Figure 1.
Overall scenario of humanoid navigation/exploration and recognition tasks.
Figure 2.
Experimental setting.
Figure 3.
Proposed algorithm for navigation and exploration mode.
Figure 4.
Proposed procedure of recognition mode.
Figure 5.
Timing diagram of stimulus events and corresponding epoch windows for EEG signals.
Figure 6.
Proposed selection rule of ERD and SSVEP-based BCI protocols.
Figure 7.
The design of the display during (a) walking straight, (b) turning head, and (c) recognition.
(d) The interface being used by a subject.
Figure 8.
The system architecture.
Table 1.
Cross-validation accuracies and ITRs of BCI protocols used for navigation/exploration.
Figure 9.
SSVEP and ERD-based BCI selection: (a) selection classification region: dotted lines indicate decision boundaries, and (b) average brain activities specified from O1 during BCI operation of subject C.
Table 2.
Real-time humanoid navigation/exploration control results.
Figure 10.
Per subject recognition accuracies over the number of flashes per object of (a) two objects on the computer monitor, (b) four objects on the computer monitor, (c) two objects through robot vision in real-time, and (d) four objects through robot vision in real-time.
Table 3.
ITRs during various object recognition scenarios.
Table 4.
Real-time humanoid navigation/exploration and recognition control results.
Figure 11.
Pictures taken during a real-time humanoid experiment: (a) walking forward, (b) looking around, and (c) recognizing an object.
A movie clip demonstrating the experimental performance has been included as Video S1.