Derivation of control activity metrics for the rule-based prediction of helicopter pilot workload

Abstract

Control activity is a recognised gauge of pilot workload and recent research has employed wavelet decomposition to classify discrete control actions into categories such as guidance and stabilisation. The aim of the present work is to extend the wavelet approach so that workload may be quantified through sets of rules based on appropriate control activity metrics. The rules are derived from data collected in piloted simulation trials of a variety of flying tasks involving a number of pilots and different helicopter configurations. Statistical tests are then applied which test the efficacy of the derived rules. The immediate aim of the research is to establish whether workload can be successfully predicted from control responses. The underlying goal however, is to be able to predict workload ratings from desktop simulations in order to provide indicative workload information at the design stage. The contribution of the current study to this objective is discussed.

Quantification and Prediction of Pilot Workload in the Helicopter/Ship Dynamic Interface

“…For different pilots, even of similar capability, it is clear that their variability means that individual rating cannot be predicted with consistent accuracy but trends and approximate values can be relied upon provided a tolerance is associated with the results. MacDonald [3] achieved rules acceptable for several pilots by trading off accuracy against generality. It is unlikely that this situation can be improved without smarter rules of some kind.…”

“…In the work above, it has been assumed that the task has been carried out at least adequately. Some variation in helicopter configuration has been accommodated in the work of MacDonald [3]. It is reasonable to assume that this would not extend to more significant changes in configurationbut this is probably acceptable since most competing investigations would be for tuning of changes in control system parameters and authority.…”

“…This is the case for the piloted trials which produced the database described in section 3. In several piloted trials, the Defence Evaluation and Research Agency (DERA) WR questionnaire [3] was employed to allocate WRs which were mapped to Cooper -Harper HQRs in a systematic manner. In many applications, workload and performance are not the only factors affecting the situation; HQRs strictly apply only to a particular task and set of environmental conditions, but in practice their use extends beyond this narrow definition.…”

“…Macdonald and Bradley [13,14] adopted a similar analysis method in an attempt to make such charts give quantitative information by deriving rules which relate the distribution of smooth-ramp events (Fig. 2) in control responses to the WR.…”

“…The detailed rationale for these values and a full description of the attack chart regions may be found in Macdonald [3]. The metric derived from the attack chart is a vector fx 1 , x 2 , x 3 , x 4 g, where x 1 is the number of events in region 1, etc.…”