Development of Automated Performance Measures for Introductory Air Combat Maneuvers

Abstract

An exploratory study was conducted to develop automated performance measurement (APM) techniques for introductory air combat maneuvers. The method and results of the study can be summarized as follows: (1) The study used Northrop's LAS/WAVS air combat simulator, mechanized to simulate the F4J. (2) Seven F4J student pilots and six F4J instructor pilots each flew 16 barrel roll attacks against an autopilot-controlled bogey, for a total of 208 simulator runs. (3) A total of 552 objective performance measures and an average of 35 subjective performance measures were obtained on each run. (4) Using the simulator's replay capability, 64 of the original 208 runs were evaluated independently by three different instructors to obtain estimates of inter-observer reliability. Sixteen of these 64 runs were evaluated a second time by these same three instructors to obtain estimates of intra-observer reliability. (5) A master APM measures tape was constructed, which contains all of the subjective and objective measures obtained for each run in a format that permits statistical analyses to be conveniently performed on any desired subset of subjects, evaluators, or performance variables. (6) Several different univariate and multivariate statistical analyses were performed on selected subsets of the data, using the BMDP programs developed by UCLA. The objectives of this task were: a) to identify a preliminary set of automated measures which could be used to augment instructor evaluation of performance in introductory air-to-air tactics; and b) to assess the relative effectiveness of alternative statistical techniques for the measure identification process.