CN105676639A - Parallel multi-modal brain control method for complete grabbing operation of artificial hand - Google Patents

Abstract

The invention discloses a parallel multi-modal brain control method for complete grabbing operation of an artificial hand. On the basis of intelligent artificial hand control methods in the prior art that use EEG signals, the method integrates the reliability of a brain control paradigm that is evoked by vision on the basis of scene animations and of a brain control paradigm that is driven by expressions to improve the correct rate of complete grabbing operation of an artificial hand. EEG signals generated by visual evocation when a subject looks at different scene animations and EEG signals generated by different expressions are used as control information sources at the same time. By increasing individual differences between different types of EEG signals, the method achieves precise control of the artificial hand and further improves the accuracy and information transmission rate of a brain control artificial hand system.

Description

A Parallel Multimodal Brain Control Method for Complete Grasping Operation of Prosthetic Hand

technical field

The invention relates to the field of intelligent robots, in particular to a parallel multi-modal brain control method for a complete grasping operation of a prosthetic hand.

Background technique

With the continuous development of my country's economy, the total number of all kinds of disabled people and the proportion of the total population have increased, among them, amputation, amyotrophic lateral sclerosis (Amyotrophic Lateral Sclerosis, ALS) and other lack of physical mobility or muscle control ability There are a large number of disabled people, and their families often invest a lot of manpower and financial resources in care. In recent years, artificial hand control strategies based on EEG control sources have been gradually adopted by people. It uses EEG signals as a signal source to drive prosthetic hand movements, and establishes a direct connection between the two, thus helping some disabled people with spinal cord or peripheral nerve damage but intact central nervous system to obtain the ability to communicate with the outside world, not only in line with The national policy will also reduce the burden on the society and will surely produce huge economic benefits.

Hybrid brain control technology is a new research direction proposed on the basis of traditional brain-computer interface technology. It adds new physiological electrical signals to the traditional single-paradigm brain-computer interface to obtain a hybrid brain-control system. Among them, the new physiological electrical signals include oculoelectricity, electrocardiogram, blood flow transformation and other signal modes. The hybrid brain control method can be divided into parallel mode and serial mode according to its mixing mode. Among them, the hybrid brain control method based on parallel mode can realize the cooperative control strategy of two kinds of EEG signals, which not only increases the reliability of the system, but also improves the control accuracy of the whole system.

At present, domestic scholars are still using EEG signals under a single paradigm as the control source of intelligent prosthetic hands, and have not conducted in-depth research on the prosthetic hand control technology based on hybrid brain control technology. Therefore, the present invention proposes a parallel multi-modal brain control method for the complete grasping operation of a prosthetic hand according to the difference in the characteristics of EEG signals under the visually induced brain control paradigm of scene animation and the expression driven brain control paradigm. In order to solve the key problems such as the correct rate, information transmission rate, and reliability of the brain-controlled system for the complete grasping operation of the intelligent prosthetic hand.

Contents of the invention

The purpose of the present invention is to overcome the above-mentioned deficiencies in the prior art. On the basis of the existing EEG signal control intelligent prosthetic hand method, the reliability of the brain control paradigm based on scene animation visual induction and expression driving is comprehensively utilized to improve the complete grasping of the prosthetic hand. The accuracy of the operation is the purpose, and a parallel multi-modal brain control method for the complete grasping operation of the prosthetic hand is provided.

To achieve the above object, the present invention is achieved by taking the following technical solutions:

A parallel multi-modal brain control method for a complete grasping operation of a prosthetic hand, comprising the following steps:

(1) Wear the EEG acquisition module for the subject, in which all electrodes are in the standard position of the international 10/20 system, and collect the EEG of the O1 and O2 channels in the occipital lobe area of the cerebral cortex according to the generation mechanism of the visually induced EEG signal in the scene animation According to the expression-driven EEG signal generation mechanism, the EEG signals of the F7 and F8 channels of the lateral frontal cortex were collected.

(2) Embed in the visual eliciting module of the scene animation in advance, which is driven by four different facial expressions at the same time through the simple daily drinking water operation of a disabled person, and four continuous action scene images obtained by decomposition. Among them, the faces of the disabled in each image showed a different expression, and each action scene image was gray-scale processed to obtain two black and white reversed color images, which were alternately presented in front of the subjects for visual induction. The alternating changing frequency of the two black and white reversed color pictures, that is, the flickering frequency is the same as the pulse width modulation frequency of the group of pictures, and the pulse width frequency modulation range is 2-30 Hz.

The four continuous action scene images obtained by decomposition are respectively defined as scene animation one, scene animation two, scene animation three, and scene animation four; the four different expression drives are respectively eyebrow raising, frowning, left Curl your lips, curl your lips to the right. Among them, raised eyebrows appear in scene animation 1; frowns appear in scene animation 2; left-handed mouth appears in scene animation 3; right-handed mouth appears in scene animation 4.

(3) The subjects expressed their control intentions by gazing at different scene animation visual evoked interfaces and driving different expressions at the same time. The EEG acquisition module simultaneously extracts the subject's O1, O2, F7, and F8 four-channel EEG signals for off-line analysis.

(4) First, the EEG acquisition module amplifies and filters the collected EEG signals of four channels O1, O2, F7, and F8, and then wirelessly transmits them to the signal processing module via the Bluetooth module. The signal processing module adopts power spectrum The density method analyzes the frequency spectrum of the four-channel EEG signal, and extracts the energy spectral density value of the alpha frequency band. Calculate the energy spectral density of the alpha frequency bands of the O1 and O2 channel EEG signals generated by the scene animation visual induction and the expression drive at the same time, and the energy spectral density of the alpha frequency bands of the F7 and F8 channel EEG signals generated by the expression drive Specific weight ω.

Secondly, using the parallel control method, according to the characteristics of EEG signals generated under the scene animation steady-state visual evoked and expression-driven brain control paradigms, the time-frequency domain eigenvalues of the four-channel alpha bands are simultaneously extracted through the power spectral density method. Among them, the time-frequency domain feature extraction algorithm of the alpha frequency band of the scene animation visually induced EEG signal is fast Fourier transform; the time-frequency domain feature extraction algorithm of the expression-driven EEG signal alpha frequency band is the wavelet transform modulus mean.

(5) determine the number of training samples of the characteristic EEG signals produced under two different brain control paradigms according to the weight ω calculated in step (4), and input it into the BP neural network classifier for training;

(6) After the sample training is completed, the subject expresses the control intention by gazing at the animated visual eliciting interface of different scenes and driving different expressions at the same time. The EEG acquisition module simultaneously extracts the four-channel EEG of the subject O1, O2, F7, and F8 After the signal, return to step (4), carry out online target recognition, adopt BP neural network to carry out online classification according to the training result of step (5) and the extracted EEG characteristic signal;

(7) The recognition result obtained in step (6) is wirelessly transmitted to the drive control module in the prosthetic arm tube through the communication module, and the drive control module controls the prosthetic hand body module to complete the basic actions.

In the above method, the four kinds of expressions are eyebrow raising, frowning, left-handed mouth, and right-handed mouth, and each expression is repeated at least 3 times each time it is driven. The eyebrow raising and scene animation one controls the opening of the artificial hand; frowning and the scene animation two control the grasping of the artificial hand; the left curling mouth and the scene animation three control the internal rotation of the fake wrist joint; the right curling mouth and the scene animation four control the external rotation of the fake wrist joint.

The present invention aims at controlling the prosthetic hand with the EEG signal under the traditional single paradigm, and its superiority lies in:

Aiming at the brain-controlled prosthetic hand method under a single paradigm, the present invention proposes a parallel multi-modal brain-controlled method for the complete grasping operation of the prosthetic hand. The EEG signals generated at the same time are used as the control information source at the same time. By improving the individual differences between different types of EEG signals, the precision control of the prosthetic hand is realized, and the accuracy and information transmission rate of the brain-controlled prosthetic hand system are further improved.

Description of drawings

FIG. 1 is a schematic diagram of the layout of the EEG acquisition module of the present invention.

Fig. 2 is a schematic diagram of visual induction based on scene animation.

(a) Schematic diagram of prosthetic hand opening mouth and eyebrow raising scene;

(b) Schematic diagram of the prosthetic hand grasping and frowning scene;

(c) Schematic diagram of prosthetic wrist pronation and left-handed mouth;

(d) Schematic diagram of the fake wrist external rotation and right-handed mouth scene.

In each scene, the left side is the action scene; the middle and the right side are two black and white reversed color pictures after grayscale processing;

Fig. 3 is a diagram of the stimulation interface of the scene animation visually evoked by the intelligent prosthetic hand.

Fig. 4 is a schematic diagram of the control system of the present invention.

Wherein (a) is a flow chart of the control method;

(b) is the structural block diagram of the control device.

detailed description

A parallel multimodal brain-controlled prosthetic device based on the complete grasping operation of the prosthetic hand, including: an EEG signal acquisition module worn on the subject's head, a portable signal processing module placed on the subject's waist, and a prosthetic hand driven Control module, prosthetic hand body module, bluetooth transmission module, communication module and scene animation visual eliciting module placed within the visual range of the subject, the scene animation visual eliciting module plays different actions decomposed and processed to the subject A flickering image of the scene to elicit EEG signals from the subject.

In the above scheme, the EEG acquisition module adopts the special portable EEG cap EMOTIV with its own amplification filter, and selects the F7, F8, O1, O2 channel signals under the international standard 10/20. The portable signal processing module adopts an embedded microprocessor. The EEG acquisition module and the signal processing module are wirelessly connected through the Bluetooth transmission module, and the signal processing module and the drive control module of the prosthetic hand are connected wirelessly through the communication module. The scene animation visual eliciting module adopted by the device is one of a computer monitor, a TV screen, a mobile phone or a tablet computer.

Referring to Figure 1 and Figure 4(b), in Figure 1, the O1 and O2 positions in the occipital lobe area of the subject's head and the F7 and F8 positions in the lateral frontal cortex area were collected according to the scene animation visual evoked and expression-driven mechanism For EEG signals, reference electrodes were placed behind the ears on both sides. The prosthetic hand control device involved in the present invention includes an EEG signal acquisition module 310 placed on the subject's head, and a scene animation visual induction module 370 placed within the visual range of the subject. Prioritize the use of portable 16-channel wireless EEG acquisition equipment, and select the EEG signals at the O1 and O2 positions in the occipital lobe area and the F7 and F8 positions in the lateral frontal cortex area under the international standard 10/20. The EEG signal collection module amplifies and filters the collected EEG signals and sends them to the portable signal processing module 330 through the Bluetooth transmission module 320 . The signal processing module is responsible for feature extraction and pattern recognition of EEG signals, and the recognition results are transmitted to the drive control module 350 placed in the arm tube of the prosthetic hand body module 360 through TTL serial communication technology, and the drive control module converts the recognition results into motor control command to complete the prosthetic hand target movement.

Refer to Figure 2. The visual eliciting unit of the scene animation decomposes the drinking process driven by different expressions of the disabled into opening and eyebrow raising of the artificial hand, grasping and frowning of the artificial hand, internal rotation of the artificial wrist joint and left-handed mouth, and external rotation of the artificial wrist joint. Four different target action scenes with right-handed lips, and pre-implanted into the scene animation visual eliciting module;

Scenario 1 (a) The initial state of the artificial hand of the handicapped is opened, and at the same time, the eyebrow-raising expression is driven, and the artificial hand is controlled to complete the opening process. Scenario 2 (b) The handicap's prosthetic hand grasps the cup while driving the frown expression, and controls the prosthetic hand to complete the action process of grabbing the cup. Scene 3 (c) The artificial wrist of the handicapped is turned inward while the expression of left-handed mouth is driven, and the prosthetic hand is controlled to complete the action process of wrist internal rotation; Scene 4 (d) The artificial wrist of the disabled is turned outward while the expression of right-handed mouth is driven, and the artificial hand is controlled to complete the wrist movement process. The action process of external rotation, so as to realize a complete set of water drinking process.

Referring to Figure 4(a), the present invention is a parallel multi-modal brain control method for a complete grasping operation of a prosthetic hand. When the scene animation visual induction and expression drive are performed on the subject at the same time, the EEG acquisition device collects the occipital lobe and The EEG signals in the lateral frontal cortex were amplified and preprocessed with band-pass filtering. The signal processing equipment extracts the power spectral density and frequency-domain eigenvalues of the preprocessed EEG signals, and identifies different basic action types of the prosthetic hand according to the time-frequency domain eigenvalues. Finally, the prosthesis controls the prosthetic hand to complete four basic actions according to the recognition results. Specifically include the following steps:

(1) Step S110, the subject made one of the four simple expressions of eyebrow raising, frowning, left-handed mouth, right-handed mouth while watching the scene animation visual evoked module (Figure 3), and simultaneously collected O1, O2, F7 , F8 position EEG signal. In this embodiment, the four basic movements of the intelligent prosthetic hand are hand opening, hand grasping, wrist joint internal rotation, and wrist joint external rotation.

(2) Step S120, preprocessing the collected EEG signals. In this embodiment, the collected EEG signal is first amplified, and then 2-50 Hz band-pass filtering is performed.

(3) Step S130, extracting time-frequency domain feature values of the EEG signal. According to their own needs, the subjects randomly stared at the desired visual scene animation visual evoked interface and made one of the four corresponding simple expressions, then extracted the EEG signals at O1, O2, F7, and F8 positions, and used power spectral density The method extracts the time-frequency domain eigenvalues of four-channel alpha bands.

$\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; e</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; t</span>}}$

r(k)=E[x(t)x ^* (n+t)](2)

Among them, r(k) is the autocorrelation function of EEG signals of different channels, and x(t) is the EEG signals of different channels;

Calculate the energy spectral density of the alpha frequency bands of the characteristic EEG signals of the O1 and O2 channels generated by the visual induction of the scene animation and the specific weight ω of the energy spectral density of the alpha frequency bands of the characteristic EEG signals of the F7 and F8 channels generated by the expression drive:

$\mathrm{<span\; class="notranslate">\; \&omega;</span>}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; )</span>}{{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; f</span>}\mathrm{<span\; class="notranslate">\; 7</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; f</span>}\mathrm{<span\; class="notranslate">\; 8</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

Among them, PO1(i) and PO2(i) represent the energy spectral densities corresponding to the alpha bands of the O1 and O2 channel EEG signals generated by the visual evoked scene animation, respectively, and PF7(i) and PF8(i) represent the energy spectral density corresponding to the expression-driven The energy spectral density of the alpha band of the generated EEG signals of F7 and F8 channels.

After calculating the weight ω, calculate the alpha frequency band time-frequency domain eigenvalues of the four-channel EEG signals respectively. In this embodiment, the fast Fourier transform method is used to calculate the O1 and O2 channel alpha frequency bands generated by the visual induction of the scene animation. The frequency domain eigenvalues of the F7 and F8 channel alpha frequency band time-frequency domain characteristics generated by the expression drive are calculated by using the wavelet transform modulus mean modulo method. In addition to the fast Fourier transform and wavelet transform used in this embodiment, feature extraction methods such as principal component analysis and common-space mode can also be used for extracting feature values.

(4) Step S140, according to the energy proportion weight ω of the alpha frequency band under four different control sources, determine the number of samples under each channel in the classifier training samples, and carry out the training of the classifier. During the training process, identify the corresponding The basic movement types of the prosthetic hand. In this embodiment, the BP neural network is used to perform pattern recognition on the eigenvalues. According to the four types of basic action types of the artificial hand that are trained after weight calculation, that is, before the subject formally uses the intelligent artificial hand, he needs to use four parallel multi- The time-frequency domain eigenvalues of the modal EEG signals trained the BP neural network classifier, and the output result of the classifier controlled the prosthetic hand to complete four basic actions. The recognition results of the four basic actions are shown in Table 1.

Table 1 EEG recognition results

(5) Step S150, the drive control module in the prosthetic arm tube completes four basic actions according to the recognition results.

(6) Step S160, after the prosthetic hand body module completes the four basic movements, feedback can be realized through visual information and biological perception.

Referring to Fig. 4(b), based on the method in Fig. 4(a), the present invention provides a corresponding control device, including: EEG signal acquisition module 310, Bluetooth transmission module 320, signal processing module 330, communication module 340, drive Control module 350 , prosthetic hand body module 360 , and scene animation visual induction module 370 .

In this device, the electroencephalogram signal collection module 310 collects the characteristic electroencephalogram signals driven by expressions under the international standard 10/20, and selects the positions behind the ears on both sides as reference positions. The portable signal processing module 330 adopts an embedded microprocessor. The EEG acquisition module is connected with the signal processing module through the Bluetooth transmission module. The signal processing module is connected with the driving control module of the prosthetic hand through the communication module. The scene animation visual eliciting module adopted by the device is one of a computer monitor, a TV screen, a mobile phone or a tablet computer.

When the scene animation visual eliciting module 370 is working, the EEG acquisition module 310 is used to collect EEG signals at O1, O2, F7, and F8 positions. The 16-channel wireless EEG cap Emotiv is selected, and its own software is used for amplification and filtering. The Bluetooth transmission module 320 is connected with the signal processing module 330, and transmits the EEG signals collected by the EEG collection module to the signal processing module. The signal processing module 330 receives the EEG signal transmitted from the Bluetooth transmission module 320 to perform feature extraction and pattern recognition. Fast Fourier transform and wavelet transform modulus mean method are used to extract eigenvalue vectors, and BP neural network method is used to identify four basic movements of the prosthetic hand. The signal processing module can be selected as an embedded microprocessor that is carried around. The communication module 340 is composed of a generating terminal module and a receiving terminal module, and completes the initial setting of the transmitting terminal and the receiving terminal through AT commands. Wherein, the transmitting end is connected with the signal processing module 330 , and the receiving end is connected with the drive control module 350 of the prosthetic hand. The transmitting terminal module transmits the recognition result to the driving control module of the prosthetic hand through the receiving terminal module through the TTL serial port communication. The drive control module 350 of the prosthetic hand is composed of a motor control sub-module and a motor drive sub-module. Receive the recognition result of the basic motion type of the prosthetic hand transmitted via the communication module. The recognition result is transmitted to the motor control sub-module and then converted into a 0,1 level control instruction and transmitted to the motor drive sub-module. The motor drive sub-module controls the prosthetic hand body module 360 to complete four basic actions according to the level control command. If the recognition result is the grasping action of the artificial hand, the drive control module drives the artificial hand to complete the grasping action.

The prosthetic hand body module 360 drives different motors to rotate according to the action commands sent by the drive control module 350, and finally realizes four basic actions of the prosthetic hand.

The above-mentioned embodiments are only to illustrate the technical concept and characteristics of the present invention, and its purpose is to enable those familiar with this technology to understand the content of the present invention and implement it accordingly, and cannot limit the protection scope of the present invention.

Claims (2)

1. A parallel multimodal brain control method for the complete grasping operation of a prosthetic hand, characterized in that it comprises the following steps: (1) Wear the EEG acquisition module for the subject, in which all electrodes are in the standard position of the international 10/20 system, and collect the EEG of the O1 and O2 channels in the occipital lobe area of the cerebral cortex according to the generation mechanism of the visually induced EEG signal in the scene animation Signal, collect the EEG signals of the F7 and F8 channels of the lateral frontal cortex according to the expression-driven EEG signal generation mechanism; (2) Implanting in the scene animation visual eliciting module in advance, the whole process driven by a disabled person's daily drinking water simple operation and four different facial expressions at the same time, decomposed and obtained four continuous action scene images, each of which In the image, the face of the disabled presents a different expression. After each action scene image is processed in grayscale, two black and white reversed color pictures are obtained, which are alternately presented in front of the subject for visual evoking. The two pictures The alternating frequency of black and white reversed color pictures, that is, the flickering frequency is the same as the pulse width modulation frequency of the group of pictures, and the pulse width frequency modulation range is 2-30Hz; The four continuous action scene images obtained by decomposition are respectively defined as scene animation one, scene animation two, scene animation three, and scene animation four; Right-handed lips, among them, raised eyebrows appear in scene animation 1; frowns appear in scene animation 2; left-handed lips appear in scene animation 3; right-handed lips appear in scene animation 4; (3) The subject expresses the control intention by gazing at the animated visual evoked interface of different scenes and driving different expressions at the same time. The EEG acquisition module simultaneously extracts the subject's O1, O2, F7, and F8 four-channel EEG signals for offline analysis ; (4) First, the EEG acquisition module amplifies and filters the collected EEG signals of four channels O1, O2, F7, and F8, and then wirelessly transmits them to the signal processing module via the Bluetooth module. The signal processing module adopts power spectrum The density method analyzes the frequency spectrum of the four-channel EEG signal, extracts the energy spectral density value of the alpha frequency band, and calculates the energy of the alpha frequency band of the O1 and O2 channel EEG signals generated by the scene animation visual induction and expression drive at the same time. The spectral density value and the specific weight ω of the energy spectral density value of the alpha frequency band of the F7 and F8 channel EEG signals driven by expressions; Secondly, using the parallel control method, according to the characteristics of EEG signals generated under the scene animation steady-state visual evoked and expression-driven brain control paradigms, the time-frequency domain feature values of the four-channel alpha frequency bands are simultaneously extracted by the power spectral density method. Among them, the scene animation The time-frequency domain feature extraction algorithm for the alpha frequency band of the visually induced EEG signal is Fast Fourier Transform; the time-frequency domain feature extraction algorithm for the alpha frequency band of the expression-driven EEG signal is the wavelet transform modulus mean; (5) determine the number of training samples of the characteristic EEG signals produced under two different brain control paradigms according to the weight ω calculated in step (4), and input it into the BP neural network classifier for training; (6) After the sample training is completed, the subject expresses the control intention by gazing at the animated visual eliciting interface of different scenes and driving different expressions at the same time. The EEG acquisition module simultaneously extracts the four-channel EEG of the subject O1, O2, F7, and F8 After the signal, return to step (4), carry out online target recognition, adopt BP neural network to carry out online classification according to the training result of step (5) and the extracted EEG characteristic signal; (7) The recognition result obtained in step (6) is wirelessly transmitted to the drive control module in the prosthetic arm tube through the communication module, and the drive control module controls the prosthetic hand body module to complete the basic actions. 2. A parallel multi-modal brain control method for a complete grasping operation of a prosthetic hand according to claim 1, wherein the four facial expressions are eyebrow raising, frowning, left-handed mouth, right-handed mouth, and each time actuation Each expression is repeated at least three times, the eyebrow raising and scene animation 1 controls the opening of the artificial hand; frowning and scene animation 2 controls the grasping of the fake hand; left-handed mouth and scene animation 3 control the internal rotation of the fake wrist joint; Control external rotation of the prosthetic wrist joint.