You are here

ECE PhD Defense: "FlashLife(tm), A Context-Aware code-VEP based Brain Computer Interface for Daily Life using EEG Signals," Hooman Nezamfar


140 The Fenway, Room 378

March 21, 2016 3:00 pm
March 21, 2016 3:00 pm


Loosing communication and control abilities imposes many restrictions on how, when and if different tasks can be done by the affected individuals. 

Unfortunately, accidents and many harmful diseases such as Amyotrophic Lateral Sclerosis (ALS) enforce such disabilities on individuals on a daily basis around the globe. Advances in the technology and medicine have made it possible to know more and do more in terms of assistive technology during the past few decades. However, still most of the assistive devices are designed for specific tasks, such as typing or control.

In this dissertation, we introduce FlashLife(tm), a context aware language independent brain interface, suitable for everyday needs of an individual with disabilities. FlashLife(tm) provides control and communication abilities all through the same stimulation method using a single EEG electrode or eye tracking. In addition, use of the context information along with a probabilistic classification and decision making mechanism adds more robustness and flexibility at the same time. The stimulation paradigm provides highly accurate and fast classifications making use of short Calibration sessions. FlashLife(tm) provides performance estimates for each individual for different tasks taking advantage of the Calibration data.

The stimulation paradigm has been put into use by different applications to do different tasks. A short list of applications is, FlashType(tm) for typing, FlashNav(tm) for navigation, FlashGrab(tm) for object manipulation and FlashPlay(tm) for entertainment in a virtual environment.

FlashType(tm); A context aware language independent typing brain interface. It provides the user with a cursor, capable of navigating throughout a grid of letters or symbols. This keyboard consists of three main parts, Static Keyboard, Character Suggestion, and Word Prediction. By default, using a 6-gram language model and typing history, 7 highly probable characters and 3 most probable words are estimated and presented to the user.

FlashNav(tm); A context aware navigation brain interface. It can be used to navigate a wheelchair or control a robot remotely. Information such as environment map, objects and locations of interest and user habits can be used to boost the probabilistic decision making performance.

FlashGrab(tm); A context aware object manipulation brain interface. Using Baxter, a low cost humanoid robot, and image processing techniques, graspable objects are detected and labeled. Depending on the number of graspable handles, a direct or a multistep decision will be made by the user.

FlashPlay(tm); An interface to a virtual environment such as a maze or a floor map. Training and entertaining the user are the main goals. A series of Mastery tasks have been designed with different difficulty levels, taking advantage of the probabilistic classifier and the virtual environment, to help the users to build the habit of using the system and attending to the stimuli effectively.

Advisor: Professor Deniz Erdogmus 

Professor Deniz Erdogmus 
Professor Dana Brooks
Professor Gunar Schirner
Professor Frank Guenther (Boston University)