1 ONR Principle Investigators: Dr. Joe DiVita, Code Dr Glenn Osga, Code Dr. David Kieras - University of Michigan Dr. Tom Santoro - NSMRL Groton CT Mr. Rob Morris, Code Contributors: Dr. Hung T. Nguyen- New Mexico State University Modeling of Human-Computer Interaction: Application to Command & Control Presented at Systems Design Technical Group Meeting HFS Annual Conference, Denver Colorado, Oct. 2003

2 Based upon Stepwise models as defined in: Psychology of Human-Computer Interaction, Card, Moran, and Newell (1983). Goals: What Must be Accomplished Operators: Elementary Perceptual, Motor, or Cognitive Acts. Methods: Step by Step Procedure for a Goal Selection Rules: Basis for Choosing Methods GOMS Components

3 GLEAN: GOMS Language Evaluation and Analysis Tool Simulated Interaction Devices Auditory Input Declarative and Procedural Knowledge in Long Term memory Visual Input Cognitive Processor GOMS Language Interpreter Working Memory Auditory Processor Visual Processor Vocal Motor Processor Manual Motor Processor Task Environment Model-based Evaluation David Kieras University of Michigan to appear in J. Jacko & A. Sears (Eds), Human-Computer Interaction Handbook, Lawrence Erlbaum Associates, in press

4 Define the Goals: How are they accomplished ? How might they be accomplished? What are the alternatives? 2. Write the Methods in GOMSL, 3. Build the HCI and Task Environment in C++, 4. Run the Scenario(s) & Review Results. 1. Task Analysis Using Models for Design Trade-Off Studies Analysis Procedure:

5 Display Design Components

6 Task Manager Task Queue Systems Status CommunicationsCommunications Task Manager & Status Display

7 Strike Plan Overview Task Progress Process Visualization - Tactical Tomahawk

8 Avg Time Avg Time Method for Goal Frequency Frequency SPOKEN SYNTHETIC Respond_to New_Air Update_Air Trk Review Air_ID Review Track_profile Conduct Threat_Assessment Request Escort Request Visual ID Issue Query Issue Warning New_Track_Verbal_Rpt Update_Track_Verbal_Rpt Hook Track Comparison Results

9 AWC operator selects task Task arrives on TM display AWC operator sends report Text to Speech completed AWC receives acknowledgment Mean Waiting time in the queue Team 1 data = 51.2 s GOMSL Model = 55.2s Mean Service Time Team 1 data = 6.7 s GOMSL Model = 10.3 s Mean time = 10.1 s GOMSL Model = 5.5 s GOMSL Model = GOMSL Model = 9.43GOMSL Model = 9.10 Actual Team Results GOMSL Model Results GOMSL Model with fast working memory Example: New Track Report Task Flow

10 AWC Operator IQC1 Operator AIC Operator Tasks performed - Output Flow Tasks performed - Output flow Network Queueing Model of Team 1 Task Flow. Level I* & II*, ordered to send. VID Level I & II’s Tasks Entering: New track Report Update track Report Level 1Query Level II Warn VID Cover Engage Illuminate

11 AWC IQC1 AIC  11 22 33  1 incoming tasks General Open Network Queueing Model Tasks passed between operators  2 incoming tasks  3 incoming tasks Load to each node: i=i= i ii i =  i + j p ji i = the effective arrival rate to node i. p ji = probability that a task, after receiving service by node j, proceeds to node i. Ave #, N, of tasks in the whole system:  i - N= i /( i )  Ave time, T, of tasks in the system:  i - 1 T= i /( i )   ii Network Stats:  i = service time of task ñ) = i)i)  i ni  (1- P( Probability of a particular state (n 1, n 2, n 3 ) tasks: ii = rate of incoming tasks

12 Team 2 takes 36% longer to complete New Track Reports

13 Modeling Plans Expand Models Basic models constructed for Air Defense mission and Land Attack with Tomahawk. Models based upon future HCI designs being incorporated into Tomahawk. Expand individual models into tactical team models. Design Feedback Provide design feedback on best features to improve performance. Team Design Compare team work allocation, flow, process with various team configurations for future systems.