Brain interfaces
Taking visual EEG interaction beyond its original host.
This work rebuilds NextMind acquisition and visual decoding for Python, Linux and Android. It provides the platform foundations for experiments in gaze, EEG selection and robot control.
01 / The interaction
What must move with a portable brain interface?
The original interaction combines changing visual targets with the wearer’s EEG response. Portable implementations need to recover the device stream, preserve timing, calibrate the decoder and translate a selection into an application action.
Visual target timing and EEG acquisition are part of the same decoding problem.
02 / Research & engineering
What went into it.
Device acquisition
Independent acquisition recovers eight signed 24-bit EEG channels at a nominal 250 Hz. Packet counters help detect interruptions.
Python integration
A Windows Python wrapper exposes calibration and selection events from the original native engine.
Linux & Android
Independent device-protocol and cVEP decoding work extends the workflow to Linux and Android.
Mobile interaction
An Android app supports connection, calibration, configurable visual targets and UDP output.
VR drone interaction
The earlier Mindrone setup maps visual targets to flight commands and headset rotation to yaw. Interface iterations balance marker visibility with a clear first-person camera view.
Controller research
A Tello interface combines decoding, video, visual controls and command policies, including pause/land handling and stale-input expiry.
03 / Progress
Built. Tested. Next.
Device and platform foundations
Acquisition, platform ports and configurable target control provide working building blocks.
Controller policy
Nine Android policy tests used scripted target identities to check software behavior.
From Visual Attention to Robot Commands
A working manuscript brings together the physical Reach, earlier VR drone and portable interface work.
Physical and timed evaluation
Validate the Android controller with hardware and synchronize gaze, EEG, commands and motion.
The Android controller has software-level checks; physical flight with that build remains unverified. The manuscript is a working draft, not a peer-reviewed publication.
Continue exploring