Limits...
Exploiting Task Constraints for Self-Calibrated Brain-Machine Interface Control Using Error-Related Potentials.

Iturrate I, Grizou J, Omedes J, Oudeyer PY, Lopes M, Montesano L - PLoS ONE (2015)

Bottom Line: This paper presents a new approach for self-calibration BCI for reaching tasks using error-related potentials.The results show that it is possible to have a usable BCI control from the beginning of the experiment without any prior calibration.Furthermore, comparisons with simulations and previous results obtained using standard calibration hint that both the quality of recorded signals and the performance of the system were comparable to those obtained with a standard calibration approach.

View Article: PubMed Central - PubMed

Affiliation: Chair in Brain-Machine Interface (CNBI) and Center for Neuroprosthetics (CNP), École Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland; Instituto de Investigación en Ingeniería de Sistemas (I3A), Universidad de Zaragoza, Zaragoza, Spain.

ABSTRACT
This paper presents a new approach for self-calibration BCI for reaching tasks using error-related potentials. The proposed method exploits task constraints to simultaneously calibrate the decoder and control the device, by using a robust likelihood function and an ad-hoc planner to cope with the large uncertainty resulting from the unknown task and decoder. The method has been evaluated in closed-loop online experiments with 8 users using a previously proposed BCI protocol for reaching tasks over a grid. The results show that it is possible to have a usable BCI control from the beginning of the experiment without any prior calibration. Furthermore, comparisons with simulations and previous results obtained using standard calibration hint that both the quality of recorded signals and the performance of the system were comparable to those obtained with a standard calibration approach.

No MeSH data available.


Experimental protocol.(a) Experimental protocol. The protocol showed a 5 × 5 grid with a virtual cursor (green circle) and a goal location (shadowed in red). (b) The cursor could perform five different actions (from top to bottom, move one position up, down, left, right, or performing a goal-reached action). (c) Correct actions (i.e. optimal policy) from each state for the goal exemplified in (a). Extracted from [18] with permission.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4488878&req=5

pone.0131491.g001: Experimental protocol.(a) Experimental protocol. The protocol showed a 5 × 5 grid with a virtual cursor (green circle) and a goal location (shadowed in red). (b) The cursor could perform five different actions (from top to bottom, move one position up, down, left, right, or performing a goal-reached action). (c) Correct actions (i.e. optimal policy) from each state for the goal exemplified in (a). Extracted from [18] with permission.

Mentions: The usability of the self-calibration approach was tested under an experimental protocol demonstrated in a prior work [18]. The protocol consisted of a 5 × 5 squares grid, a virtual cursor, and a goal location (see Fig 1). The cursor performs five different instantaneous actions: move one position up, down, left, right; and a goal-reached action, represented as concentric circumferences (see Fig 1b). The time between two actions (inter-action interval) was random within the range [3,3.5] s. The role of the subjects was to assess the cursor actions as correct when the cursor performed (i) a movement towards the goal position, or (ii) a goal-reached action over the goal position; or as incorrect otherwise (see Fig 1c). These assessments generated the associated evoked potentials for both correct and error conditions (i.e., error-related potentials, ErrP [21]). Note that the goal position was known by the user, but it was unknown by the device. The users were instructed not to move their eyes during the cursor actions, and to restrain blinks only to the resting periods. The experiment duration was fixed to 500 actions performed by the device, with rest intervals every 10 actions. Every time the device reached a target, the target was changed to a different position, with the sequence of target positions the same for all the subjects. The total length of the experiment was around 50 minutes. For more details about the protocol, please refer to [18].


Exploiting Task Constraints for Self-Calibrated Brain-Machine Interface Control Using Error-Related Potentials.

Iturrate I, Grizou J, Omedes J, Oudeyer PY, Lopes M, Montesano L - PLoS ONE (2015)

Experimental protocol.(a) Experimental protocol. The protocol showed a 5 × 5 grid with a virtual cursor (green circle) and a goal location (shadowed in red). (b) The cursor could perform five different actions (from top to bottom, move one position up, down, left, right, or performing a goal-reached action). (c) Correct actions (i.e. optimal policy) from each state for the goal exemplified in (a). Extracted from [18] with permission.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4488878&req=5

pone.0131491.g001: Experimental protocol.(a) Experimental protocol. The protocol showed a 5 × 5 grid with a virtual cursor (green circle) and a goal location (shadowed in red). (b) The cursor could perform five different actions (from top to bottom, move one position up, down, left, right, or performing a goal-reached action). (c) Correct actions (i.e. optimal policy) from each state for the goal exemplified in (a). Extracted from [18] with permission.
Mentions: The usability of the self-calibration approach was tested under an experimental protocol demonstrated in a prior work [18]. The protocol consisted of a 5 × 5 squares grid, a virtual cursor, and a goal location (see Fig 1). The cursor performs five different instantaneous actions: move one position up, down, left, right; and a goal-reached action, represented as concentric circumferences (see Fig 1b). The time between two actions (inter-action interval) was random within the range [3,3.5] s. The role of the subjects was to assess the cursor actions as correct when the cursor performed (i) a movement towards the goal position, or (ii) a goal-reached action over the goal position; or as incorrect otherwise (see Fig 1c). These assessments generated the associated evoked potentials for both correct and error conditions (i.e., error-related potentials, ErrP [21]). Note that the goal position was known by the user, but it was unknown by the device. The users were instructed not to move their eyes during the cursor actions, and to restrain blinks only to the resting periods. The experiment duration was fixed to 500 actions performed by the device, with rest intervals every 10 actions. Every time the device reached a target, the target was changed to a different position, with the sequence of target positions the same for all the subjects. The total length of the experiment was around 50 minutes. For more details about the protocol, please refer to [18].

Bottom Line: This paper presents a new approach for self-calibration BCI for reaching tasks using error-related potentials.The results show that it is possible to have a usable BCI control from the beginning of the experiment without any prior calibration.Furthermore, comparisons with simulations and previous results obtained using standard calibration hint that both the quality of recorded signals and the performance of the system were comparable to those obtained with a standard calibration approach.

View Article: PubMed Central - PubMed

Affiliation: Chair in Brain-Machine Interface (CNBI) and Center for Neuroprosthetics (CNP), École Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland; Instituto de Investigación en Ingeniería de Sistemas (I3A), Universidad de Zaragoza, Zaragoza, Spain.

ABSTRACT
This paper presents a new approach for self-calibration BCI for reaching tasks using error-related potentials. The proposed method exploits task constraints to simultaneously calibrate the decoder and control the device, by using a robust likelihood function and an ad-hoc planner to cope with the large uncertainty resulting from the unknown task and decoder. The method has been evaluated in closed-loop online experiments with 8 users using a previously proposed BCI protocol for reaching tasks over a grid. The results show that it is possible to have a usable BCI control from the beginning of the experiment without any prior calibration. Furthermore, comparisons with simulations and previous results obtained using standard calibration hint that both the quality of recorded signals and the performance of the system were comparable to those obtained with a standard calibration approach.

No MeSH data available.