System based on brain wave control and wheelchair thereof

Abstract

The invention discloses a system based on brain wave control and a wheelchair thereof, and relates to the field of brain-controlled wheelchairs, the system comprises a BCI module, a control module and an equipment terminal, the BCI module comprises an electroencephalogram collector, an electroencephalogram signal acquisition unit and an electroencephalogram signal transmission unit; the control module comprises an upper computer, a brain-like chip, an electroencephalogram signal feature extraction unit and an electroencephalogram signal feature classification unit; the equipment terminal comprises a wheelchair body, a first wheelchair peripheral and a second wheelchair peripheral; the BCI module is used for collecting, recording and analyzing the electroencephalogram signals and transmitting real-time data to the control module; the control module is used for extracting and classifying the features of the electroencephalogram signals and accurately matching the electroencephalogram signals with wheelchair action instructions. The robot improves the freedom of action, reduces the operation burden, improves the safety, expands the application scene, achieves the modular design and autonomous learning universality, and brings brand new life experience to the disabled and the old.

Description

Technical Field

[0001] The present invention mainly relates to the technical field of brain-controlled wheelchairs, and specifically to a system and a wheelchair based on brain wave control. Background Art

[0002] Brain control technology, also known as brain-computer interface technology, is a technology that collects, processes and analyzes signals generated by the human brain to achieve information exchange and control between the human brain and external devices. Brain control technology collects the user's brain signals (such as brain waves) and inputs them into a computer for real-time processing to mine information that reflects the human brain's intentions. This information is then converted into control instructions for real-time control of external devices. The control effect of the external device is fed back to the user, who can adjust the brain control strategy based on this feedback to further improve control efficiency. In this process, the human brain plays the role of a "commander", and the "commands" issued are implicit in the brain signals, which are interpreted by the computer.

[0003] Due to the mobility difficulties of many disabled and elderly people in today's society, the vast majority of people use wheelchairs for their daily lives. The many inconveniences in the mechanical operation of ordinary wheelchairs bring tremendous pressure to their lives and spirits. Among them, people with mobility difficulties: those with clear brains, semi-paralysis or full body paralysis, unable to move, walk, and even unable to take care of themselves, are prone to sudden death, and must be taken care of by relatives, which brings serious life burdens to their families and themselves. Summary of the invention

[0004] Based on this, the purpose of the present invention is to provide a system and a wheelchair based on brain wave control to solve the technical problems raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A system based on brain wave control and a wheelchair thereof, comprising a BCI module, a control module, and a device terminal, wherein the BCI module comprises an electroencephalogram collector, an electroencephalogram signal acquisition unit, and an electroencephalogram signal transmission unit; the control module comprises a host computer and a brain-like chip, an electroencephalogram signal feature extraction unit, and an electroencephalogram signal feature classification unit; the device terminal comprises a wheelchair body, a first wheelchair peripheral, and a second wheelchair peripheral;

[0007] The BCI module is used to collect, record and analyze brain electrical signals, and transmit real-time data to the control module;

[0008] The control module is used to extract and classify the features of the EEG signal, accurately match the EEG signal with the wheelchair action instructions, and control the specific actions of the wheelchair body;

[0009] The device terminal is used as a means of transportation for users with limited mobility, and ensures the safety of the transportation behavior during the transportation behavior.

[0010] Preferably, the EEG signal acquisition unit is used to collect task EEG signals from the brain of a user sitting on the wheelchair body, and transmit them to the brain-like chip in real time through the EEG signal transmission unit;

[0011] During the acquisition and processing of the EEG signal acquisition unit, the regularization term is introduced to select EEG signal features using the Lasso regression model to reduce redundant EEG signal features.

[0012] According to claim 2, a system and wheelchair based on brain wave control are characterized in that the task EEG signals are mainly generated by the user's brain activities and brain imagination, and the task EEG signals mainly include the following four types:

[0013] Task 1: The EEG signal generated by the user imagining the wheelchair moving forward;

[0014] Task 2: The EEG signal generated by the user imagining the wheelchair turning left;

[0015] Task 3: The EEG signal generated by the user imagining the wheelchair turning to the right;

[0016] Task 4: The EEG signals generated by the user imagining the wheelchair stopping and moving.

[0017] Preferably, the specific formula of the Lasso regression model is as follows:

[0018]

[0019] in,

[0020] n: is the number of samples of EEG signal features;

[0021] p: is the number of EEG signal features;

[0022] y i : is the response variable of the i-th sample of the EEG signal feature;

[0023] x ij : is the jth eigenvalue of the i-th sample of the EEG signal feature;

[0024] β0: is the intercept term;

[0025] β j : is the coefficient of the jth feature of the EEG signal;

[0026] λ is a regularization parameter, which is used to control the strength of regularization. The larger the λ is, the stronger the regularization effect is and the simpler the model is. On the contrary, the smaller the λ is, the weaker the regularization effect is and the more complex the model is.

[0027] |β j |: indicates β j This is the key part of L1 regularization, which makes the unimportant EEG signal feature coefficients tend to 0, thereby realizing EEG signal feature selection.

[0028] Preferably, during the fitting process of the Lasso regression model, the model fitting error is determined by calculating the mean absolute error to evaluate the fitting effect of the model;

[0029] The mean absolute error is expressed as follows:

[0030]

[0031] in,

[0032] n: is the number of EEG signal feature samples,

[0033] y i : is the true value of the i-th sample of the EEG signal feature;

[0034] is the predicted value of the i-th sample of the EEG signal feature;

[0035] The smaller the MAE, the more accurate the model prediction is, that is, the smaller the difference between the predicted value and the actual value is.

[0036] Preferably, the EEG signal feature extraction unit receives in real time the screened and recorded task EEG signals sent by the EEG signal transmission unit, and classifies the feature signals, i.e., four task EEG signals, through the EEG signal feature classification unit and accurately matches the four task EEG signals with the wheelchair body action instructions.

[0037] Preferably, the task EEG signal is classified and matched to the action instruction by the brain-like chip as follows:

[0038] Type 1 signal: The EEG signal generated by the user's imagination of the wheelchair moving forward matches the wheelchair moving forward;

[0039] Type II signal: The EEG signal generated by the user imagining the wheelchair turning left matches the wheelchair moving a certain distance to the left;

[0040] The third type of signal: The EEG signal generated by the user's imagination of the wheelchair turning to the right matches the wheelchair moving a certain distance to the right;

[0041] The fourth type of signal: The EEG signal generated by the user's imagination of the wheelchair stopping and moving matches the wheelchair stopping.

[0042] Preferably, the first wheelchair peripheral is a visual sensor, and the second wheelchair peripheral is a laser radar; the first wheelchair peripheral and the second wheelchair peripheral are connected to the host computer by electrical signals;

[0043] The first wheelchair peripheral and the second wheelchair peripheral jointly formulate and generate a wheelchair stop motion signal;

[0044] When the visual sensor and the laser radar detect that there are no obstacles ahead, the host computer normally executes the generated signals of type one to four;

[0045] When the visual sensor and the laser radar detect an obstacle ahead, the host computer preferentially executes the first wheelchair peripheral and the second wheelchair peripheral to jointly generate a wheelchair stop motion signal.

[0046] Preferably, an adjustment assembly is provided on one side of the wheelchair body, and the adjustment assembly includes a support rod provided on one side of the wheelchair body, a limit block provided on one side of the support rod, an adjustment rod penetrating one end of the limit block and slidably connected, an adjustment screw embedded on one side of the limit block, and a placement horizontal plate provided on the side wall of one end of the adjustment rod and rotatably connected;

[0047] The first wheelchair peripheral and the second wheelchair peripheral are arranged on the top of the placement horizontal plate.

[0048] Preferably, a limit stop rod is provided at the top of one end of the adjusting rod close to the placement transverse plate, and a limit support rod is provided at one side of the adjusting rod and at the bottom of the placement transverse plate.

[0049] In summary, the present invention mainly has the following beneficial effects:

[0050] Brain-controlled wheelchairs allow users to directly control the movement of the wheelchair through brain electrical signals, without relying on traditional joysticks or other auxiliary equipment, greatly improving the user's freedom of movement and autonomy; compared with traditional wheelchairs, brain-controlled wheelchairs reduce the user's operating burden. Users can control the wheelchair through simple thinking activities without complicated physical operations;

[0051] In addition, it has an obstacle avoidance function. By using visual sensors and laser radar to identify obstacles, the wheelchair can automatically avoid obstacles and reduce the risk of collision.

[0052] Through big data analysis, machine learning algorithms, and the use of mathematical models, the brain-controlled wheelchair can continuously optimize its control algorithm and recognition accuracy, improve operation and ensure action safety.

[0053] It has a wide range of applications. Brain-controlled wheelchairs are not only suitable for severely disabled people such as quadriplegia, but also for the elderly and people with physical disabilities. It has good social benefits. The promotion and application of brain-controlled wheelchairs will help improve the quality of life of the disabled and the elderly, reduce social burdens, and improve the overall social welfare. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 It is the overall system framework diagram of the present invention;

[0055] Figure 2 It is a schematic diagram of a specific implementation of the present invention;

[0056] Figure 3 It is a specific framework diagram of the BCI module of the present invention;

[0057] Figure 4 It is a specific framework diagram of the control module of the present invention;

[0058] Figure 5 It is an axonometric diagram of the overall structure of the device terminal of the present invention;

[0059] Figure 6 It is an axonometric diagram of the overall structure of the adjustment assembly of the present invention;

[0060] Figure 7 It is an exploded view of the overall structure of the adjustment component of the present invention;

[0061] Figure 8 It is an axonometric view of the adjustment component structure of the present invention.

[0062] Description of the drawings: 10. BCI module; 20. Control module; 30. Equipment terminal; 40. Adjustment component; 11. EEG collector; 12. EEG signal acquisition unit; 13. EEG signal transmission unit; 21. Brain-like chip; 22. EEG signal feature extraction unit; 23. EEG signal feature classification unit; 31. Wheelchair body; 32. First wheelchair peripheral; 33. Second wheelchair peripheral; 41. Support rod; 42. Limit block; 43. Adjustment rod; 44. Adjustment screw; 45. Placement cross plate; 431. Limit stop rod; 432. Limit support rod. DETAILED DESCRIPTION

[0063] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0064] Example

[0065] like Figure 1 , 2, 3, and 4, including a BCI module 10, a control module 20, and a device terminal 30, the BCI module 10 includes an electroencephalogram collector 11, an electroencephalogram signal acquisition unit 12, and an electroencephalogram signal transmission unit 13; the BCI module 10 is used to collect, record, and analyze electroencephalogram signals, and transmit real-time data to the control module 20; the electroencephalogram signal acquisition unit 12 is used to collect task electroencephalogram signals of the brain of a user sitting on a wheelchair body 31, and transmit them to the brain-like chip 21 in real time through the electroencephalogram signal transmission unit 13;

[0066] During the acquisition and processing of the EEG signal acquisition unit 12, the Lasso regression model is used to select the EEG signal features by introducing the regularization term to reduce the redundant EEG signal features; the task EEG signal is mainly generated by the user's brain activity and brain imagination. The task EEG signal mainly includes the following four types:

[0067] Task 1: The EEG signal generated by the user imagining the wheelchair moving forward;

[0068] Task 2: The EEG signal generated by the user imagining the wheelchair turning left;

[0069] Task 3: The EEG signal generated by the user imagining the wheelchair turning to the right;

[0070] Task 4: EEG signals generated by the user imagining the wheelchair stopping and moving;

[0071] The specific formula of the Lasso regression model is as follows:

[0072]

[0073] in,

[0074] n: is the number of samples of EEG signal features;

[0075] p: is the number of EEG signal features;

[0076] y i : is the response variable of the i-th sample of the EEG signal feature;

[0077] x ij : is the jth eigenvalue of the i-th sample of the EEG signal feature;

[0078] β0: is the intercept term;

[0079] β j : is the coefficient of the jth feature of the EEG signal;

[0080] λ is a regularization parameter, which is used to control the strength of regularization. The larger the λ is, the stronger the regularization effect is and the simpler the model is. On the contrary, the smaller the λ is, the weaker the regularization effect is and the more complex the model is.

[0081] |β j |: indicates β j The absolute value of is the key part of L1 regularization, which makes the unimportant EEG signal feature coefficients tend to 0, thereby realizing EEG signal feature selection;

[0082] During the fitting process of the Lasso regression model, the mean absolute error (MAE) is calculated to determine the model fitting error and evaluate the model fitting effect.

[0083] The mean absolute error MAE is shown as follows:

[0084]

[0085] in,

[0086] n: is the number of EEG signal feature samples,

[0087] y i : is the true value of the i-th sample of the EEG signal feature;

[0088] is the predicted value of the i-th sample of the EEG signal feature;

[0089] The smaller the MAE, the more accurate the model prediction is, that is, the smaller the difference between the predicted value and the actual value is.

[0090] It should be noted that, in this embodiment, the specification parameters of the brain-like chip 21 in the control module 20 of the brain-controlled wheelchair are: brain-like multi-core architecture, computer storage one, CNN / SN, high-performance processor, MIPI&TFT-LCD&LVDS, 28nm process, tens of millions of neural prominence, easy expansion at scale, support for up to 4.6 million neurons, mimetic computing, programmable dynamics simulation solution, support for asynchronous sparse computing, chiplet-based four-core Internet architecture, self-developed low-latency inter-chip extended communication technology, multiple computing precisions: FP32 / NT16 / INT8, energy efficiency exceeds 500 times of CPU / GPU;

[0091] The BCI module 10 is disposed at the position of the headrest in the wheelchair body 31 to facilitate the collection, recording and analysis of the user's EEG signals;

[0092] In the screening process, in order to avoid the interference of EEG signals of non-collection tasks in the acquisition of multi-channel EEG signals and affect the subsequent precise control of the wheelchair body 31, the Lasso regression model is cited. By introducing the L1 regularization term, the unimportant feature coefficients tend to 0, thereby realizing feature selection and filtering unnecessary collection and recording information, which greatly facilitates the extraction and classification of the subsequent control module 20; and by calculating the mean absolute error MAE, the model fitting error is judged, the model fitting effect is evaluated, and excessive filtering is further avoided to exclude effective EEG signals;

[0093] Among them, the main tasks are to obtain: the EEG signals generated when the user imagines the wheelchair moving forward; Task 2: the EEG signals generated when the user imagines the wheelchair turning left; Task 3: the EEG signals generated when the user imagines the wheelchair turning right; Task 4: the EEG signals generated when the user imagines the wheelchair stopping and moving.

[0094] like Figure 1 , 2 As shown in , 3 and 4, the control module 20 includes a host computer IPC and a brain-like chip 21, an EEG signal feature extraction unit 22, and an EEG signal feature classification unit 23; the device terminal 30 includes a wheelchair body 31, a first wheelchair peripheral 32, and a second wheelchair peripheral 33; the control module 20 is used to extract and classify the features of the EEG signal, and accurately match the EEG signal with the wheelchair action command to control the specific action of the wheelchair body 31;

[0095] The EEG signal feature extraction unit 22 receives the screened and recorded task EEG signals sent by the EEG signal transmission unit 13 in real time, and classifies the feature signals, i.e., the four task EEG signals, through the EEG signal feature classification unit 23, and accurately matches the four task EEG signals with the action instructions of the wheelchair body 31;

[0096] The task EEG signals are classified and matched with action instructions through the brain-like chip 21 as follows:

[0097] Type 1 signal: The EEG signal generated by the user's imagination of the wheelchair moving forward matches the wheelchair moving forward;

[0098] Type II signal: The EEG signal generated by the user imagining the wheelchair turning left matches the wheelchair moving a certain distance to the left;

[0099] The third type of signal: The EEG signal generated by the user's imagination of the wheelchair turning to the right matches the wheelchair moving a certain distance to the right;

[0100] Four types of signals: The EEG signal generated by the user's imagination of the wheelchair stopping and moving matches the wheelchair stopping;

[0101] The first wheelchair peripheral 32 is a visual sensor, and the second wheelchair peripheral 33 is a laser radar; the first wheelchair peripheral 32 and the second wheelchair peripheral 33 are connected to the host computer IPC electrical signal;

[0102] The first wheelchair peripheral 32 and the second wheelchair peripheral 33 jointly formulate and generate a wheelchair stop motion signal;

[0103] When the visual sensor and lidar detect that there are no obstacles ahead, the host computer IPC normally executes the generated signals of types 1 to 4;

[0104] When the visual sensor and the laser radar detect an obstacle ahead, the host computer IPC preferentially executes the first wheelchair peripheral 32 and the second wheelchair peripheral 33 to jointly generate a wheelchair stop motion signal.

[0105] It should be noted that, in this embodiment, when the brain-like chip 21 receives the signal sent by the BCI module 10, it extracts and classifies the signal:

[0106] Type 1 signal: The EEG signal generated by the user's imagination of the wheelchair moving forward matches the wheelchair moving forward;

[0107] Type II signal: The EEG signal generated by the user imagining the wheelchair turning left matches the wheelchair moving a certain distance to the left;

[0108] The third type of signal: The EEG signal generated by the user's imagination of the wheelchair turning to the right matches the wheelchair moving a certain distance to the right;

[0109] Four types of signals: The EEG signal generated by the user's imagination of the wheelchair stopping and moving matches the wheelchair stopping;

[0110] During the experiment, we use the sigaction function. The sample code is as follows:

[0111]

[0112]

[0113] Before the next execution instruction is issued, the current action instruction is still executed, except for special circumstances; the special circumstances refer to that in order to further ensure the safety of the wheelchair body 31 during use, we install the first wheelchair peripheral 32 and the second wheelchair peripheral 33 on the wheelchair body 31, namely the visual sensor and the laser radar, which are responsible for detecting obstacles in the environment and sending the detection information to the upper computer IPC.

[0114] When the visual sensor and lidar detect that there are no obstacles ahead, the host computer IPC normally executes the generated signals of types 1 to 4;

[0115] When the visual sensor and lidar detect an obstacle ahead, a real-time signal is sent to the host computer IPC. The host computer IPC receives the signal and generates a wheelchair stop movement instruction, which has the highest priority.

[0116] like Figure 5 , 6 As shown in FIGS. 7 and 8, the device terminal 30 is used as a walking tool for users with limited mobility, and ensures the safety of walking behavior during walking. An adjustment component 40 is provided on one side of the wheelchair body 31, and the adjustment component 40 includes a support rod 41 provided on one side of the wheelchair body 31, a limit block 42 provided on one side of the support rod 41, an adjustment rod 43 penetrating one end of the limit block 42 and slidably connected, an adjustment screw 44 embedded on one side of the limit block 42, and a placement horizontal plate 45 provided on the side wall of one end of the adjustment rod 43 and rotatably connected;

[0117] The first wheelchair peripheral 32 and the second wheelchair peripheral 33 are arranged on the top of the horizontal plate 45;

[0118] A limit stopper 431 is provided on the top of one end of the adjusting rod 43 close to the placement transverse plate 45 , and a limit support rod 432 is provided on one side of the adjusting rod 43 and at the bottom of the placement transverse plate 45 .

[0119] It should be noted that, in this embodiment, in order to facilitate the user to use the brain-controlled wheelchair, an adjustment component 40 is provided, so that the first wheelchair peripheral 32 located at the front side of the wheelchair body 31 and the second wheelchair peripheral 33 are conveniently placed on the horizontal plate 45 to change their positions, so that the user can sit on the wheelchair body 31 conveniently;

[0120] The support rod 41 is welded with the limit block 42, the adjustment rod 43 crosses the inner wall of the limit block 42, and the adjustment rod 43 is positioned and fastened by the adjustment screw 44. The adjustment rod 43 can adjust the position of the first wheelchair peripheral 32 and the second wheelchair peripheral 33 on the front side of the wheelchair body 31, so that the user can sit on it more comfortably.

[0121] The adjusting rod 43 is provided with a limit stopper rod 431 and a limit support rod 432. The limit support rod 432 can support the placement cross plate 45. The limit stopper rod 431 is provided to limit the rotation and provide auxiliary support for the placement cross plate 45 when the placement cross plate 45 is rotated.

[0122] The working principle of the present invention is:

[0123] The specifications of the brain-like chip 21 in the control module 20 of the brain-controlled wheelchair are as follows: brain-like multi-core architecture, computer storage 1, CNN / SN, high-performance processor, MIPI&TFT-LCD&LVDS, 28nm process, tens of millions of neural prominence, easy expansion at scale, support for up to 4.6 million neurons, mimetic computing, programmable dynamics simulation solution, support for asynchronous sparse computing, chiplet-based four-core Internet architecture, self-developed low-latency inter-chip extended communication technology, multiple computing precisions: FP32 / NT16 / INT8, energy efficiency exceeds 500 times of CPU / GPU;

[0124] The BCI module 10 is disposed at the position of the headrest in the wheelchair body 31 to facilitate the collection, recording and analysis of the user's EEG signals;

[0125] In the screening process, in order to avoid the interference of EEG signals of non-collection tasks in the acquisition of multi-channel EEG signals and affect the subsequent precise control of the wheelchair body 31, the Lasso regression model is cited. By introducing the L1 regularization term, the unimportant feature coefficients tend to 0, thereby realizing feature selection and filtering unnecessary collection and recording information, which greatly facilitates the extraction and classification of the subsequent control module 20; and by calculating the mean absolute error MAE, the model fitting error is judged, the model fitting effect is evaluated, and excessive filtering is further avoided to exclude effective EEG signals;

[0126] Among them, the main acquisition task is: the EEG signal generated by the user's imagination of the wheelchair moving forward; Task 2: The EEG signal generated by the user's imagination of the wheelchair turning left; Task 3: The EEG signal generated by the user's imagination of the wheelchair turning right; Task 4: The EEG signal generated by the user's imagination of the wheelchair stopping;

[0127] When the brain-like chip 21 receives the signal sent by the BCI module 10, it extracts and classifies the signal:

[0128] Type 1 signal: The EEG signal generated by the user's imagination of the wheelchair moving forward matches the wheelchair moving forward;

[0129] Type II signal: The EEG signal generated by the user imagining the wheelchair turning left matches the wheelchair moving a certain distance to the left;

[0130] The third type of signal: The EEG signal generated by the user's imagination of the wheelchair turning to the right matches the wheelchair moving a certain distance to the right;

[0131] Four types of signals: The EEG signal generated by the user's imagination of the wheelchair stopping and moving matches the wheelchair stopping;

[0132] During the experiment, we use the sigaction function. The sample code is as follows:

[0133]

[0134] Before the next execution instruction is issued, the current action instruction is still executed, except for special circumstances; the special circumstances refer to that in order to further ensure the safety of the wheelchair body 31 during use, we install the first wheelchair peripheral 32 and the second wheelchair peripheral 33 on the wheelchair body 31, namely the visual sensor and the laser radar, which are responsible for detecting obstacles in the environment and sending the detection information to the upper computer IPC.

[0135] When the visual sensor and lidar detect that there are no obstacles ahead, the host computer IPC normally executes the generated signals of types 1 to 4;

[0136] When the visual sensor and laser radar detect an obstacle ahead, a real-time signal is sent to the host computer IPC, which receives the signal and generates a wheelchair stop motion instruction, which has the highest priority;

[0137] Finally, in order to facilitate the user to use the brain-controlled wheelchair, an adjustment component 40 is provided to facilitate the first wheelchair peripheral 32 located at the front side of the wheelchair body 31 and the second wheelchair peripheral 33 to be placed on the horizontal plate 45 for conveniently changing the position, so that the user can sit on the wheelchair body 31;

[0138] The support rod 41 is welded with the limit block 42, the adjustment rod 43 crosses the inner wall of the limit block 42, and the adjustment rod 43 is positioned and fastened by the adjustment screw 44. The adjustment rod 43 can adjust the position of the first wheelchair peripheral 32 and the second wheelchair peripheral 33 on the front side of the wheelchair body 31, so that the user can sit on it more comfortably.

[0139] The adjusting rod 43 is provided with a limit stopper rod 431 and a limit support rod 432. The limit support rod 432 can support the placement cross plate 45. The limit stopper rod 431 is provided to limit the rotation and provide auxiliary support for the placement cross plate 45 when the placement cross plate 45 is rotated.

[0140] The above embodiments are only for illustrating the technical idea of ​​the present invention, and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution in accordance with the technical idea proposed by the present invention shall fall within the protection scope of the present invention.

Claims

1. A brain wave control system and a wheelchair thereof, comprising a BCI module (10), a control module (20), and a device terminal (30), characterized in that The BCI module (10) includes an electroencephalogram collector (11), an electroencephalogram signal acquisition unit (12), and an electroencephalogram signal transmission unit (13); the control module (20) includes a host computer (IPC) and a brain-like chip (21), an electroencephalogram signal feature extraction unit (22), and an electroencephalogram signal feature classification unit (23); the device terminal (30) includes a wheelchair body (31), a first wheelchair peripheral (32), and a second wheelchair peripheral (33); The BCI module (10) is used to collect, record and analyze brain electrical signals, and transmit real-time data to the control module (20); The control module (20) is used to extract and classify the features of the brain electrical signal, accurately match the brain electrical signal with the wheelchair action instruction, and control the specific action of the wheelchair body (31); The device terminal (30) is used as a means of transportation for users with limited mobility, and ensures the safety of the transportation behavior during the transportation behavior.

2. A brainwave-controlled system and wheelchair according to claim 1, characterized in that: The electroencephalogram signal acquisition unit (12) is used to collect task electroencephalogram signals from the brain of a user sitting on the wheelchair body (31), and transmit the signals to the brain-like chip (21) in real time through the electroencephalogram signal transmission unit (13); During the acquisition and processing process of the electroencephalogram signal acquisition unit (12), a regularization term is introduced to select electroencephalogram signal features using a Lasso regression model, thereby reducing redundant electroencephalogram signal features.

3. A brainwave-controlled system and wheelchair according to claim 2, characterized in that: The task EEG signals are mainly generated by the user through brain activities and imagination. The task EEG signals mainly include the following four types: Task 1: The EEG signal generated by the user imagining the wheelchair moving forward; Task 2: The EEG signal generated by the user imagining the wheelchair turning left; Task 3: The EEG signal generated by the user imagining the wheelchair turning to the right; Task 4: The EEG signals generated by the user imagining the wheelchair stopping and moving.

4. The brainwave-controlled system and wheelchair according to claim 2, characterized in that: The specific formula of the Lasso regression model is as follows: in, n: is the number of samples of EEG signal features; p: is the number of EEG signal features; y i : is the response variable of the i-th sample of the EEG signal feature; x ij : is the jth eigenvalue of the i-th sample of the EEG signal feature; β0: is the intercept term; β j : is the coefficient of the jth feature of the EEG signal; λ is a regularization parameter, which is used to control the strength of regularization. The larger the λ is, the stronger the regularization effect is and the simpler the model is. On the contrary, the smaller the λ is, the weaker the regularization effect is and the more complex the model is. |β j |: indicates β j This is the key part of L1 regularization, which makes the unimportant EEG signal feature coefficients tend to 0, thereby realizing EEG signal feature selection.

5. The brainwave-controlled system and wheelchair according to claim 1, characterized in that: During the fitting process of the Lasso regression model, the mean absolute error (MAE) is calculated to determine the model fitting error and evaluate the model fitting effect; The mean absolute error (MAE) is shown below: in, n: is the number of EEG signal feature samples, y i : is the true value of the i-th sample of the EEG signal feature; is the predicted value of the i-th sample of the EEG signal feature; The smaller the MAE, the more accurate the model prediction is, that is, the smaller the difference between the predicted value and the actual value is.

6. The brainwave-controlled system and wheelchair according to claim 1, characterized in that: The EEG signal feature extraction unit (22) receives in real time the filtered and recorded task EEG signals sent by the EEG signal transmission unit (13), and classifies the feature signals, i.e., the four task EEG signals, through the EEG signal feature classification unit (23), and accurately matches the four task EEG signals with the action instructions of the wheelchair body (31).

7. The brainwave-controlled system and wheelchair according to claim 6, characterized in that: The task EEG signal is classified and matched to the action instruction by the brain-like chip (21) as follows: Type 1 signal: The EEG signal generated by the user's imagination of the wheelchair moving forward matches the wheelchair moving forward; Type II signal: The EEG signal generated by the user imagining the wheelchair turning left matches the wheelchair moving a certain distance to the left; The third type of signal: The EEG signal generated by the user's imagination of the wheelchair turning to the right matches the wheelchair moving a certain distance to the right; The fourth type of signal: The EEG signal generated by the user's imagination of the wheelchair stopping and moving matches the wheelchair stopping.

8. The brainwave-controlled system and wheelchair according to claim 1, characterized in that: The first wheelchair peripheral (32) is a visual sensor, and the second wheelchair peripheral (33) is a laser radar; the first wheelchair peripheral (32) and the second wheelchair peripheral (33) are connected to the host computer (IPC) via electrical signals; The first wheelchair peripheral device (32) and the second wheelchair peripheral device (33) jointly formulate and generate a wheelchair stop movement signal; When the visual sensor and the laser radar detect that there are no obstacles ahead, the host computer (IPC) normally executes the generated signals of type 1 to 4; When the visual sensor and the laser radar detect an obstacle ahead, the host computer (IPC) preferentially executes the first wheelchair peripheral (32) and the second wheelchair peripheral (33) to jointly generate a wheelchair stop motion signal.

9. The brainwave-controlled system and wheelchair according to claim 1, characterized in that: An adjustment assembly (40) is provided on one side of the wheelchair body (31), and the adjustment assembly (40) comprises a support rod (41) provided on one side of the wheelchair body (31), a limit block (42) provided on one side of the support rod (41), an adjustment rod (43) penetrating one end of the limit block (42) and slidably connected thereto, an adjustment screw (44) embedded on one side of the limit block (42), and a placement horizontal plate (45) provided on a side wall of one end of the adjustment rod (43) and rotatably connected thereto; The first wheelchair peripheral device (32) and the second wheelchair peripheral device (33) are arranged on the top of the placement horizontal plate (45).

10. The brainwave-controlled system and wheelchair according to claim 9, characterized in that: A limit stopper rod (431) is provided at the top of one end of the adjusting rod (43) close to the placement transverse plate (45), and a limit support rod (432) is provided at one side of the adjusting rod (43) and at the bottom of the placement transverse plate (45).

Images