The development of a tuning paradigm for flight simulator motion drive algorithms

This study investigated the tuning of flight simulator motion drive algorithms. The purpose of the study was to develop a tuning paradigm for the subjective pilot-in-the-loop tuning procedure which is currently used to tune flight simulator motion drive algorithms.; A tuning paradigm was developed using knowledge obtained from a linearized analysis of the University of Toronto Classical Washout Algorithm (UTCWA) and a washout filter tuning expert. This tuning paradigm optimizes the motion based on pilot evaluation of a set of tuning maneuvers. The resulting paradigm was captured in an expert system called PROTEST. PROTEST is a hybrid procedure/declarative expert system. The declarative aspect of PROTEST is rule-based with both classical and fuzzy logic rules. Knowledge, available to the expert by observation of the simulator motion during tuning, was provided to PROTEST by an on-line analysis of the simulator motion and motion algorithm states.; Two sets of experiments were run to test the abilities of PROTEST and hence the underlying tuning paradigm. In the first set of experiments four general aviation pilots tuned a set of maneuvers. The maneuvers were tuned independently to eliminate tuning interactions between the maneuvers. At the end of the tuning process, the pilots compared the resulting motion to a case previously tuned by an expert. In the second experiment, two commercial airline pilots tuned four tuning maneuvers in sequence, with the results from one maneuver forming the initial tuning set for the next maneuver. At the end of the tuning process the resulting motion was also compared to the same case previously tuned by an expert. The tuning decisions that were made by PROTEST were closely examined.; In both sets of experiments PROTEST exhibited expert tuning behaviour. The motion obtained using PROTEST was generally rated better than the case previously tuned by an expert. The author (the expert who developed PROTEST) generally agreed with PROTEST's tuning decisions. It was also discovered that PROTEST has limitations in its knowledge. These limitations require that PROTEST occasionally be redirected by the program operator.