In order to improve work efficiency and cope with complex control tasks that are difficult for a single person to complete, post-90s college students try brain-brain collaborative writing for the first time. Operator? Operation. ? Brain-brain cooperation is not a simple problem of 1+ 1=2. Based on the brain control system with the largest instruction set in the world, the team innovatively designed a brain-brain cooperative operation mode based on time-frequency mixed brain information coding. ? Brain wave module will reach 2 16 instruction sets, which greatly improves the dimension and output speed of brain control instructions again, making? Nezha? Intelligent arms can perform higher-order brain control actions.
Compared with the first generation intelligent manipulator system developed by 20 19, its control efficiency and information transmission rate are more than doubled compared with that of single operator, and the spatial resolution of brain control view is from 9? 12 pixels increased to 12? More than 18 pixels, can be competent for more complex tasks.
The human brain is composed of thousands of neurons, and brain waves are electrical signals generated by the activities between these neurons. The connections between these neurons are either stimulated or inhibited; Thinking activity reflects the connection between these neurons. Neurons in the brain receive signals from other neurons. When the energy accumulation of these signals exceeds a certain value, brain waves will be generated. In order to detect brain waves, people usually put electrodes on people's scalp to detect brain wave signals, and then use related equipment to collect and process brain waves. The uncertainty and randomness of single-lead EEG signals in brain waves are poor, and the nonlinear research is limited, which leads to poor recognition effect. Multi-lead EEG signals contain more information about brain activity, which can better reflect the overall information of brain activity. In the study of decoding speech directly from brain waves, the brain-computer interface system is limited to decoding monosyllables, or when volunteers read about 100 words continuously, less than 40% of the words can be correctly decoded.
Inspired by the machine, the research team trained a circular neural network. In this study, four volunteers were asked to read 30 to 50 sentences aloud. A large number of microelectrodes are distributed in the lateral cerebral cortex, which can monitor the corresponding brain nerve activity. These brainwave data are encoded into a series of sequences, which are input into the artificial intelligence system and then decoded into corresponding English sentences. This study proves the high accuracy and natural speech speed of cortical EEG decoding. In a volunteer's brainwave decoding task, only 3% of each sentence needs to be corrected on average, which is better than the average error rate of 5% of professional manual stenographers. [