YeetMind

It was our goal to build an AI device that integrates a personal EEG and smart car to remotely monitor users awareness, fatigue and attention. Our goal is simple, increase driver awareness and save lives! This technology could be integrated into a fleet of drivers ensuring that drivers are not overly fatigued and are paying attention while driving whilst simutaneously providing a web app for external monitoring.

Using the computation of the Jetson nano we were able to implement a powerful AI on the edge system that make real time EEG data inferences from three different support vector machine models.

A Muse2 EEG device is used with Fourier transforms to produce a vector of frequencies and amplitudes. This vector is passed into an SVM to classify brain states, which are used to control the car.

We also measure fatigue over time and pull the car over to the side of the road if driver has lost attention for too long.

The SVM is trained to detect the following states, and their corresponding control mappings:

eyes open -> drive normal speed
eyes closed -> slow down
fatigue level : 1 -> 3 Scale
attention to road : Binary classifier

A huge challenge we had to overcome was finding a functional SDK for the muse headset that was also compatible with armhf. Our solution was to connect the muse headset via the web-browser and forward the data using socket.io and a Javascript ROS library.

We will be applying what we learned to help the makers of eegedu.com make an integrated Muse web SDK so others can easily integrate Muse data into their own projects via the web browser, no dependencies required! Again, we cannot thank the Mathewson brothers enough for their advise and eegedu which we used to collect data and served as a backbone for our webapp.