Guhan Sundar

EEG · MEG · WEARABLES · DATA INFRA — 2012 → NOW

Senior design — EEG motor imagery BCI

research · 2017-09 → 2018-05 · 8m

Built a brain-computer interface allowing navigation of a virtual environment by decoding brain state from a wearable EEG headset, with MATLAB feature extraction and classification integrated into a VR simulation game.

UC San Diego

What

A brain–computer interface that let a person navigate a virtual environment by imagining movement. A wearable EEG headset recorded motor-imagery activity; a MATLAB pipeline extracted features using common spatial patterns, classified them with a linear support vector machine, cross-validated the classifier, and translated its output into simulated keyboard and mouse input driving a VR navigation game.

Why it mattered

This is the hinge. Before it, neuroscience was incidental — a lab job, an interesting elective. Building a system that turned an intention into motion on a screen made it deliberate, and it exposed exactly where the difficulty sits: not in the classifier, which is textbook, but in the quality of the signal going into it. Everything since has been an attack on the input side of that pipeline.

My role

Collected the raw motor-imagery data, wrote the MATLAB feature-extraction and classification code, and integrated the classifier with the VR simulation.

Outcome

A working end-to-end system — headset to features to classifier to on-screen navigation — delivered as the capstone senior design project. The MATLAB pipeline is public.

Assets