WiMi Hologram Cloud Inc., a leading global Hologram Augmented Reality (AR) Technology provider, today announced that it had made significant progress in the field by developing a humanoid control system based on hybrid signals BCI.
The technical implementation path of the system consists of several steps. First, multiple sensors need to be used to record multiple complementary signals such as Electromyography (EMG), Electroencephalogram (EEG), Electrooculography (EOG) and Event-related Desynchronization (ERD), Steady-state Visual Evoked Potentials (SSVEP), and Near-infrared Spectroscopy (NIRS).
The signals recorded by these sensors need to be pre-processed to remove interfering signals, noise reduction, etc. Then, the signals are subjected to feature extraction, signal classification, and other operations using machine learning algorithms to decode the BCI signals accurately. Finally, the decoding results are mapped to the humanoid robot control to realize the power of the humanoid robot.
The system offers several advantages over traditional BCI techniques. First, by recording and analyzing multiple complementary signals, brain activity information can be more comprehensively obtained, thus improving the accuracy and robustness of decoding. Second, data fusion technology can further enhance the robustness and reliability of the system and avoid recognition errors caused by the specificity of a single signal.
In addition, machine learning algorithms can further improve the decoding speed and accuracy, thus increasing the information transmission rate. Finally, the humanoid control system based on hybrid BCI technology can achieve more natural and precise control, which can be applied to many fields, such as the robot and disabled assistance.
Specific advantages of the hybrid BCI technique over the conventional BCI technique are as follows:
- Improved accuracy and robustness: Hybrid BCI technology utilizes multiple complementary signal sources, such as EMG, EEG, EOG, and SSVEP, to improve accuracy and robustness through data fusion techniques. Numerous sources can provide more comprehensive and reliable information than a single source, thus improving the accuracy and robustness of the system.
- Enhanced information transfer rate: In traditional BCI technology, a single source may need to provide more information for high-speed human-computer interaction. Hybrid BCI technology, however, combines multiple sources to enhance the information transfer rate, resulting in faster and more natural human-computer exchange.
- Improved Suitability and Operability: Hybrid BCI technology takes advantage of multiple signal sources, which can improve the suitability and operability of the system. For example, some users may need help interacting effectively with a single source, but combining multiple sources can provide more options and make it easier to achieve effective interaction.
- Improved training efficiency: In traditional BCI techniques, training a single signal source usually requires much time and effort. In contrast, hybrid BCI techniques can take advantage of multiple signal sources and improve training efficiency through data fusion techniques, resulting in faster and more reliable interactions.
The technical framework of hybrid BCI is mainly based on techniques of signal acquisition, signal pre-processing, feature extraction, feature selection, and classifier training. By combining multiple signal sources and machine learning algorithms, higher control accuracy and robustness can be achieved. The system uses numerous signal sources, including EMG, EEG, EOG, and NIR spectra.
Combining these signal sources through data fusion techniques improves the accuracy and robustness of the control system on the one hand. At the same time, the system also has high-speed information transmission capability, which enables users to achieve natural and efficient human-computer interaction through simple thought commands.
The technical framework and the specific implementation path of the system can be divided into the following steps:
- Signal acquisition: multiple complementary signal sources are acquired using numerous sensors. These sources can provide different information, such as muscle movement, brain activity, attention, etc.
- Signal pre-processing: The acquired signals are pre-processed, such as denoising, filtering, feature extraction, etc., to improve the quality and accuracy of the signals. For example, standard pre-processing methods such as average removal, band-pass filtering, wavelet transform, etc., can reduce signal noise and extract useful features.
- Feature extraction: Machine learning algorithms extract features from the pre-processed signal, such as time domain features, frequency domain features, wavelet transforms, etc. These features can provide important information about brain or muscle movements.
- Feature selection: Feature selection is performed based on the importance of the features to reduce the number of features and computational complexity. For example, a regularization-based sparsification method can be used to select important features.
- Classifier training: The classifier is trained using a training set, such as Support Vector Machine (SVM), Random Forest, etc. The classifier can map the input signal to a specified action or command.
- System integration: Integration of all components into a complete system, including signal acquisition, pre-processing, feature extraction, feature selection, and classifier training. The system can communicate with external devices such as robots, prostheses, or game controllers and send commands or actions to the devices.
The humanoid control system also has good applicability and operability and can be adapted to the needs and characteristics of different users. The system is also highly efficient, allowing users to complete training quickly and quickly achieve reliable interaction.
This humanoid control system has a wide range of application prospects. For example, it can be applied to the rehabilitation and assistive treatment of people with disabilities. By monitoring and recognizing the muscle and brain signals of people with disabilities, the humanoid robot can be precisely controlled to help people with disabilities live and work more autonomously.
In addition, the technology can be widely used in manufacturing. Monitoring and recognizing signals from employees’ muscles and brains enables precise control of robots in production lines, thereby improving productivity and product quality. It will be applied to medical care, smart home, and entertainment to give people a more convenient and efficient life experience.
In addition, WIMI will continue to promote technological innovation and research and development to continuously improve product performance and functions to provide users with better services and experiences. With the continuous development and application of hybrid BCI technology, the system will bring users a more intelligent and efficient human-computer interaction experience and significantly contribute to the development of BCI technology.