Analysis Techniques For Man-Machine Systems Design

NATO UNCLASSIFIEDAC/243(Panel 8)TR/7 - 88 -Volume 2on the measurement of subjective workload to verify that system performance requirements will be met. As noted inVolume 1, detailed considerations of the use of prototyping, simulation and field trials are outside the scope of thisreview. The links between analytical techniques and other approaches are discussed where possible.human tasks & task sequences,pacing times &events. & performance requirements4 -task tasks associate with high/ow tmes or erormanceanalysisredictinpf noperatoreoper \ act ignsequene \ acins \ tms/ features/Of tsks \ dictated by byh/diaeytaskF inputs & R a the between operato macoutputs, timesd o inte rface ad wo pormace d ifrequencies,\\ //performancerequirements6 -A variety of approaches are -available -to predict operator performance analytically (Meister. 1985). Most of them usethe concept of operator "workload" rather than a measure of performance itself (Hart & Wickens, 1990). Someestablished techniques such as the time Line analysis of workload (5.1) are normative, based on a single task analysissequence. Network simulations of operator tasks, such as PERT. SAINT and MicroSAINT (5.2) can be madestochastic so that the external mission events, and the task times and sequences, are drawn from a distribution oftimes and probabilities. Network simulations can link workload to system performance because they can produceestimates of task success in terms of percentage of tasks completed. or time to complete, using models of humanperformance, as well as producing estimates of operator workload (Chubb, Laughery & Pritsker, 1987). NATOAC/243 Panel 8/RSG.9 has reviewed and reported on those techniques and on a variety of models of humanperformance (McMillan et al., 1989, 1991). One specific approach, SIMWAM (5.3) is included here, because of theextensive use made of it for weapon systems development.Performance predictions can also be made using subjective workload measurement techniques (5.4). For example,although the SWAT technique (5.5) was developed for the evaluation of an existing, or simulated, system, it can beapplied pomjecavely (Pa-SWAT) to the prediction of operator workload based on a task analysis. The NASA TaskLoad Index (TLX) (5.6) can be applied projectively also.In following either a simulation modelling approach or a subjective workload approach to predicting operatorperformance, it must be remembered that the relationship between workload and performance is not straightforward.As task demands vary with time, the operators may increase or decrease their effort to compensate, or their effort maydecline due to fatigue (Fig. 5.2). Thus at the outset of a mission, the task demand may lie within an operator'sNATO UNCLASSIFIED- 88 -