ERIC | Devpost

Inspiration

After acquiring an Emotiv EEG Headset at TreeHacks, our team was excited by the idea of using it towards some type of mental health app. This, combined with our aversion to loneliness, and our desire to be able to share our minds at any time, led to ERIC (Emotionally Responsive Interactive Character). ERIC is a smart web-psychotherapist who is there for you at any hour, minute, and second of the day.

How it works

The ERIC system is an extension of ELIZA, the result of a natural-language processing research project conducted at MIT by Joseph Weizenbaum in the 1960s. Our goal was to make the experience of conversing with ERIC as human as possible. The beauty of ERIC comes in his ability to read your mood through the Emotiv EEG Headset to prompt more relevant conversation. The experience is made even more tactile by the Leap-Motion enhanced playground that allows you to fiddle with virtual blocks during your psychotherapeutic experience! Lastly, ERIC uses text-to-speech technology to provide both vocal and textual feedback.

Challenges I ran into

1) Getting Emotiv to work via web client despite very sparse, unorganized documentation
2) Connecting the LeapMotion with our Three.js WebGL environment
3) Finding smart ways to enhance ERIC's conversational skills

Accomplishments that I'm proud of

1) Overcoming the aforementioned challenges
2) Creating our own collision detection system for our virtual blocks + Leap-Motion hands
3) Really pretty "feel" to our app

What I learned

Two members of our team were fairly new to web development. We were able to learn basic HTML+CSS web design, some JavaScript, and even some 3D web programming! Kelly, after programming for the first time implemented our block system using Three.js and Physi.js!! Liem, a freshman in high school, owned our web and graphic design, and contributed to our front-end code as well.

What's next for ERIC

We plan to further push ERIC's interactive capabilities using more complex NLP strategies. We also wanted to incorporate speech-to-text recognition, but decided not to given our time constraints and the loud, hectic nature of hackathon pitch expos.