CN109765823A - Ground crawler-type unmanned vehicle control method based on arm electromyography signal - Google Patents
Ground crawler-type unmanned vehicle control method based on arm electromyography signal Download PDFInfo
- Publication number
- CN109765823A CN109765823A CN201910051882.7A CN201910051882A CN109765823A CN 109765823 A CN109765823 A CN 109765823A CN 201910051882 A CN201910051882 A CN 201910051882A CN 109765823 A CN109765823 A CN 109765823A
- Authority
- CN
- China
- Prior art keywords
- signal
- electromyography signal
- unmanned vehicle
- crawler
- type unmanned
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The ground crawler-type unmanned vehicle control method based on arm electromyography signal that the invention discloses a kind of, to overcome the problems, such as shortage flexibility of the existing technology, equipment complexity, vulnerable to external environmental interference, its step: 1) correctly wear myoelectric sensor and carry out Initialize installation: (1) user is worn over myoelectric sensor in the middle position of hand and elbow, electrode is set to be close to skin, it is attached to reference electrode on musculus flexor carpi ulnaris, it is ensured that electromyography signal can be collected；(2) it initializes, first makes myoelectric sensor and crawler-type unmanned vehicle Initialize installation, i.e., test whether that establishing reliable wireless communication connect with crawler-type unmanned vehicle by signal processing apparatus in certain region；2) acquisition electromyography signal identifies different gestures and is sent to crawler-type unmanned vehicle；3) crawler-type unmanned vehicle resolve command and the movement of the order is executed；4) content of video camera shooting is passed back signal processing apparatus, and be shown in computer by wireless transmission.
Description
Technical field
The present invention relates to a kind of control methods of crawler-type unmanned vehicle, and more precisely, the present invention relates to one kind to be based on hand
The ground crawler-type unmanned vehicle control method of arm electromyography signal.
Background technique
With the continuous development of science and technology with innovation, the human-machine interface technology based on human biological signal is people
One of research hotspots in fields such as machine interaction, robot control.In addition, as artificial intelligence field achieves great achievement,
Artificial intelligence is widely used in all trades and professions, and especially military ground crawler-type unmanned vehicle is quickly grown, and is battle field information
Change and develops the condition that brings great convenience.
But existing technology is generally by computer operating whole system, realize programme path, avoiding barrier, from
The functions such as dynamic traveling.It perceives the environment around car body by fuselage sensing system, hides the barrier and vehicle on road surface
With pedestrian etc., but it is applied to lack certain flexibility when military field is fought.
In addition, though gesture identification based on computer vision is the research side of the mainstream of domestic and international gesture identification at present
To, but the technology is highly prone to illumination and the influence of position, and due to more complex equipment limit its in gesture interaction side
The period of expansion in face.The present invention controls ground crawler-type unmanned vehicle by electromyography signal, increase one it is convenient, fast, flexible
Controlling soil moist.
Summary of the invention
The technical problem to be solved by the present invention is to overcome shortage flexibility of the existing technology, equipment complexity,
The problem of vulnerable to external environmental interference, provides a kind of ground crawler-type unmanned vehicle control method based on arm electromyography signal.
In order to solve the above technical problems, the present invention is achieved by the following technical scheme: described based on arm myoelectricity
The ground crawler-type unmanned vehicle control method of signal includes:
1) it correctly wears myoelectric sensor and carries out Initialize installation；
(1) myoelectric sensor is worn over the position among hand and elbow by user, and electrode is made to be close to skin, while being made with reference to electricity
Pole is attached on musculus flexor carpi ulnaris, it is ensured that can collect electromyography signal；
(2) it initializes, first makes myoelectric sensor and crawler-type unmanned vehicle Initialize installation, i.e., in certain area
It is communicated in domain by the first signal processing apparatus with the single-chip microcontroller on crawler-type unmanned vehicle to test whether to establish reliably
Wireless communication connection；
2) acquisition electromyography signal identifies different gestures and is sent to crawler-type unmanned vehicle；
3) crawler-type unmanned vehicle resolve command and the movement of the order is executed；
4) content by video camera shooting passes second time signal processing apparatus by wireless transmission, and is shown in computer and works as
In.
Acquisition electromyography signal described in technical solution, which identifies different gestures and is sent to crawler-type unmanned vehicle, to be referred to: 1) being adopted
Collect the electromyography signal of user；
A. electromyography signal is acquired by myoelectric sensor and reference electrode first；
B. the data that the collected data of myoelectric sensor and reference electrode are collected into are sent to the first signal processing device
In setting；
2) the first signal processing apparatus carries out gesture identification and sends.
First signal processing apparatus described in technical solution carries out gesture identification and sends to refer to:
A. judge whether collected signal is useful signal, if the signal of acquisition is invalid electromyography signal, then just
Continue to acquire, until electromyographic signal collection success；
Starting point threshold value f is set thusHWith stop threshold value fL, it is made to remain fH>fLState, only when myoelectricity believe
Number be greater than fHWhen, it is judged as gesture and starts to execute, effective electromyography signal can be acquired；
B. a series of signal processing is passed through to collected electromyography signal；
C. the feature for extracting signal is to carry out feature extraction to the signal of active segment to obtain the feature of discrimination to be promoted
The redundancy of signal；
D. classifier carries out the classification of motion, judges whether to can recognize that electromyography signal, if being unable to identify that myoelectricity
Signal, then acquisition will be re-started；If can recognize that electromyography signal, which is converted into definition of gesture
Movement number；
E. movement number is sent it on crawler-type unmanned vehicle by 433MHz communication；
F. judge whether electromyography signal changes, so re-recognize the electromyography signal if changing, and repeat step
2) i.e. the process that the first signal processing apparatus (3-1) carries out gesture identification and sends, if no change has taken place, continuing will
The gesture generates the number for being converted to definition, sends by wirelessly；In case of changing, then will just change turning for gesture
Corresponding definition number is changed to send.
Collected electromyography signal is referred to by a series of signal processing described in technical solution:
Signal is carried out to carry out denoising before feature extraction, by using small echo μ rule threshold function table to electromyography signal
Radio-frequency component carry out de-noising；μ restrains threshold function table are as follows:
dr=sign (| w |-α τ), | w | > τ
Wherein, w indicates coefficient of wavelet decomposition；The key step that small echo μ restrains threshold denoising is as follows:
(1) WAVELET PACKET DECOMPOSITION is carried out to electromyography signal；
(2) threshold process is carried out to the high frequency coefficient decomposed；
(3) low frequency coefficient and the high frequency coefficient after threshold process are subjected to wavelet reconstruction, the original letter after obtaining de-noising
Number.
The feature that signal is extracted described in technical solution is to carry out feature extraction to the signal of active segment to have obtained differentiation
The redundancy that the feature of degree carrys out promotion signal refers to:
The extraction of the electromyography signal feature is by selecting the absolute value mean value (MAV) of time domain, standard deviation (SD),
Root (RMS), the frequency of average power (MPF) of frequency domain and median frequency (MF) feature of totally 5 parameters as electromyography signal:
In formula: XiFor the electromyography signal after interception, Psd (f) indicates the power spectrum of electromyography signal.
It is to establish optimal classification on the basis of off-line data that classifier described in technical solution, which carries out the classification of motion,
Device model, and the disaggregated model is applied in test set and is verified, optimal classification effect is obtained, the model then used
Parameter carries out processing classification in real time to online data；
The off-line data is 10 people's electromyography signals of acquisition, and acquires individual electromyography signal 300 times, and mark gesture
Label is that the movement of the gesture is numbered；
(1) individual electromyography signal, i.e., the signal type of 5 kinds different gestures, to corresponding to each gesture are acquired respectively first
Acquisition training set data put on 1~No. 5 class label respectively；
(2) 5 category features for extracting 5 kinds of hand signals are the absolute value mean value (MAV) of time domain, standard deviation (SD), respectively
The frequency of average power (MPF) and median frequency (MF) of root (RMS), frequency domain constitute electromyography signal sample；
(3) it will be input in SVM classifier and classify after Fusion Features；
(4) ten foldings cross-check to obtain the optimal sorter model of classifier classifying quality；
The electromyography signal sample is that the electromyography signal of acquisition arm surface is pre-processed, filtered and extracted myoelectricity letter
Number characteristic, composition characteristic vector classifies for SVM；
Described being input in SVM classifier after Fusion Features is classified, and the execution method tested is as follows
It is shown:
The sample of some classification is successively classified as one kind when training, other remaining samples are classified as another kind of, such 5 classes
Other sample has just constructed 5 SVM.Unknown sample is classified as that class with maximum classification function value when classification, to say
It is bright facilitate a bent wrist, clench fist, stretch the palm, shooting, sky are pinched five movement class data and are replaced respectively with A, B, C, D, E；
(1) vector corresponding to A collects as positive, B, C, and vector corresponding to D, E is carried out as negative collection using the training set
Training obtains training result f1 (x)；Vector corresponding to B collects as positive, A, C, and vector corresponding to D, E is used as negative collection
The training set is trained to obtain training result f2 (x)；Vector corresponding to C collects as positive, A, B, and vector corresponding to D, E is made
Be negative collection, is trained to obtain training result f3 (x) using the training set；Vector corresponding to D is as positive collection, A, B, C, E institute
Corresponding vector is trained to obtain training result f4 (x) as negative collection using the training set；Vector corresponding to E is as just
Collect, A, B, vector corresponding to C, D is trained to obtain training result f5 (x) as negative collection using the training set；
(2) when test, corresponding test vector is utilized respectively this five training results and is tested.It obtains every
The result type f1 (x) of a test, f2 (x), f3 (x), f4 (x), f5 (x)；
(3) choose five value in maximum one be used as classification results；
The described ten foldings crosscheck, which refers to, is divided into 10 parts for data set, in turn will wherein 9 parts be used as training data, 1 part
As test data, tested.
The identification electromyography signal is according to when carrying out different gesture identifications, and the contribution degree of each muscle group is different,
Therefore characteristic value also can be different；Classified by classifier, and judges to belong to a certain classification, and then identify specific hand
Gesture；
Table 1
| Movement number | 01 | 02 | 03 | 04 | 05 |
| Gesture | Bent wrist | It clenches fist | Stretch the palm | Shooting | Sky is pinched |
| Function | Stop | Advance | It retreats | Turn left | It turns right |
The definition of gesture is as follows: controlling crawler-type unmanned vehicle by 5 manipulation instructions；Table 1 is gesture and meaning
Corresponding relationship is controlled unmanned vehicle using gesture and carries out displacement movement.
Judge whether electromyography signal changes the method referred to through sliding window energy threshold described in technical solution
Active segment detection is carried out to the surface electromyogram signal of gesture motion, specific algorithm is as follows:
1. the absolute value for calculating the electromyography signal in all channels it is average and；
2. ask the absolute value of electromyography signal in window to be averaged the quadratic sum of sum using the rectangular slide window of a length of N of window,
It is f respectively that beginning and end threshold value, which is arranged,HAnd fL(fH>FL), when electromyography signal is greater than starting point threshold value, determine hand
Gesture starts；When electromyography signal instantaneous energy all in 100ms is less than terminal threshold value, determine that gesture motion terminates；
3. starting detection gesture movement if electromyography signal is greater than the threshold value of starting point in active segment after gesture terminates
Whether change.
Crawler-type unmanned vehicle resolve command described in the technical solution and movement for executing the order refers to:
1) single-chip microcontroller carries out wireless communication, singlechip interruption initializes；
2) whether single-chip microcontroller is sky by detecting flag bit constantly come decision instruction queue, if it is non-empty, then carrying out
Instruction fetch；
A. a byte is taken out from queue heads, judges whether the byte is one of five numbers of definition, if it is not, that
It is abandoned, because may be what interference generated；If it is one among five numbers, then being deposited into caching array
Etc. pending；
B. judge whether queue is sky, if non-empty, continue the process for repeating above-mentioned instruction fetch；If it is sky, that
Continuing following 3) step judges that caching array instructs and carries out various operations；
3) judgement caching array instructs and is moved ahead, retreated, turned left, right by single-chip microcontroller control crawler-type unmanned vehicle
The operation for turning, stopping；
Described is that content judgement is defined by above-mentioned table 1 by single-chip microcontroller control crawler-type unmanned vehicle, as received
When number is 2, just know that user uses the movement clenched fist, therefore crawler-type unmanned vehicle is promoted come driving motor by single-chip microcontroller
Advance.
By the content of video camera shooting by passing second signal processing unit back described in technical solution, and
It is shown in computer and refers to:
1) video camera starts to be shot, and acquires image information；
2) acquired image is digitized, obtains H.264 data flow, and send in such a way that 433MHz is wirelessly communicated
To second signal processing unit；
A. coding parameter is initialized first, and obtains input-buffer address and output buffer address；
B. the data of a frame are passed to, to the Image Acquisition carrier chrominance signal, the data of acquisition are encoded；
C. by the image taken using 16*16 as macro block, the pixel value of each macro block is calculated；
D. the pixel value for then calculating each macro block of the frame copies code stream head and generates message queue；
E. mode of the message queue in crawler-type unmanned vehicle by 433MHz wireless communication passes image information back second
Signal processing apparatus；
3) after second signal processing unit receives message queue, the data flow received is divided into two-way, is sealed all the way
Dress turns and the storage of FLV file is written, and another way carries out computer video live streaming according to the broadcasting process of current shooting；
If 4) computer issues the request of live streaming, it is sent to after data will be obtained using RTMP protocol packing data
The player of client；If computer issues the instruction of playback, according to the period of playback, corresponding document, solution are read
Analyse data flow.
Compared with prior art the beneficial effects of the present invention are:
1. using myoelectricity in the ground crawler-type unmanned vehicle control method of the present invention based on arm electromyography signal
Sensor gets rid of traditional electrode and is attached on fixed acupuncture point the drawbacks of acquiring electromyography signal, only a reference electrode is needed to be affixed on
At musculus flexor carpi ulnaris, this avoid electrode acquisition swingings of signal, inconvenient for use etc. to influence；
2. the ground crawler-type unmanned vehicle control method of the present invention based on arm electromyography signal is in off-line data
On the basis of, optimal sorter model is established, the model parameter then used carries out processing point in real time to online data
The mode of class, since the electromyography signal for different people Different Individual can slightly have difference, Classification and Identification is can be improved in this method
Accuracy rate, so as to the more acurrate intention for identifying user.
3. the ground crawler-type unmanned vehicle control method of the present invention based on arm electromyography signal is through electromyography signal
Differentiate, the displacement that ground crawler-type unmanned vehicle is controlled by gesture is mobile, compared to the prior art there is flexibility, conveniently
Property advantage, and identify cost is relatively low, it is insensitive to illumination and position, be conducive in Military Application using be swift in response,
Fast, low feature is interfered.
Detailed description of the invention
The present invention will be further described below with reference to the drawings:
Fig. 1 is the schematic block diagram of the ground crawler-type unmanned vehicle control of the present invention based on arm electromyography signal；
Fig. 2 is the flow chart element of the ground crawler-type unmanned vehicle control method of the present invention based on arm electromyography signal
Figure；
Fig. 3 is the myoelectricity letter in the ground crawler-type unmanned vehicle control method of the present invention based on arm electromyography signal
The flow diagram of number collecting work；
Fig. 4 be the ground crawler-type unmanned vehicle control method of the present invention based on arm electromyography signal in foundation most
The method flow block diagram of excellent sorter model；
Fig. 5 is that the ground in the ground crawler-type unmanned vehicle control method of the present invention based on arm electromyography signal is carried out
The flow diagram of belt unmanned vehicle reception command analysis；
Fig. 6 is the instruction fetch in the ground crawler-type unmanned vehicle control method of the present invention based on arm electromyography signal
Process flow diagram flow chart；
Fig. 7 is that the ground in the ground crawler-type unmanned vehicle control method of the present invention based on arm electromyography signal is carried out
Belt unmanned vehicle video information sends, handles and show to the schematic block diagram of computer screen；
Fig. 8 is the passback view in the ground crawler-type unmanned vehicle control method of the present invention based on arm electromyography signal
Frequency information process block diagram；
In figure: 1. myoelectric sensors, 2. reference electrodes, 3. signal processing apparatus, the first signal processing apparatus of 3-1., 3-2.
Second signal processing unit, 4. computer
Specific embodiment
The present invention is explained in detail with reference to the accompanying drawing:
Refering to fig. 1, to realize the ground crawler-type unmanned vehicle control method based on arm electromyography signal, the present invention provides
A kind of ground crawler-type unmanned vehicle control based on arm electromyography signal, the ground based on arm electromyography signal are carried out
The unmanned vehicle control of belt includes myoelectric sensor 1, reference electrode 2, the first signal processing apparatus (3- for acquiring electromyography signal
1), second signal processing unit (3-2), the single-chip minimum system being mounted on crawler-type unmanned vehicle, camera and computer,
Wherein: the first signal processing apparatus (3-1) is identical as the structure of second signal processing unit (3-2).
The myoelectric sensor 1 of the acquisition electromyography signal is made of the myoelectric sensor of eight tunnel medical grades, the eight roads flesh
Electric transducer is sequentially connected end to end by retractable belt, is all fixed on plastic shell center per myoelectric sensor all the way
Position on, plastic shell is 2 centimetres long, and width is 0.6 centimetre, the angle with a thickness of 0.3 centimetre, between two neighboring plastic shell
It is 150 degree；
The reference electrode 2 uses silver/silver chlorate surface electrode, when electric current passes through electrode electrolyte interface, electrode
Characteristic is similar to unpolarized electrode, and signal transduction effect is preferable, while the very strong conducting resinl of toughness below patch, can incite somebody to action
Patch, which is fixed on muscle, prevents patch movement from interfering；
First signal processing apparatus (3-1), second signal processing unit (3-2) are by model
Main control chip, minimum system circuit and the basic peripheral circuit composition of STM32F407VET6.
The main control chip of the model STM32F407VET6 is the ARM Cortex M4 kernel with FPU, can be with
Complete the data processing to collected electromyography signal.
Minimum system in first signal processing apparatus (3-1), second signal processing unit (3-2) includes 26M
Crystal oscillating circuit, RST reset key circuit.
The basic peripheral circuit refers to E10-433MS1W module, LNA low-noise amplifier, wherein E10-
433MS1W module use SI4463 radio frequency chip, strong interference immunity, performance stabilization, long transmission distance, penetrate diffracting power by force
Feature；The effect of LNA low-noise amplifier is to improve receiving sensitivity.
The single-chip minimum system being mounted on crawler-type unmanned vehicle is by the master control of model STC15W408S
Chip and basic peripheral circuit are constituted.
Minimum system on the crawler-type unmanned vehicle further includes 11.0592MHz crystal oscillating circuit, RST reset key electricity
Road.
The basic peripheral circuit includes the triode Q9012 for being connected to single-chip microcontroller P1.5 pin and P5.4 pin, and two
Pole pipe IN4148, electromagnetic relay and E10-433MS1W module, LNA low-noise amplifier.The wherein work of diode IN4148
With being to protect triode not breakdown, the effect of electromagnetic relay is with the rotation of 220V current control alternating current generator.
The camera is the camera of Haikang prestige view model DS-2CD1301-L, and being equipped with POE power supply cable is 8 cores
0.52 oxygen-free copper.
The computer uses model DESKTOP-JN54QV0 Dell laptop, and use makes
Windows10 operating system；
The myoelectric sensor 1, reference electrode 2 use wired connection with the first signal processing apparatus 3-1 respectively, and first
Antenna on the antenna and crawler-type unmanned vehicle of signal processing apparatus 3-1, which uses, to be wirelessly connected, the antenna on crawler-type unmanned vehicle
With second signal processing unit 3-2 using being wirelessly connected, second signal processing unit 3-2 and computer 4 use the wired company of USB
It connects.
Myoelectric sensor 1, reference electrode 2 and the first signal processing apparatus 3-1 refers to acquisition using wired connection
The electrical signal line of myoelectric sensor of eight medical grades of electromyography signal is connected with PE0~PE7 of STM32F407VET6；It is attached to
Patch reference electrode 2 on musculus flexor carpi ulnaris is connected with the PE8 of STM32F407VET6.
The step of described ground crawler-type unmanned vehicle control method based on arm electromyography signal, is as follows:
1. correctly wearing myoelectric sensor and carrying out Initialize installation
1) user correctly wears myoelectric sensor 1, and myoelectric sensor 1 is worn over the position among hand and elbow, makes electrode
It is close to skin, while is attached to reference electrode 2 on musculus flexor carpi ulnaris, it is ensured that electromyographic signal collection success；
Described ensures that electromyographic signal collection successfully refers to: since its feature has regularity, low frequency characteristic, for same
When one gesture carries out the same electromyographic signal collection, it can be observed that fluctuation of the electromyography signal in regularity, amplitude-frequency characteristic
It can be judged as basic identical, and the most energy of human body surface myoelectric signal concentrates between 10Hz~300Hz, makes
Myoelectric sensor 1, reference electrode 2 capture the variation of real-time electromyography signal.
2) it initializes, i.e. myoelectric sensor 1 and crawler-type unmanned vehicle Initialize installation, then in certain area
It is communicated with the single-chip microcontroller on crawler-type unmanned vehicle by the first signal processing apparatus 3-1 in domain come test whether to establish can
It is connected by wireless communication；
The communication is the communication using 433MHz, since 433MHz wireless communication can pass through
The directional aerial for matching high benefit, adjusts tranmitting frequency appropriate, can significantly improve the power density on communication direction, from
And improve communication distance.The present invention is remote come what is realized by the higher generation power of directional aerial simultaneous selection for selecting high gain
Distance communication；
2. acquisition electromyography signal identifies different gestures and is sent to crawler-type unmanned vehicle refering to Fig. 3
1) electromyography signal of user is acquired
A. electromyography signal is acquired by myoelectric sensor 1 and reference electrode 2 first；
B. the data that the collected data of myoelectric sensor 1 and reference electrode 2 are collected into are sent to the first signal processing
In device 3-1；
2) the first signal processing apparatus (3-1) carries out gesture identification and sends
A. judge whether collected signal is useful signal, if the signal of acquisition is invalid electromyography signal, then just
Continue to acquire, until electromyographic signal collection success；
The useful signal makes the signal for referring to collected signal regularity, low frequency characteristic, is if collecting signal
Irregularities have high frequency characteristics, are judged as invalid signal at this time；Starting point threshold value f is set thusHWith stop threshold
Value FL, it is made to remain fH>FLState, prevent in this way it is unconscious shake be mistaken to effective gesture, only when myoelectricity believe
Number be greater than fHWhen, it is judged as gesture and starts to execute, effective electromyography signal can be acquired.
B. a series of signal processing is passed through to collected electromyography signal
The a series of signal processing refers to due to can be inevitably by free surrounding space in collection process
The interference of ground electromagnetic interference, the interference that recorder generates and subject itself generated, these interference will affect subsequent identification
As a result, it is therefore desirable to carry out denoising before carrying out feature extraction to signal；By using small echo μ rule threshold function table to flesh
The radio-frequency component of electric signal carries out de-noising；μ restrains threshold function table are as follows:
dr=sign (| w |-α τ), | w | > τ
Wherein, w indicates coefficient of wavelet decomposition；The key step that small echo μ restrains threshold denoising is as follows:
(1) WAVELET PACKET DECOMPOSITION is carried out to electromyography signal；
(2) threshold process is carried out to the high frequency coefficient decomposed；
(3) low frequency coefficient and the high frequency coefficient after threshold process are subjected to wavelet reconstruction, the original letter after obtaining de-noising
Number；
C. the feature for extracting signal is to carry out feature extraction to the signal of active segment to obtain the feature of discrimination to be promoted
The redundancy of signal
The extraction of the electromyography signal feature is by selecting the absolute value mean value (MAV) of time domain, standard deviation (SD),
Root (RMS), the frequency of average power (MPF) of frequency domain and median frequency (MF) feature of totally 5 parameters as electromyography signal:
In formula: XiFor the electromyography signal after interception, Psd (f) indicates the power spectrum of electromyography signal；
D. classifier carries out the classification of motion, judges whether to can recognize that electromyography signal, if being unable to identify that myoelectricity
Signal, then will carry out re-starting acquisition；If can recognize that electromyography signal, which is converted into gesture
The movement of definition is numbered；
The classifier carry out the classification of motion be establish optimal sorter model on the basis of off-line data, and
The disaggregated model is applied in test set and is verified, optimal classification effect is obtained, the model parameter then used to
Line number is classified according to processing in real time is carried out.
The off-line data is 10 people's electromyography signals of acquisition, and acquires individual electromyography signal 300 times, and mark gesture
Label is that the movement of the gesture is numbered.
Refering to Fig. 4, the method process for establishing optimum classifier model is as follows:
(1) individual electromyography signal, i.e., the signal type of 5 kinds different gestures, to corresponding to each gesture are acquired respectively first
Acquisition training set data put on 1~No. 5 class label respectively；
(2) 5 category features for extracting 5 kinds of hand signals are the absolute value mean value (MAV) of time domain, standard deviation (SD), respectively
The frequency of average power (MPF) and median frequency (MF) of root (RMS), frequency domain constitute electromyography signal sample；
(3) it will be input in SVM classifier and classify after Fusion Features；
(4) ten foldings cross-check to obtain the optimal sorter model of classifier classifying quality；
The electromyography signal sample is that the electromyography signal of acquisition arm surface is pre-processed, filtered and extracted myoelectricity letter
Number characteristic, composition characteristic vector classifies for SVM；
It is input in SVM classifier and classifies after the Fusion Features, and the following institute of the method for inspection tested
Show:
The sample of some classification is successively classified as one kind when training, other remaining samples are classified as another kind of, such 5 classes
Other sample has just constructed 5 SVM.Unknown sample is classified as that class with maximum classification function value when classification.(to say
It is bright facilitate a bent wrist, clench fist, stretch the palm, shooting, sky are pinched five movement class data and are replaced respectively with A, B, C, D, E)
(1) vector corresponding to A collects as positive, B, C, and vector corresponding to D, E is carried out as negative collection using the training set
Training obtains training result f1 (x)；Vector corresponding to B collects as positive, A, C, and vector corresponding to D, E is used as negative collection
The training set is trained to obtain training result f2 (x)；Vector corresponding to C collects as positive, A, B, and vector corresponding to D, E is made
Be negative collection, is trained to obtain training result f3 (x) using the training set；Vector corresponding to D is as positive collection, A, B, C, E institute
Corresponding vector is trained to obtain training result f4 (x) as negative collection using the training set；Vector corresponding to E is as just
Collect, A, B, vector corresponding to C, D is trained to obtain training result f5 (x) as negative collection using the training set；
(2) when test, corresponding test vector is utilized respectively this five training results and is tested.It obtains every
The result type f1 (x) of a test, f2 (x), f3 (x), f4 (x), f5 (x)；
(3) choose five value in maximum one be used as classification results；
The described ten foldings crosscheck, which refers to, is divided into 10 parts for data set, in turn will wherein 9 parts be used as training data, 1 part
As test data, tested.
The identification electromyography signal is according to when carrying out different gesture identifications, and the contribution degree of each muscle group is different,
Therefore characteristic value also can be different；Classified by classifier, and judges to belong to a certain classification, and then identify specific hand
Gesture；
The definition of gesture is as follows: controlling crawler-type unmanned vehicle by 5 manipulation instructions；Table 1 is gesture and meaning
Corresponding relationship is controlled unmanned vehicle using gesture and carries out displacement movement；
Table 1
| Movement number | 01 | 02 | 03 | 04 | 05 |
| Gesture | Bent wrist | It clenches fist | Stretch the palm | Shooting | Sky is pinched |
| Function | Stop | Advance | It retreats | Turn left | It turns right |
E. movement number is sent it on crawler-type unmanned vehicle by 433MHz communication；
F. judge whether electromyography signal changes, so re-recognize the electromyography signal if changing, and repeat step
2) the first signal processing apparatus (3-1) carries out gesture identification and transmission process, if no change has taken place, continues the hand
Gesture generates the number for being converted to definition, sends by wirelessly；In case of changing, then will just change being converted to for gesture
Corresponding definition number is sent；
It is described to judge whether electromyography signal changes and refer to through the method for sliding window energy threshold to gesture motion
Surface electromyogram signal carry out active segment detection, specific algorithm is as follows:
4. the absolute value for calculating the electromyography signal in all channels it is average and；
5. ask the absolute value of electromyography signal in window to be averaged the quadratic sum of sum using the rectangular slide window of a length of N of window,
It is f respectively that beginning and end threshold value, which is arranged,HAnd fL(fH>fL), when electromyography signal is greater than starting point threshold value, determine hand
Gesture starts；When electromyography signal instantaneous energy all in 100ms is less than terminal threshold value, determine that gesture motion terminates.
6. starting detection gesture movement if electromyography signal is greater than the threshold value of starting point in active segment after gesture terminates
Whether change；
3. refering to Fig. 5, crawler-type unmanned vehicle resolve command and the movement for executing the order
1) single-chip microcontroller carries out wireless communication, singlechip interruption initializes；
2) whether single-chip microcontroller is sky by detecting flag bit constantly come decision instruction queue, if it is non-empty, then carrying out
Instruction fetch；
A. refering to Fig. 6, a byte is taken out from queue heads, judges whether the byte is one of five numbers of definition, such as
Fruit is not, then being abandoned, because may be what interference generated；If it is one among five numbers, then being deposited
It is pending to enter to cache array etc.；
B. judge whether queue is sky, if non-empty, continue the process for repeating above-mentioned instruction fetch；If it is sky, that
Continue 3) to judge that caching array is instructed and executed in next step；
3) judgement caching array instructs and is moved ahead, retreated, turned left, right by single-chip microcontroller control crawler-type unmanned vehicle
The operation for turning, stopping；
The crawler-type unmanned vehicle resolve command is to complete under single-chip minimum system, while being equipped with E10-
433MS1W module, reception and transmission for 433MHz wireless communication；
Described is that content judgement is defined by above-mentioned table 1 by single-chip microcontroller control crawler-type unmanned vehicle, as received
When number is 2, just know that user uses the movement clenched fist, therefore crawler-type unmanned vehicle is promoted come driving motor by single-chip microcontroller
Advance；
The MCU driving motor is controlled by STC15W408AS single-chip microcontroller P1.5 and P5.4 pin, is passed through
Triode Q9012 saturation conduction, by principle, so that electromagnetic relay is had electric current by (no current passes through), and then control electricity
Magnetic relay is attracted conducting (disconnection) to reach control motor control purpose.Pass through differential control when left-hand rotation and right-hand rotation,
It moves forward and backward, is rotating and reverse for alternating current generator.
4. by the content of video camera shooting by passing second signal processing unit 3-2 back, and being shown refering to Fig. 7
Show in computer display screen
1) video camera starts to be shot, and acquires image information；
2) acquired image is digitized, obtains H.264 data flow, and send in such a way that 433MHz is wirelessly communicated
To second signal processing unit 3-2；
A. coding parameter is initialized first, and obtains input-buffer address and output buffer address；
B. the data of a frame are passed to, to the Image Acquisition carrier chrominance signal, the data of acquisition are encoded；
C. by the image taken using 16*16 as macro block, the pixel value of each macro block is calculated；
D. the pixel value for then calculating each macro block of the frame copies code stream head and generates message queue；
E. mode of the message queue in crawler-type unmanned vehicle by 433MHz wireless communication passes image information back second
Signal processing apparatus 3-2；
It is used in the scheme in frame, the form of interframe compression, 9 kinds of model predictions can be carried out to each macro block information,
After analyzing image, determines in nine kinds with a kind of prediction mode similar in original image, then each macro block of whole image is carried out
Prediction process, and then subtract each other acquisition residual values with intra-prediction image and original image, then by predictive information and residual values one
It rises and saves, decoding when can obtain original image.Speculate then in conjunction with the before and after frames of a certain frame current to be compressed
This part data method, in this way for available more efficient data.
3) after second signal processing unit 3-2 receives message queue, the data flow received is divided into two-way, is carried out all the way
Encapsulation turns and the storage of FLV file is written, and another way carries out computer video live streaming according to the broadcasting process of current shooting, is written
FLV file storing process is refering to Fig. 8；
The broadcasting process is responsible for the playing request of response computation generator terminal, is sent to using RTMP agreement by USB
Computer makes broadcasting be divided into two kinds of broadcasting and playback, that is, obtains H.264 data, parsing H.264 data；
If 4) computer 4 issues the request of live streaming, sent after data will be obtained using RTMP protocol packing data
To the player of client；If computer 4 issues the instruction of playback, according to the period of playback, corresponding document is read,
Parse data flow；
The data that image information is returned technical transmission by the photographing module in real time are H.264 data flows, belong to byte
Stream, the step are realized using message queue.
Embodiment 1:
Specific implementation process is as follows:
User first correctly wears myoelectric sensor 1 and reference electrode 2, and by identical first signal processing of 2 structures
Device 3-1 and second signal processing unit 3-2 is opened.If wanting to issue the instruction advanced, user needs to make the gesture clenched fist, this
When myoelectric sensor 1, reference electrode 2 will acquire the electromyography signal of individual subscriber, the first signal processing apparatus 3-1 will be by that will acquire
To electromyography signal carry out denoising, extract five category feature values of the electromyography signal, be the absolute value mean value of time domain respectively
(MAV), standard deviation (SD), root mean square (RMS), frequency domain frequency of average power (MPF) and median frequency (MF), then according to five
Category feature value is classified by the established sorter model of off-line data, judges that the gesture is simultaneously handle of clenching fist after classification
The gesture information is converted to corresponding number 02, then by character 2 by 433MHz wireless communication in a manner of be sent to crawler type without
On people's vehicle, when the antenna on crawler-type unmanned vehicle is properly received the signal, the single-chip microcontroller on crawler-type unmanned vehicle is compared
Judgement, that is, judge the instruction for advance, at this time single-chip microcontroller by P1.5 and P5.4I 0 mouthful of control alternating current generator operation, i.e.,
P1.5 pin is low level, and P5.4 pin is high level, the triode saturation conduction for promoting P1.5 pin to connect, electromagnetic relay
It is attracted, promotes alternating current generator that main story occurs, crawler-type unmanned vehicle travels forward.The camera being mounted on unmanned vehicle at this time just will
The image data of acquisition, and compression processing is carried out to the image information, second is sent in such a way that 433MHz is wirelessly communicated
In signal processing apparatus 3-2, computer 4 sends the instruction that request plays, and second signal processing unit 3-2 is passed through by decoding
USB interface is presented in computer 4, completes the process of entire method.
Claims (9)
1. a kind of ground crawler-type unmanned vehicle control method based on arm electromyography signal, it is characterised in that, it is described based on hand
The ground crawler-type unmanned vehicle control method of arm electromyography signal includes:
1) it correctly wears myoelectric sensor and carries out Initialize installation；
(1) myoelectric sensor (1) is worn over the position among hand and elbow by user, and electrode is made to be close to skin, while being made with reference to electricity
Pole (2) is attached on musculus flexor carpi ulnaris, it is ensured that can collect electromyography signal；
(2) it initializes, first makes myoelectric sensor (1) and crawler-type unmanned vehicle Initialize installation, i.e., in certain area
It is communicated in domain by the first signal processing apparatus (3-1) with the single-chip microcontroller on crawler-type unmanned vehicle to test whether to establish
Reliable wireless communication connection；
2) acquisition electromyography signal identifies different gestures and is sent to crawler-type unmanned vehicle；
3) crawler-type unmanned vehicle resolve command and the movement of the order is executed；
4) content by video camera shooting passes second time signal processing apparatus (3-2) by wireless transmission, and is shown in computer
(4) in.
2. the ground crawler-type unmanned vehicle control method described in accordance with the claim 1 based on arm electromyography signal, feature exist
In the acquisition electromyography signal, which identifies different gestures and is sent to crawler-type unmanned vehicle, to be referred to:
1) electromyography signal of user is acquired；
A. electromyography signal is acquired by myoelectric sensor (1) and reference electrode (2) first；
B. the data that myoelectric sensor (1) collected data and reference electrode (2) are collected into are sent to the first signal processing
In device (3-1)；
2) the first signal processing apparatus (3-1) carries out gesture identification and sends.
3. the ground crawler-type unmanned vehicle control method based on arm electromyography signal, feature exist according to claim 2
In first signal processing apparatus (3-1) carries out gesture identification and send to refer to:
A. judge whether collected signal is useful signal, if the signal of acquisition is invalid electromyography signal, then continuing to
Acquisition, until electromyographic signal collection success；
Starting point threshold value f is set thusHWith stop threshold value fL, it is made to remain fH>fLState, only when electromyography signal is greater than
fHWhen, it is judged as gesture and starts to execute, effective electromyography signal can be acquired；
B. a series of signal processing is passed through to collected electromyography signal；
C. the feature for extracting signal is that the feature for obtaining discrimination to the signal progress feature extraction of active segment carrys out promotion signal
Redundancy；
D. classifier carries out the classification of motion, judges whether to can recognize that electromyography signal, if being unable to identify that electromyography signal,
Acquisition will so be re-started；If can recognize that electromyography signal, which is converted into the dynamic of definition of gesture
It numbers；
E. movement number is sent it on crawler-type unmanned vehicle by 433MHz communication；
F. judge whether electromyography signal changes, so re-recognize the electromyography signal if changing, and repeat step 2) i.e.
The process that first signal processing apparatus (3-1) carries out gesture identification and sends continues if no change has taken place by the hand
Gesture generates the number for being converted to definition, sends by wirelessly；In case of changing, then will just change being converted to for gesture
Corresponding definition number is sent.
4. the ground crawler-type unmanned vehicle control method described in accordance with the claim 3 based on arm electromyography signal, feature exist
In described to refer to collected electromyography signal by a series of signal processing:
Signal is carried out to carry out denoising before feature extraction, by using small echo μ rule threshold function table to the height of electromyography signal
Frequency ingredient carries out de-noising；μ restrains threshold function table are as follows:
dr=sign (| w |-α τ), | w | > τ
Wherein, w indicates coefficient of wavelet decomposition；The key step that small echo μ restrains threshold denoising is as follows:
(1) WAVELET PACKET DECOMPOSITION is carried out to electromyography signal；
(2) threshold process is carried out to the high frequency coefficient decomposed；
(3) low frequency coefficient and the high frequency coefficient after threshold process are subjected to wavelet reconstruction, the original signal after obtaining de-noising.
5. the ground crawler-type unmanned vehicle control method described in accordance with the claim 3 based on arm electromyography signal, feature exist
In the feature of the extraction signal is to carry out feature extraction to the signal of active segment to obtain the feature of discrimination to promote letter
Number redundancy refer to:
The extraction of the electromyography signal feature is absolute value mean value (MAV), the standard deviation (SD), root mean square by selecting time domain
(RMS), the frequency of average power (MPF) of frequency domain and median frequency (MF) feature of totally 5 parameters as electromyography signal:
In formula: XiFor the electromyography signal after interception, Psd (f) indicates the power spectrum of electromyography signal.
6. the ground crawler-type unmanned vehicle control method described in accordance with the claim 3 based on arm electromyography signal, feature exist
In it is to establish optimal sorter model, and this point on the basis of off-line data that the classifier, which carries out the classification of motion,
Class model is applied in test set and is verified, and obtains optimal classification effect, and the model parameter then used is to online data
Carry out processing classification in real time；
The off-line data is 10 people's electromyography signals of acquisition, and acquires individual electromyography signal 300 times, and mark gesture label
It is numbered for the movement of the gesture；
(1) individual electromyography signal is acquired respectively first, i.e., the signal type of 5 kinds different gestures is adopted to corresponding to each gesture
Collection training set data puts on 1~No. 5 class label respectively；
(2) 5 category features for extracting 5 kinds of hand signals, are absolute value mean value (MAV), the standard deviation (SD), root mean square of time domain respectively
(RMS), the frequency of average power (MPF) and median frequency (MF) of frequency domain constitutes electromyography signal sample；
(3) it will be input in SVM classifier and classify after Fusion Features；
(4) ten foldings cross-check to obtain the optimal sorter model of classifier classifying quality；
The electromyography signal sample is that the electromyography signal of acquisition arm surface is pre-processed, filtered and extracted electromyography signal
Characteristic, composition characteristic vector, classifies for SVM；
Described being input in SVM classifier after Fusion Features is classified, and the execution method tested is as follows:
Training when the sample of some classification is successively classified as one kind, other remaining samples be classified as it is another kind of, such 5 classifications
Sample has just constructed 5 SVM.Unknown sample is classified as that class with maximum classification function value when classification, for the side of explanation
Just bent wrist, clench fist, stretch the palm, shooting, sky are pinched five movement class data and are replaced respectively with A, B, C, D, E；
(1) vector corresponding to A collects as positive, B, C, and vector corresponding to D, E is trained as negative collection using the training set
Obtain training result f1 (x)；Vector corresponding to B collects as positive, A, C, and vector corresponding to D, E uses the instruction as negative collection
Practice collection to be trained to obtain training result f2 (x)；Vector corresponding to C collects as positive, A, B, and vector corresponding to D, E is as negative
Collection, is trained to obtain training result f3 (x) using the training set；Vector corresponding to D collects as positive, A, B, corresponding to C, E
Vector as negative collection, be trained to obtain training result f4 (x) using the training set；Vector corresponding to E collects as positive, A,
Vector corresponding to B, C, D is trained to obtain training result f5 (x) as negative collection using the training set；
(2) when test, corresponding test vector is utilized respectively this five training results and is tested.Obtain each survey
The result type f1 (x) of examination, f2 (x), f3 (x), f4 (x), f5 (x)；
(3) choose five value in maximum one be used as classification results；
The described ten foldings crosscheck, which refers to, is divided into 10 parts for data set, in turn will wherein 9 parts be used as training data, 1 part of conduct
Test data is tested.
The identification electromyography signal is according to when carrying out different gesture identifications, and the contribution degree of each muscle group is different, therefore
Characteristic value also can be different；Classified by classifier, and judges to belong to a certain classification, and then identify specific gesture；
Table 1
The definition of gesture is as follows: controlling crawler-type unmanned vehicle by 5 manipulation instructions；Table 1 is that gesture is corresponding with meaning
Relationship is controlled unmanned vehicle using gesture and carries out displacement movement.
7. the ground crawler-type unmanned vehicle control method described in accordance with the claim 3 based on arm electromyography signal, feature exist
In described to judge whether electromyography signal changes the surface referred to by the method for sliding window energy threshold to gesture motion
Electromyography signal carries out active segment detection, and specific algorithm is as follows:
1. the absolute value for calculating the electromyography signal in all channels it is average and；
2. ask the absolute value of electromyography signal in window to be averaged the quadratic sum of sum using the rectangular slide window of a length of N of window,
It is f respectively that beginning and end threshold value, which is arranged,HAnd fL(fH>fL), when electromyography signal is greater than starting point threshold value, determine that gesture is opened
Begin；When electromyography signal instantaneous energy all in 100ms is less than terminal threshold value, determine that gesture motion terminates；
3. whether starting detection gesture movement if electromyography signal is greater than the threshold value of starting point in active segment after gesture terminates
It changes.
8. the ground crawler-type unmanned vehicle control method described in accordance with the claim 1 based on arm electromyography signal, feature exist
In, the crawler-type unmanned vehicle resolve command and the movement for executing the order refers to:
1) single-chip microcontroller carries out wireless communication, singlechip interruption initializes；
2) whether single-chip microcontroller is sky by detecting flag bit constantly come decision instruction queue, if it is non-empty, then carrying out fetching
It enables；
A. a byte is taken out from queue heads, judges whether the byte is one of five numbers of definition, if it is not, so will
It is abandoned, because may be what interference generated；If it is one among five numbers, waited then being deposited into caching array
It executes；
B. judge whether queue is sky, if non-empty, continue the process for repeating above-mentioned instruction fetch；If it is sky, then after
Continue following 3) step judge cache array instruct and carry out various operations；
3) judgement caching array instructs and is moved ahead, retreated, turned left, turn right, stopped by single-chip microcontroller control crawler-type unmanned vehicle
Operation only；
Described is that content judgement, the number such as received are defined by above-mentioned table 1 by single-chip microcontroller control crawler-type unmanned vehicle
When being 2, just know that user uses the movement clenched fist, therefore crawler-type unmanned Chinese herbaceous peony is promoted come driving motor by single-chip microcontroller
Into.
9. the ground crawler-type unmanned vehicle control method described in accordance with the claim 1 based on arm electromyography signal, feature exist
In the content by video camera shooting is shown in calculating by passing second signal processing unit (3-2) back
Refer in machine (4):
1) video camera starts to be shot, and acquires image information；
2) acquired image is digitized, obtains H.264 data flow, and be sent in such a way that 433MHz is wirelessly communicated the
Binary signal processing unit (3-2)；
A. coding parameter is initialized first, and obtains input-buffer address and output buffer address；
B. the data of a frame are passed to, to the Image Acquisition carrier chrominance signal, the data of acquisition are encoded；
C. by the image taken using 16*16 as macro block, the pixel value of each macro block is calculated；
D. the pixel value for then calculating each macro block of the frame copies code stream head and generates message queue；
E. mode of the message queue in crawler-type unmanned vehicle by 433MHz wireless communication passes image information back second signal
Processing unit (3-2)；
3) after second signal processing unit (3-2) receives message queue, the data flow received is divided into two-way, is sealed all the way
Dress turns and the storage of FLV file is written, and another way carries out computer video live streaming according to the broadcasting process of current shooting；
If 4) computer (4) issues the request of live streaming, it is sent to after data will be obtained using RTMP protocol packing data
The player of client；If computer (4) issues the instruction of playback, according to the period of playback, corresponding document is read,
Parse data flow.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910051882.7A CN109765823A (en) | 2019-01-21 | 2019-01-21 | Ground crawler-type unmanned vehicle control method based on arm electromyography signal |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910051882.7A CN109765823A (en) | 2019-01-21 | 2019-01-21 | Ground crawler-type unmanned vehicle control method based on arm electromyography signal |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109765823A true CN109765823A (en) | 2019-05-17 |
Family
ID=66454836
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910051882.7A Pending CN109765823A (en) | 2019-01-21 | 2019-01-21 | Ground crawler-type unmanned vehicle control method based on arm electromyography signal |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109765823A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111651046A (en) * | 2020-06-05 | 2020-09-11 | 上海交通大学 | Gesture intention recognition system without hand action |
| CN114081513A (en) * | 2021-12-13 | 2022-02-25 | 苏州大学 | Electromyographic signal-based abnormal driving behavior detection method and system |
| CN114625246A (en) * | 2022-02-14 | 2022-06-14 | 深圳市心流科技有限公司 | Gesture combination triggering method and device, intelligent bionic hand and storage medium |
| CN115114962A (en) * | 2022-07-19 | 2022-09-27 | 歌尔股份有限公司 | Control method and device based on surface electromyogram signal and wearable device |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150070270A1 (en) * | 2013-09-06 | 2015-03-12 | Thalmic Labs Inc. | Systems, articles, and methods for electromyography-based human-electronics interfaces |
| CN107230334A (en) * | 2017-05-05 | 2017-10-03 | 北京理工大学 | A kind of Portable unmanned car ground control terminal |
| CN107529041A (en) * | 2017-09-30 | 2017-12-29 | 江西洪都航空工业集团有限责任公司 | A kind of long-distance monitoring method for unmanned agricultural vehicle |
| CN108829252A (en) * | 2018-06-14 | 2018-11-16 | 吉林大学 | Gesture input computer character device and method based on electromyography signal |
| CN108900928A (en) * | 2018-07-26 | 2018-11-27 | 宁波视睿迪光电有限公司 | Method and device, the 3D screen client, Streaming Media Cloud Server of naked eye 3D live streaming |
| CN109168031A (en) * | 2018-11-06 | 2019-01-08 | 杭州云英网络科技有限公司 | Streaming Media method for pushing and device, steaming media platform |
-
2019
- 2019-01-21 CN CN201910051882.7A patent/CN109765823A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150070270A1 (en) * | 2013-09-06 | 2015-03-12 | Thalmic Labs Inc. | Systems, articles, and methods for electromyography-based human-electronics interfaces |
| CN107230334A (en) * | 2017-05-05 | 2017-10-03 | 北京理工大学 | A kind of Portable unmanned car ground control terminal |
| CN107529041A (en) * | 2017-09-30 | 2017-12-29 | 江西洪都航空工业集团有限责任公司 | A kind of long-distance monitoring method for unmanned agricultural vehicle |
| CN108829252A (en) * | 2018-06-14 | 2018-11-16 | 吉林大学 | Gesture input computer character device and method based on electromyography signal |
| CN108900928A (en) * | 2018-07-26 | 2018-11-27 | 宁波视睿迪光电有限公司 | Method and device, the 3D screen client, Streaming Media Cloud Server of naked eye 3D live streaming |
| CN109168031A (en) * | 2018-11-06 | 2019-01-08 | 杭州云英网络科技有限公司 | Streaming Media method for pushing and device, steaming media platform |
Non-Patent Citations (4)
| Title |
|---|
| 加玉涛等: "肌电信号特征提取方法综述", 《电子器件》 * |
| 李珊珊等: "基于 ＭＹＯ 手环的移动机器人控制系统设计", 《南昌大学学报（理科版）》 * |
| 王光旭: "基于表面肌电信号的下肢运动模式识别的研究", 《中国优秀硕士学位论文全文数据库 医药卫生科技辑》 * |
| 陈泽华: "基于表面肌电和加速度信号的时序组合动作识别方法", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111651046A (en) * | 2020-06-05 | 2020-09-11 | 上海交通大学 | Gesture intention recognition system without hand action |
| CN114081513A (en) * | 2021-12-13 | 2022-02-25 | 苏州大学 | Electromyographic signal-based abnormal driving behavior detection method and system |
| CN114081513B (en) * | 2021-12-13 | 2023-04-07 | 苏州大学 | Electromyographic signal-based abnormal driving behavior detection method and system |
| CN114625246A (en) * | 2022-02-14 | 2022-06-14 | 深圳市心流科技有限公司 | Gesture combination triggering method and device, intelligent bionic hand and storage medium |
| CN115114962A (en) * | 2022-07-19 | 2022-09-27 | 歌尔股份有限公司 | Control method and device based on surface electromyogram signal and wearable device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109765823A (en) | Ground crawler-type unmanned vehicle control method based on arm electromyography signal | |
| US10061389B2 (en) | Gesture recognition system and gesture recognition method | |
| Singh et al. | Transforming sensor data to the image domain for deep learning—An application to footstep detection | |
| CN105205436A (en) | Gesture identification system based on multiple forearm bioelectric sensors | |
| Ghosh et al. | Real-time object recognition and orientation estimation using an event-based camera and CNN | |
| CN104461013A (en) | Human body movement reconstruction and analysis system and method based on inertial sensing units | |
| CN102104771B (en) | Multi-channel people stream rate monitoring system based on wireless monitoring | |
| CN112990026B (en) | Wireless signal perception model construction and perception method and system based on countermeasure training | |
| CN105224066A (en) | A kind of gesture identification method based on high in the clouds process | |
| CN102054167A (en) | All-weather multipath channel pedestrian flow monitoring system based on wireless infrared monitoring | |
| CN110414468B (en) | Identity verification method based on gesture signal in WiFi environment | |
| CN106406297A (en) | Wireless electroencephalogram-based control system for controlling crawler type mobile robot | |
| CN105138995A (en) | Time-invariant and view-invariant human action identification method based on skeleton information | |
| CN106203497A (en) | A kind of finger vena area-of-interest method for screening images based on image quality evaluation | |
| Yang et al. | Wiimg: Pushing the limit of wifi sensing with low transmission rates | |
| Devi et al. | Web enabled paddy disease detection using Compressed Sensing | |
| KR20130073361A (en) | Apparatus and method for classifing pattern of electromyogram signals | |
| CN108065938A (en) | Animal activity monitors system and the active state recognition methods based on neutral net | |
| CN110555393A (en) | method and device for analyzing pedestrian wearing characteristics from video data | |
| CN108064745A (en) | Animal yelps monitoring system and the state identification method of yelping based on machine learning | |
| Bhattacharyya et al. | EEG controlled remote robotic system from motor imagery classification | |
| CN112380903B (en) | Human body activity recognition method based on WiFi-CSI signal enhancement | |
| CN109918994A (en) | A kind of act of violence detection method based on commercial Wi-Fi | |
| CN111652132B (en) | Non-line-of-sight identity recognition method and device based on deep learning and storage medium | |
| Zhou et al. | Deep-WiID: WiFi-based contactless human identification via deep learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190517 |