System and method for continual decoding of brain states to multi-degree-of-freedom control signals in hands free devices

1. A brain-machine interface system configured to decode neural signals to control a target device, the brain-machine interface system comprising:

at least one sensor configured to sample the neural signals; and

a computer-readable storage medium having software instructions stored therein, which, when executed by a processor, cause the processor to:

transform the neural signals into a common representational space stored in the brain-machine interface system;

provide the common representational space as a state representation to inform an Actor recurrent neural network policy of the brain-machine interface system;

generate and evaluate, utilizing a deep recurrent neural network of the brain-machine interface system having a generative sequence decoder, predictive sequences of control signals for the target device;

supply a specific control signal derived from the predictive sequences of control signals to the target device to achieve an output of the target device;

determine an intrinsic biometric-based reward signal, from the common representational space, based on an expectation of the output of the target device; and

supply the intrinsic biometric-based reward signal to a Critic model of the brain-machine interface system.

2. The brain-machine interface system of claim 1 , wherein the at least one sensor is configured to sample the neural signals invasively or non-invasively.

3. The brain-machine interface system of claim 2 , wherein a sensor of the at least one sensor comprises an invasive electrocorticographic (ECoG) device or an intracranial electroencephalography (iEEG) device.

4. The brain-machine interface system of claim 1 , wherein transforming the neural signals into the common representational space comprises:

identifying regions of the neural signals with informative activations for controlling the target device, and

performing subject-specific transforms to align the regions across different users.

5. The brain-machine interface system of claim 1 , wherein the intrinsic biometric-based reward signal is indicative of a positive emotional response when the output of the target device matches a user's intended output.

6. The brain-machine interface system of claim 1 , wherein the intrinsic biometric-based reward signal is indicative of a negative emotional response when the output of the target device does not match a user's intended output.

7. The brain-machine interface system of claim 1 , wherein the software instructions, when executed by the processor, cause the processor to generate and evaluate the predictive sequences of the control signals utilizing a tree search.

8. A non-transitory computer-readable storage medium having software instructions stored therein, which, when executed by a processor, cause the processor to:

transform neural data from an individual user into a common representational space and store the common representation space in a memory device of a brain-machine interface system;

provide the common representational space as a state representation to inform an Actor recurrent neural network policy of the brain-machine interface system;

generate and evaluate, utilizing a deep recurrent neural network of the brain-machine interface system having a generative sequence decoder, predictive sequences of control signals for a target device;

supply a specific control signal derived from the predictive sequences of control signals to the target device to produce an output of the target device;

determine an intrinsic biometric-based reward, from the common representational space, based on the individual user's expectation of the output of the target device; and

supply the intrinsic biometric-based reward to a Critic model of the brain-machine interface system.

9. The non-transitory computer-readable storage medium of claim 8 , wherein the instructions, when executed by the processor, cause the processor to transform the neural data into the common representational space by:

identifying regions of the neural data with informative activations for controlling the target device, and

performing subject-specific transforms to align the regions across different users.

10. The non-transitory computer-readable storage medium of claim 8 , wherein the instructions, when executed by the processor, cause the processor to determine the intrinsic biometric-based reward by decoding an emotional response from the individual user to the output of the target device.

11. The non-transitory computer-readable storage medium of claim 10 , wherein the intrinsic biometric-based reward is a positive emotional response when the output of the target device matches the individual user's intended output.

12. The non-transitory computer-readable storage medium of claim 10 , wherein the intrinsic biometric-based reward is a negative emotional response when the output of the target device does not match the individual user's intended output.

13. The non-transitory computer-readable storage medium of claim 8 , wherein the software instructions, when executed by the processor, cause the processor to generate and evaluate the predictive sequences of the control signals utilizing a tree search.

14. A method of controlling a target device utilizing neural data, the method comprising:

sampling, with at least one sensor, the neural data from a user;

transforming, with a processor of a brain-machine interface system, the neural data into a common representational space and storing the common representational space in a memory device of the of a brain-machine interface system;

supplying, by the processor of the brain-machine interface system, the common representational space as a state representation to inform an Actor recurrent neural network policy of the brain-machine interface system;

generating and evaluating, utilizing a deep recurrent neural network of the brain-machine interface system having a generative sequence decoder, predictive sequences of control signals for the target device;

supplying, by the processor of the brain-machine interface system, a specific control signal derived from the predictive sequences of control signals to the target device to produce an output of the target device;

determining, by the processor of the brain-machine interface system, an intrinsic reward, from the common representational space, based on the user's expectation of the output of the target device; and

supplying, by the processor of the brain-machine interface system, the intrinsic reward to a Critic model of the brain-machine interface system.

15. The method of claim 14 , wherein the evaluating the predictive sequences of the control signals comprises a tree search.

16. The method of claim 14 , wherein the transforming the neural data into the common representational space comprises:

identifying regions of the neural data with informative activations for controlling the target device, and

performing subject-specific transforms to align the regions across different users.

17. The method of claim 14 , wherein the determining the intrinsic reward comprises decoding an emotional response from the user to the output of the target device.

18. The method of claim 17 , wherein the emotional response is a positive emotional response when the output of the target device is expected.

19. The method of claim 17 , wherein the emotional response is a negative emotional response when the output of the target device is unexpected.

20. The method of claim 14 , wherein the sampling the neural data from the user is performed invasively or non-invasively.