Subliminal calibration for machine operation

This paper addresses a calibration technique for machine operation systems enhancing the operability. Humans have an ability to learn the machine operation unconsciously. Therefore, proposed technique is attempted to utilize the human ability. Optimum operability in machine operation is defined as when a operated machine with ¿human calculated dynamics model¿ coincides with its ¿machine dynamics.¿ Based on this definition, we propose calibration that brings the machine dynamics closer to the human calculated machine model. The human learning ability corresponds to approach the human calculated dynamics model to operated machine dynamics. So changes in machine dynamics appear to pose problems that both make operators uncomfortable and hinder their learning. The calibration technique for changes in machine dynamics without operator awareness quantifying perception based on cognitive scientific knowledge is proposed. We applied this calibration, named subliminal calibration, to a task involving the maneuvering of a mobile vehicle and the performance is improved. Since the scheme employs human input prediction by human model with neural network, the calibration performance depends on its prediction accuracy. A technique to improve the prediction accuracy is also discussed in this paper. Finally, the application of the subliminal calibration is mentioned.