UC Riverside

Brain-computer interfaces (BCIs) are recent developments in alternative technologies of human computer interaction. These interfaces aim to interpret the brain's activity as user intentions in active BCI systems or cognitive/ emotional state in passive BCI systems.

This dissertation focuses on implementation of BCIs in different aspects of Computer Aided Design system. Specifically, the dissertation explores the use of BCI in creating, selecting and modifying objects in CAD systems.

Geometry creation is achieved through visual imagery by recording and analyzing EEG signals when subjects imagine distinct shapes. The algorithms developed in this research successfully recreated the primitive shape that a subject imagined with an average accuracy of 44.6% (chance accuracy is 20%). The research further indicated that geometrical properties of objects such as roundness and parallel extrusion are also salient in classifying imagined objects.

Selection and modification of objects is obtained by developing algorithms based on the P300 characteristic of the EEG signal and motor imagery. The classification result indicates the proposed method can select the target object /face with an average accuracy of 74%.

Furthermore, the dissertation discusses a method to estimate user emotions of satisfaction and frustration that follow the successful or unsuccessful execution of a user command by the computer. This estimate is derived based on a combination of relative power spectral density and largest Lyapunov exponents. The results show that the algorithms can determine the user's satisfaction level with an average accuracy of 79%.