Aarav Sharma is still at work.
If you subscribe to the Make: Substack, back in July you read about Aarav Sharma and the 3D-printed, brain-controlled robot arm he started building when he was 14.
Well it turns out Sharma hasn’t stopped working yet. Since then, he’s been busy expanding control beyond the arm into a wider communication system.
Enabled by a sponsorship from OpenBCI, he’s built a six-command Steady State Visually Evoked Potential (SSVEP) system. Think of it like eye tracking from the inside, or as Sharma explains, “SSVEP is a type of brain-computer interface where different targets flash at different frequencies. When the user focuses on one target, the EEG headset detects the matching response in the brain and converts it into a command.”
Speed is helped with word suggestions from an OpenAI language model, while the user retains control of the final output. In his tests, people were able to respond with full sentences in under five seconds.
“For offline EEG data collection, the user completes around 60 trials where the system tells them which of the six flashing targets to focus on. I then test how the accuracy changes when using different amounts of EEG data, and use TFR graphs to check the detected frequencies and a confusion matrix to see which targets are being confused.”
He plans on eventually moving the UI from a computer screen to AR glasses, for a more portable and practical experience.
Read more about his BCI projects at aaravinnovate.com/bci-neurova.
ADVERTISEMENT
