Improving Constructive Simulation Interfaces

Military simulation applications are used in command and staff training, wargaming, analysis and experimentation. Constructive simulations, such as One Semi-Automated Forces (OneSAF), provide a flexible environment for building and simulating exercises, however, when building these new software applications, human factors engineering is often overlooked in the design process. At Spring 2006 SIW, Cybernet Systems Corporation presented preliminary results from a human factors analysis of OneSAF Testbed Baseline (OTB) to identify key weaknesses in its interface, and identified several systematic problems. Using GOMS-KLM to model human behavior while interacting with OneSAF, we found several interaction areas ripe for improvement including: excessive mouse movements, precise mouse movements, excessive visual search, finding objects in multiple data views at the same time, and repetitive, successive menu selections. Other problems exist, but these general design flaws are important in that they highlight areas of major performance loss and cognitive loading. With the recent distribution of OOS, we have undertaken an updated analysis of the new suite of constructive simulation applications. In order to evaluate the user performance within OOS and JSAF, we discovered the need for tools to quickly create and instrument sample interfaces for analysis. This paper discusses the tools and methods we developed in order to extend our previous work(1) to apply to general-purpose web applications. In discussing our test-harness framework, we present recommendations for user interaction and event data that must be captured, quantitative analysis tools that correlate interface arrangements with performance measures, and validation techniques of existing and evolving models of human performance. We also discuss our mixed-method analysis approach that uses theory- derived cognitive modeling techniques to conduct a formative analysis of interface alternatives and the validation of that formative analysis with empirical data.