of representing human pilot in the system he tried to develop. As a result, McRuer set about to explore the control engineering representation for what came to be called “manual control”— Defining Characters Air MIDAS 4
  5. 5. a model of the dynamical response of the human controller (Foyle, Hooey, 2008). Manual control was the dominant research drive to human performance modeling in aviation. Eventually, cognitive architectures that focused on the higher level of human behavior in a specific environment took over. Basis for Air MIDAS Air Man-Machine Integration Design and Analysis System (Air MIDAS) is one of the different HOOTL tools employed to formulate an analysis of humans in a machine- augmented systems. It is a joint development effort between the San Jose State University and National Aeronautics Space Administration (NASA) Ames research center. Initially in 1984, the United States Army and NASA Ames research center had initiated a project to come up with a modeling software that could analyze the characteristics of humans in a work environment, the outcome of which was the Core MIDAS software (Gore, 1999). The development effort of Core MIDAS software focused on the individual operator in a psychological theory framework (Gore, 1999). Beginning in 1993, a parallel development effort that led to the Air MIDAS software started. This focused on the team aspect of human behavior in the framework of psychological theory (Gore, 1999). Development of Air MIDAS was by NASA Ames research center and San Jose State University, primarily for aviation related applications such as the examination of procedural rule set changes on critical event recovery in the National Air Space (Gore, Corker, 2001). The concept of human modeling in Air MIDAS bases on a psychological concept referred to as First Principles. First Principles model of human performance bases on the mechanisms that underlies and cause human behavior (Laughery, Corker, 1997). First Principles model integrate human perceptual (data acquisition), cognitive (thought process) and motor (physical manifestations) systems thus incorporating the high-level behaviors that are characteristic of human performance. First Principles model also allow the human model Defining Characters Air MIDAS 5
  6. 6. to learn about the system and recall about the system by integrating the theoretical and practical models of human behavior in to the human model (AGARD, 1998). Consequently, First Principles model of human performance provides models of emergent human behavior based on elementary models of human behaviors such as perception, attention, working memory, long-term memory and decision-making (Gore, Corker, 1999). Defining Characters of Air MIDAS Air MIDAS runs on a personal computer (PC) using the Windows 2000 or XP operating system and requires Visual C++ 6.0, Alegro CL, and JAVA software (Foyle, Hooey, 2008). The code for the software written in the LISP, C and C++ programming languages, historically had 350,000 lines of code (Pew, Mavor, 1998). The main components of the First Principles model, the building block of Air MIDAS, comprise the simulated representation of the real world within which the agent modeled by Air MIDAS exists. A symbolic operator model (SOM) that represents the perceptual and cognitive activities of the agent exists. Another element of the SOM is the Updateable World Representation (UWR). The world representation information (environment, crew station, vehicle, aerodynamic constraints and terrain database) passes through the perceptual and attention process of the SOM to the UWR (Gore, Corker, 2001). In addition, a direct feed to the UWR from the simulated world representation is also possible. The UWR represents the agent’s working memory, domain knowledge and activity structure of the tasks to be completed (Gore, Corker, 2001). An example of working memory would be a short-term recollection of a previous work out put that is necessary to continue ones job to a next level. In addition, domain knowledge assumes that the simulated human has previous experience with the machine, and thus is applicable for a human-machine system that has matured. This UWR passes information to a scheduler within the SOM that Defining Characters Air MIDAS 6
  7. 7. determines the resources available for the completion of the activity. The environment triggers the activities (procedure) within the agent and completes the desired procedure in accordance with the availability of the resources in the agent. The scheduler invokes rules to determine the triggering procedures (Gore, Corker, 2001). Consequently, postponed, suspended, working, current or pending procedures fall in the workflow simulation. In turn, the SOM selects activities to perform, some of which interact with the representation of equipment in the simulated world and change the behavior of the relevant part of the system (Gore, Corker, 2001). After the appropriate scheduling and assigning of work takes place within the SOM, the actual action execution by the motor activity representation of the software takes place. This representation feeds forward necessary functions based on the resource available per the instructions given by the scheduler. It also feeds back actual happening to the attention and perception representation of the SOM and the planning representation of the same module. By so doing, it simulates the work procedure and the entailed workload and time taken to do multiple or single task(s) by an operator or a team of operators. From the above description, it is evident that the Air MIDAS simulation software is a closed loop object oriented software. In order for it to function, the software has to initiate an iterative process that will recognize the working environment, the available resources and the instantaneous work status. All these benchmarks, collectively termed as objects of recognition, guide the software to come to a result. In addition, the fact that process flows follow an iterative path from perception to scheduling to motor activities and back to the perception representation as feedback, testify that the system is a closed loop one. In his Master’s Thesis paper on the subject of distributed cognitive process, Brian F. Gore (1999) describes the workflow of man-machine simulation on MIDAS as follows, Defining Characters Air MIDAS 7
  8. 8. “The model designer that intends to learn about human-machine relations enters the system through the Graphical User Interface (GUI) that provides the main interaction between the designer and the system. The user selects among four functions in the system. To start working on Air MIDAS requires the user to establish a domain model. The user can then select the graphical animation or view to support one or more simulations. The user is able to specify in the simulation module the execution and display parameters for a given simulation set, and specify the data to-be-collected and analyzed in the results analysis because of running the simulations. The results analysis system also provides functionality for archival processes for various simulation sessions. The other underlying MIDAS architectural components include a human operator model, memory representation models, attentional control models, activity representation models, task activity models, and decision-making models. These seven models interact to produce human behavior that is based on environmental influences”. Gore (1999) continues the description on the workflow of the simulation process as follows, “ The domain model consists of a library of descriptors that support the creation of vehicle characteristics, environment characteristics, crew station/equipment characteristics, the human operator model (HO), mission and activity models, memory representation, attention control, task agenda and decision making models. HO allows for the production of behavior and responses for single and multiple operators in the scenarios. The HO is composed of sub-models in an integrated format including an anthropometric model, sensation and perception models, attention(and other resource models), central processing cognitive functions such as decision- making, evaluation and action selection, and finally behavioral models to guide the Defining Characters Air MIDAS 8
  9. 9. anthropometric model in the execution of action. The second model that is vital for MIDAS is the goal related model. This goal related model makes the MIDAS architecture First Principled because this is where the goals are stored within the operator”. Another trait of the Air MIDAS software is that it is event driven. The fact that Air MIDAS feeds forward actions or procedures for execution and feeds back status of the proceedings of the action makes it an event- driven rather than situation-driven simulation software (Pew, Mavor, 1998). The Air MIDAS software like any HOOTL tool runs in one of two ways when being used to analyze a man machine relationship. The first one includes a process in which variables that are pertinent to the man-machine relation established manipulate, on a predefined scenario. This involves changing a specific variable at a predefined time in the simulations. One simulation is termed as a single Monte Carlo simulation. Once these variables change, a second series of Monte Carlo simulation runs take place. Consequently, comparison of the results happens in order to verify and validate the credibility of the modeling tool, which in this case is Air MIDAS (Gore, Corker, 1999). A second procedure entails an experimental design, and has the human performance- modeling tool generate responses given the manipulations outlined in the experimental design. The modeling software selects from a common random number (CRN) table and uses this human performance value in the Monte Carlo runs (Gore, Corker, 1999). These runs are performed a number of times similar to testing a number of subjects in a HITL simulation and deducing conclusions based on the output from the subjects. The data from CRN matrix that the human performance-modeling tool uses in its generation of response values comes from human performance and human performance cognition research (Gore, Corker, 1999). The Defining Characters Air MIDAS 9
  10. 10. human performance modeling software selects from the CRN table built within each modeling tool and then uses these values in the Monte Carlo simulation run. Both the above procedures of running Air MIDAS have one goal. This is anticipating the amount of time covered to do a single or a combination of tasks. Coherent with this extraction of the human model workload occurs. As this software can anticipate both single and multiple human operator workloads, it is possible to analyze crew workload management. Thus, Air MIDAS can analyze induced changes in an existing crew operated system or future crew operated systems with adequate real world representation in terms of workload and usage time. Conclusion In this paper, Air MIDAS, which is a HOOTL simulation software has been characterized based on its defining features. One feature that differentiates this particular software from others that do the same job is the representation of the high-level behaviors that are characteristic to humans. This feature makes the evaluation carried out by Air MIDAS much more stringent. Consequently, gathering detailed results from such evaluations is inevitable. However, since the representation of detailed human character is involved in the software, it is necessary to have a database upon which the software bases such characters. This is because lack of such information will compromise the simulation process of the software and flaw the results. Information about human character like perception is gathered in ways that include HITL simulations, work place observations etc. Thus, the optimal operation of Air MIDAS is dependant on human character data. Another observation made is on the path of the workflow simulated by the software. A single accomplishment loop will involve perception, UWR, motor simulation, scheduler and external world database. In order to run this loop of activities an efficient processor is Defining Characters Air MIDAS 10
  11. 11. necessary that would guarantee results. Therefore, the use of Air MIDAS and its advancement in the future ties with the future of data processing that is the availability of powerful computers will enhance the capabilities of the Air MIDAS software. . Defining Characters Air MIDAS 11
  12. 12. References Advisory Group for Aerospace Research and Development (AGARD) (1998). A designer’s guide to human performance modeling (North Atlantic Treaty Organization AGARD Advisory Report 356). Canada Communication Group Inc.: Hull, Canada. Foyle, D.C. and Hooey, B.L. (2008). Human Performance Modeling in Aviation. Boca Raton, FL.: CRC Press. Gore, B.F. (1999). Modeling distributed cognition: System interaction in free flight. Master’s thesis: UMI, Santa Monica, CA. Gore, B.F. and Corker, K.M. (1999). System interaction in “free flight”: A modeling tool cross comparison. Proceedings of the Digital Human Modeling Conference and Exposition, Warendale, PA. Gore, B.F. and Corker, K.M. (2001). Human Performance Modeling: A Cooperative and Necessary Methodology for Studying Occupational Safety. Advances in Occupational Ergonomics and Safety. Amsterdam: IOS Press Laughery, K.R. Jr., and Corker, K.R. (1997). Computer modeling and simulation of human/system performance. Handbook of human factors and ergonomics (Second Edition.), N.Y.: Wiley and Sons, Inc., (pp. 1375-1408). Pew, R.W. and Mavor, A.S. (1998). Modeling Human and Organizational Behavior: Application to Military Simulations. Washington D.C.: National Academy Press. . . Defining Characters Air MIDAS 12