AMICO : AMbient Intelligent ClassrOom

Abstract

This research looks into the possibility of using real-time EEG data analysis in Classroom Ambient Intelligence systems to detect the level of participation of students during lectures from a technical research point of view. Classroom environments have evolved from the traditional setup where it is totally up to the lecturer to devise his own tools and techniques to foster interest in students and keep attention levels high, to smarter versions which provide some form of automation or environment enhancements with the intention of improving the overall experience of students. Therefore, the ability of future classrooms to be aware of the attention levels of their students might seem as the next logical step technology will take. Research in this area is still in its infancy, and at the time of writing consists of lab tests rather than an implementation that could be practically used in real life. The same could be said about EEG analysis, where scientists are still struggling to decipher the signals derived from combinations of humongous numbers of synapses occurring from brain neurons. The proposed proof of concept implementation uses a modified EEG cat-ears toy built around an off-the-shelf EEG reading and analysis module, NeuroSky’s TGAM. This device can read EEG signals, amplify and filter them and provide a stream of data ranging from eSense values to the actual raw brainwave signal. It uses a specifically developed protocol, built on Industry Standard IP and UDP to transmit data to other devices on the network, including a specially designed Tutor Device as well as monitoring and configuration software for system engineers to configure and fine tune the system. A modified K-Means machine learning algorithm is implemented in each individual EEG reading device, thereby taking advantage of the 160MHz processor in the ESP-12F MCUs used in the devices to implement a distributed processing concept which allows very high scalability, limited only by the underlying network infrastructure. An intuitive touch screen driven interface is used for the tutor’s device, with the main design objective being ease of use, indeed the device can be used and controlled with just one finger. With one of the main design objectives being the ability to be used in real life classrooms, the prototype system has been tested and evaluated with real life students in a classroom environment and achieved an accuracy of 82% with only 30 minutes of training. Questionnaires distributed to students and tutors revealed that students thought that the project is ambitions, some even doubting it would work at all; but with most acknowledging that such systems could be useful tools in future classrooms after using it. Similarly, tutors were somewhat sceptic of such systems at first, but quickly realized that the technology could benefit them and their students once the mentality and ethical issues associated with it are tackled. This study concluded that EEG driven Ambient Intelligent Classrooms could indeed be a reality in the future.