Detection of tactile-based error-related potentials (ErrPs) in human-robot interaction

Area of Science:

• Human-Robot Interaction
• Neuroscience
• Robotics

Background:

• Implicit error detection via electroencephalography (EEG) aids robot learning when explicit criteria are difficult.
• Error-related potentials (ErrPs) are brain signals indicating error recognition.
• Previous research primarily used visual or visuo-tactile stimuli to evoke ErrPs.

Purpose of the Study:

• To investigate ErrPs evoked solely through tactile error recognition in human-robot interaction (HRI).
• To assess the impact of motor responses on tactile ErrP classification accuracy.
• To explore the transferability of tactile ErrP detection across different response conditions.

Main Methods:

• Subjects recognized errors from an orthosis's incorrect arm movement via tactile cues.
• EEG data was recorded from eight subjects.
• ErrP detection was tested under three conditions: motor response, no motor response, and delayed motor response.

Main Results:

• High ErrP classification performance was achieved with both motor response and no motor response conditions.
• Classification performance decreased when transferring between motor response and delayed motor response conditions.
• Tactilely induced errors were detected with high accuracy from brain activity.

Conclusions:

• Tactile error recognition can reliably evoke detectable ErrPs.
• Motor responses may not be essential for accurate tactile ErrP classification.
• Further systematic investigation of tactile-based ErrP classification is warranted.