Comment on "Controllability Limit of a Human Pilot"

T experiment discussed in the recent paper by Washizu and Miyajima of the University of Tokyo is quite similar to one performed in 1960 by this writer at the Aeronautical Systems Division, Wright-Patterson Air Force Base, Ohio, and reported in Ref. 2. The purpose of this commentary is to present a critical comparison of the results given in Refs. 1 and 2. First, a brief discussion of the two experiments is necessary. Both of the experiments investigated the limits of manual control of unstable, second-order dynamic systems. Reference 2 considered only those systems having an oscillatory transient response, whereas Ref. 1 also considered those having one divergent mode as well. Compensatory displays were used in both of the experiments. That is, the display showed only the system error in both of the cases. Schematically, both of the systems were of the single loop form, as shown in Fig. 1, although Ref. 2 presents evidence that higher forms of human operator adaptation are possible which invalidate this conceptual model. The experiments differed in that those of Ref. 1 required lateral manipulator motions, whereas those of Ref. 2 were longitudinal. It is probable that this would not produce large differences between the two sets of experimental results, since, in either case, the biophysical systems involved are quite similar in detail. Grossly different testing methodologies were employed, however. The experiments reported in Ref. 1 used only an initial transient system input and required that the operator maintain stable control for 20 sec. Substantially no practice was permitted. Only three trial runs were made for each combination of controlled element parameters tested. On the other hand, the experiments reported in Ref. 2 used a random-appearing system input and required that stable control be maintained for 2 min. Controlled element dynamics were varied incrementally from the easier to the more difficult control problems. The subject was permitted as much practice as required. Nearly 900 trial runs were made in establishing two controlled elements believed to represent limits of manual control. The experiment of Ref. 2 was devised in order to examine the validity of a speculative, theoretical prediction of the controllability limits given in Ref. 3. The predicted limits of Ref. 3, together with the experimental limits and revised predictions from Ref. 2, are shown in Fig. 2. Typical experimental results given in Ref. 1 are also shown. Only the oscillatory controlled element is considered. Obviously the limits of control as given by Refs. 1 and 2 disagree widely, particularly at higher values of controlled element natural frequency (stiffness). In the opinion of this