Prediction of helicopter simulator sickness

Machine learning methods from artificial intelligence are used to identify information in sampled accelerometer signals and associative behavioral patterns which correlate pilot simulator sickness with helicopter simulator dynamics. In this work, accelerometers were installed in the simulator cab, enabling a complete record of the flight dynamics and the pilot's control response as a function of time. When given the results of performance measures administered to detect simulator sickness symptoms, the problem was to find functions of the recorded data which could be used to help predict the simulator sickness level and susceptibility. Methods based upon inductive inference were used, which yield decision trees whose leaves indicate the degree of simulator-induced sickness. The long-term goal is to develop a 'gauge' which can provide an online prediction of simulator sickness level when given a pilot's associative behavioral patterns (learned expectations). This will allow informed decisions to be made on when to terminate a hop and provide an effective basis for determining the training and flight restrictions to be placed upon the pilot after simulator use.<
>