The brain-computer interface (BCI) uses different brain signals, recording methods, and signal-processing algorithms to build a link between the brain and external software/hardware platforms. With the development of computer hardware (e.g., GPU) and algorithms (e.g., machine learning, deep learning), BCI is becoming more practical and stable. Currently, the potential applications based on BCI include, but are not limited to, brain-controlled robot/exoskeleton/wheelchair, brain-controlled platforms for ALS patients, stroke/trauma rehabilitation, etc. With the upgrading of electrodes and chips, the safety of BCI is gradually improving. Minimally invasive and non-invasive devices are frequently used in BCIs. From the algorithm perspective, with the development of machine/deep learning, the decoding speed and accuracy are significantly increased.
Neuralink has successfully manufactured a very small chip with thin electrodes. But it is far from perfect. One difficulty ahead is perfecting microwires that can survive the “corrosive” context of a living brain for years. Another problem is that no matter how tiny the electrode is, it is still invasive. Other invasive electrodes, such as Utah, also face the same problem. Non-invasive BCI is always the first choice for clinical and scientific research and is more conducive to the promotion for daily and home use. How to make non-invasive BCI comparable to invasive BCI is a problem that needs to be solved urgently. From an algorithm perspective, the development of deep learning has brought more possibilities and solutions to BCI decoding, e.g, Shenoy successfully used RNN to implement brain-to-text handwriting, Tonio Ball constructed the deep/shallow CNN for EEG Decoding and Visualization. However, the application of deep learning in BCI decoding still faces several problems: 1. much higher computational cost than traditional machine learning algorithm (e.g., support vector machine, random forest), 2. poor interpretability, 3. low reproducibility and transferability, and 4. evaluation on large dataset and population.
Frontiers in Neurorobotics publishes rigorously, peer-reviewed research in the science and technology of embodied autonomous systems as well as their applications. We encourage all potential contributors to submit their high-quality original research papers related to BCI to this Research Topic. The article can be a review, research paper, or case report. The scope includes all aspects of BCI research, including but not limited to, new software/hardware development in BCI, new decoding algorithms (based on deep learning, machine learning, etc), new chip, new flexible EEG electrode, wearable systems, BCI-based rehabilitation, BCI-based emotion detection, BCI based on virtual reality (VR), ERP-based BCI, SSVEP-based BCI, BCI in stroke, BCI in epilepsy, BCI in attention deficit hyperactivity disorder (ADHD), BCI in other diseases.
The brain-computer interface (BCI) uses different brain signals, recording methods, and signal-processing algorithms to build a link between the brain and external software/hardware platforms. With the development of computer hardware (e.g., GPU) and algorithms (e.g., machine learning, deep learning), BCI is becoming more practical and stable. Currently, the potential applications based on BCI include, but are not limited to, brain-controlled robot/exoskeleton/wheelchair, brain-controlled platforms for ALS patients, stroke/trauma rehabilitation, etc. With the upgrading of electrodes and chips, the safety of BCI is gradually improving. Minimally invasive and non-invasive devices are frequently used in BCIs. From the algorithm perspective, with the development of machine/deep learning, the decoding speed and accuracy are significantly increased.
Neuralink has successfully manufactured a very small chip with thin electrodes. But it is far from perfect. One difficulty ahead is perfecting microwires that can survive the “corrosive” context of a living brain for years. Another problem is that no matter how tiny the electrode is, it is still invasive. Other invasive electrodes, such as Utah, also face the same problem. Non-invasive BCI is always the first choice for clinical and scientific research and is more conducive to the promotion for daily and home use. How to make non-invasive BCI comparable to invasive BCI is a problem that needs to be solved urgently. From an algorithm perspective, the development of deep learning has brought more possibilities and solutions to BCI decoding, e.g, Shenoy successfully used RNN to implement brain-to-text handwriting, Tonio Ball constructed the deep/shallow CNN for EEG Decoding and Visualization. However, the application of deep learning in BCI decoding still faces several problems: 1. much higher computational cost than traditional machine learning algorithm (e.g., support vector machine, random forest), 2. poor interpretability, 3. low reproducibility and transferability, and 4. evaluation on large dataset and population.
Frontiers in Neurorobotics publishes rigorously, peer-reviewed research in the science and technology of embodied autonomous systems as well as their applications. We encourage all potential contributors to submit their high-quality original research papers related to BCI to this Research Topic. The article can be a review, research paper, or case report. The scope includes all aspects of BCI research, including but not limited to, new software/hardware development in BCI, new decoding algorithms (based on deep learning, machine learning, etc), new chip, new flexible EEG electrode, wearable systems, BCI-based rehabilitation, BCI-based emotion detection, BCI based on virtual reality (VR), ERP-based BCI, SSVEP-based BCI, BCI in stroke, BCI in epilepsy, BCI in attention deficit hyperactivity disorder (ADHD), BCI in other diseases.
