The brain-computer interface, or neural interface, is a technology that allows you to process electrical signals from the cerebral cortex, amplify and transfer them to a computer, then synchronize with any control device or computer application using processing algorithms.
About our technology
We have developed a new model and methodology for a non-invasive brain-computer interface:
Real-time recognition of up to 8 individual commands
Quick training an operator
- The classifier is able to recognize the neutral state, the state of concentration, relaxation, imaginary leg movements, the transfer of attention inside the head, mental silence
- The recognition frequency is about 10 Hz.
- For the state-command formation, the operator does not use the movement of the eyes and head, reduction of facial muscles, teeth clenching, etc.
Management preparation method
- The unprepared person can study in 10 minutes to manage two commands
- An experienced user is able to use 8 commands.
- With good preparedness, it is possible to combine management using a neural interface with physical activity to increase the productivity of performing certain tasks.
- A tutorial for beginners lasts from 30 minutes to learn 2-3 states of consciousness
- The tutorial for advanced users lasts from 14 hours for learning 4-7 states of consciousness
- Optimum training's time is from 30 to 60 minutes a day
- Steady management is achieved through regular training.
- First, the classifier must be trained neutral state, on the basis of which the noise is removed from the signal. This is not a state of relaxation, but a state of a person in the absence of purposeful mental activity.
- When adding new channels there is no need to retrain the classifier
- Training consists in the formation of such states of consciousness that the classifier is able to recognize. These include the state of concentration, relaxation, imaginary limb movements, transfer of attention inside the head, mental silence
- As a quick start, it is easy to learn two states - relaxation and concentration
- Learning consists of periodically repeating states a number of times (1-3)
After training, the classifier will give out discrete commands (for example, from 1 to 8), which are used to manage various virtual and real objects.
In the BioEcho software implemented several mechanisms for the formation of more stable and advanced meta-commands:
- Simple commands: each state is interpreted as one command (for example, suitable for real-time control of a quadcopter)
- Linear command: a command is formed only when a certain number of identical commands are received in a row. Example: expected "1" x 5, sequence "1 1 1 1 0 1 1 1 0 0 1 1 0 1" will not create command, sequence "1 0 1 0 1 1 1 1 1" will create one.
- Battery: a command is formed when a specified number of commands are accumulated. Example: expected "1" x 5, command works in cases "1 0 1 0 1 1 0 1", "1 2 3 3 2 1 1 1 0 3 2 3 1"
- Combinatorics: Allows you to increase the number of meta-commands from a limited set of commands. So, using only two commands (for example, "1" and "2"), you can get 4 commands: 11, 12, 21, 22. The number of meta-commands is n! where n is the number of input commands.