A Method for Adjusting Operating Characteristics According to User Preferences using a Stress Evaluation Model based on Electroencephalogram Analysis

This paper presents a method for adjusting operating characteristics according to user preferences using a stress evaluation model based on electroencephalogram analysis. The stress evaluation model was constructed using a neural network (NN). A multirate sample entropy was used as a feature of electroencephalogram (EEG), which was the input of NN, and the output of NN was the stress value. Supervised stress values were collected from subjects using a questionnaire method. Because EEG equipment is expensive and its measurement is complicated, an alternative stress evaluation model was constructed using NN. The input of this model is the sample entropy of operating waveforms, and the output is the stress value. The output of the EEG-based evaluation model was used as the supervised value of this model. The operating characteristics were modeled as a second-order delay system, and the gain values were automatically adjusted by using NNs. One of the inputs of the NN is the output value of the evaluation model based on the operating waveform, and the output is the corrected value of the gain. Subjects were given the task of remotely operating an omnidirectional mobile robot along a specified path. After repeating the task several times, we confirmed that the robot could be operated according to the preferences of the subjects.