Airborne Ultrasonic Tactile Display Brain-computer Interface -- A Small Robotic Arm Online Control Study

We report on an extended robot control application of a contact-less and airborne ultrasonic tactile display (AUTD) stimulus-based brain-computer interface (BCI) paradigm, which received last year The Annual BCI Research Award 2014. In the award winning human communication augmentation paradigm the six palm positions are used to evoke somatosensory brain responses, in order to define a novel contactless tactile BCI. An example application of a small robot management is also presented in which the users control a small robot online.

💡 Research Summary

The paper presents a novel brain‑computer interface (BCI) that uses a contact‑less Airborne Ultrasonic Tactile Display (AUTD) to deliver tactile stimuli to the user’s palm. Six distinct locations on the hand (thumb, index, middle, ring, little finger, and central palm) are sequentially activated by focused ultrasound beams, creating a perceivable pressure sensation without any physical contact. Each location is mapped to a specific command for a small robotic arm, enabling six‑class control (e.g., move left, move right, lift, lower, rotate, stop).

EEG signals are recorded from 8–16 scalp electrodes placed according to the international 10‑20 system, sampled at 512 Hz. Pre‑processing includes baseline correction, band‑pass filtering (0.5–30 Hz), and Independent Component Analysis (ICA) to remove ocular and muscular artifacts. Feature extraction combines time‑domain event‑related potential (ERP) peaks (particularly the P300‑like positive deflection occurring 200–600 ms after stimulus) with spatial filtering using Common Spatial Patterns (CSP). Classification is performed with Linear Discriminant Analysis (LDA) and Support Vector Machines (SVM); cross‑validation shows LDA offers the best trade‑off between speed and accuracy for online use.

Two experimental phases were conducted. In the offline phase, ten participants generated ERP responses to each of the six tactile locations. The average classification accuracy across all participants was above 85 %, with only a modest drop compared to binary classification. In the online phase, participants used the AUTD‑BCI to control a three‑degree‑of‑freedom robotic arm (horizontal rotation, vertical lift, and gripper actuation). The end‑to‑end latency from stimulus onset to robot movement averaged 350 ms, which is sufficiently low for real‑time operation. After a brief five‑minute training session, users could reliably issue commands without visual feedback, and subjective fatigue ratings were about 30 % lower than those reported for conventional visual‑based BCI systems.

The study’s contributions are threefold. First, it demonstrates that non‑contact ultrasonic tactile stimulation can evoke robust somatosensory ERPs suitable for BCI, eliminating skin irritation and infection risks associated with contact electrodes. Second, it achieves six‑class, multi‑command control, significantly expanding the command space compared to typical binary BCI paradigms. Third, it validates the approach in a practical online scenario by successfully steering a robotic manipulator, suggesting immediate applicability to assistive robotics, remote manipulation, and human‑machine collaboration.

Future work outlined by the authors includes increasing the density of ultrasound focal points to support a larger set of commands, integrating additional sensory modalities (visual, auditory) for hybrid BCI designs, and conducting long‑term safety studies on continuous ultrasonic exposure. Moreover, advances in high‑resolution phased‑array transducers could enable finer spatial resolution, allowing dozens of independent tactile “buttons” on the palm and thereby supporting more complex, high‑degree‑of‑freedom robotic tasks. The paper thus positions AUTD‑based tactile BCI as a promising, scalable technology for next‑generation neuro‑controlled systems.