Context Recognition in Seat Environment

Type: Master Thesis
Student: Claudia Lorenz
Advisor: Cornelia Setz, Johannes Schumm
Project: SEAT

The comfort of an airplane passenger is strongly related to his mental state. If the mental state can be determined automatically, a cabin environment might be created, which responds to the individual emotional condition of the passenger. This thesis took a rst step in this direction by designing experiments and algorithms to detect a passenger's emotional state from physiological signals, such as electrocardiogram (ECG), electromyogram (EMG), electrooculogram (EOG), galvanic skin response (GSR), respiration and nger temperature. To elicit specic emotions in an experimental environment, dierent lm clips were used, which were evaluated and selected in a pre-study with 14 subjects. As ground truth for the elicited emotions, two questionnaires, provided in the literature, were deployed. The five emotions amusement, anger, contentment, neutral and sadness were elicited during the main experiment with 20 subjects. Appropriate software algorithms allowed the feature extraction of 41 features from the six recorded signals. Afterwards, dierent feature selection methods and classication algorithms were used to classify the data. The support vector machine and the linear discriminant analysis combined with the sequential forward floating selection reached a recognition rate of 66% in classifying five emotional states.