Lower extremity mixed motion exoskeleton training system and virtual reality device

Abstract

The present disclosure provides a lower limb mixed motion exoskeleton training system, which can be applied to the technical field of lower limb training. The lower limb mixed motion exoskeleton training system comprises a display, an exoskeleton, a collector and a controller. In a first stage, the controller controls the display to display a baseline collection image and reads a baseline electroencephalogram signal from the collector. In a second stage, the controller controls the display to display a first image in a first step image set in a first display area of the display and alternately switch a plurality of first images at a first switching frequency, so as to repeatedly display a walking motion corresponding to a first lower limb at a first walking frequency. The controller reads a sensory motor area electroencephalogram signal of a sensory motor area in the second stage from the collector, determines a target wearing part from the first wearing part and the second wearing part based on the extracted occipital electroencephalogram features, and drives the target wearing part according to the sensory motor area electroencephalogram signal and the baseline electroencephalogram signal, so as to drive a target lower limb of a target object wearing the target wearing part to move.

Description

Technical Field

[0001] This disclosure relates to the field of lower limb training technology, and more specifically, to a lower limb hybrid exoskeleton training system and virtual reality device. Background Technology

[0002] For patients with diseases such as stroke, there are some training systems available that can help them with rehabilitation.

[0003] However, this type of training system can impose a heavy training burden on users, resulting in a poor user experience. Furthermore, it also suffers from poor rehabilitation outcomes. Summary of the Invention

[0004] In view of the above problems, this disclosure provides a lower limb hybrid exoskeleton training system and virtual reality device to improve the accuracy of assisting patients in lower limb movement.

[0005] According to a first aspect of this disclosure, a lower limb hybrid exoskeleton training system is provided, comprising: a display; an exoskeleton including a first wearable part and a second wearable part, the first wearable part and the second wearable part being worn on different lower limbs of a target object; a collector for acquiring signals from the brain of the target object; the brain of the target object including a sensorimotor region and an occipital lobe region; and a controller for: in a first stage after receiving a start command, controlling the display to display a baseline acquired image and reading baseline EEG signals from the sensorimotor region in the first stage from the collector; and in a second stage after the first stage, controlling the display to display a first image from a first step of a walking image set in a first display area of ​​the display, and switching multiple first images in turn at a first switching frequency, thereby repeatedly displaying the walking steps corresponding to the first lower limb at a first walking frequency. The system simultaneously controls the display to show the second image from the second stepping image set in the second display area of ​​the display, and switches multiple second images in turn at a second switching frequency, thereby repeatedly displaying the walking action corresponding to the second lower limb at a second walking frequency; and reads the sensorimotor EEG signal from the sensorimotor area in the second stage from the collector. When the sensorimotor EEG signal indicates that the target object has the intention to move, feature extraction is performed on the occipital lobe EEG signal of the occipital lobe brain region according to the first switching frequency, the second switching frequency, the first walking frequency, and the second walking frequency, and the target wearable part is determined from the first wearable part and the second wearable part based on the extracted occipital lobe EEG features; and drives the target wearable part according to the sensorimotor EEG signal and the baseline EEG signal to drive the target object to move the target lower limb wearing the target wearable part.

[0006] A second aspect of this disclosure provides a virtual reality device, including the aforementioned lower limb hybrid exoskeleton training system.

[0007] According to embodiments of this disclosure, by switching a first image at a first switching frequency to repeatedly display the walking motion of the first lower limb at a first walking frequency, and switching a second image at a second switching frequency to repeatedly display the walking motion of the second lower limb at a second walking frequency, the cerebral cortex of the target object can be stimulated to reduce the difficulty of the target object's motor imagination, enhance the activation of the sensorimotor area, and induce occipital lobe EEG signals indicating the lower limb that the target object wishes to use for walking. Based on the occipital lobe EEG characteristics of the occipital lobe EEG signals, the target lower limb that the target object wishes to use for walking can be accurately determined from the first and second wearable parts. Furthermore, the target wearable part worn on the target lower limb can be driven to assist the target object in movement based on the sensorimotor area EEG signals and baseline EEG signals. Based on this, the lower limb that the target object wishes to move can be matched with the wearable part controlled by the exoskeleton, enhancing the user experience of the target object and reducing the cognitive burden of driving the lower limb. Furthermore, because it can accurately drive the target wearable part to move the target lower limbs, it can perform targeted lower limb training on the target, improving the training effect and enhancing the user experience of the target. Attached Figure Description

[0008] The foregoing contents, as well as other objects, features, and advantages of this disclosure, will become clearer from the following description of embodiments with reference to the accompanying drawings, in which:

[0009] Figure 1 A schematic diagram of the structure of a lower limb hybrid exoskeleton training system according to an embodiment of the present disclosure is shown.

[0010] Figure 2 A schematic diagram illustrating a walking motion according to an embodiment of the present disclosure is shown.

[0011] Figure 3a This illustration schematically shows a successful EEG activation result notification according to an embodiment of the present disclosure;

[0012] Figure 3b This illustration schematically shows another successful EEG activation result notification according to an embodiment of the present disclosure;

[0013] Figure 3c A schematic diagram illustrating a brainwave activation result failure prompt according to an embodiment of the present disclosure is shown.

[0014] Figure 4a A schematic diagram illustrating activation status information according to an embodiment of the present disclosure is shown.

[0015] Figure 4b A schematic diagram illustrating activation status information according to an embodiment of the present disclosure is shown.

[0016] Figure 5 A schematic diagram of a baseline acquisition image according to an embodiment of the present disclosure is shown.

[0017] Figure 6 This schematically illustrates an environmental image diagram with a non-fixed viewpoint according to an embodiment of the present disclosure;

[0018] Figure 7 This diagram schematically illustrates a training pattern for lower limb movements of the target subject; and

[0019] Figure 8 A schematic diagram of another lower limb hybrid exoskeleton training system is shown. Detailed Implementation

[0020] The embodiments of the present disclosure will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the disclosure. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of the present disclosure for ease of explanation. However, it will be apparent that one or more embodiments may be practiced without these specific details. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concepts of the present disclosure.

[0021] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit this disclosure. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0022] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0023] When using expressions such as "at least one of A, B and C", they should generally be interpreted in accordance with the meaning that is commonly understood by those skilled in the art (e.g., "a system having at least one of A, B and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having A and B, a system having A and C, a system having B and C, and / or a system having A, B and C, etc.).

[0024] In realizing the inventive concept of this disclosure, a brain-computer interface paradigm with output control command function is used to train the lower limbs of the target object to promote the neuroplasticity of the target object, thereby enabling the lower limbs to regain motor function. The target object can refer to a user with a need for lower limb training.

[0025] For example, brain-computer interface paradigms with output control command functions include: Steady-State Visual Evoked Potential (SSVEP) brain-computer interface paradigm and Motor Imagery (MI) brain-computer interface paradigm.

[0026] The SSVEP brain-computer interface paradigm has high classification accuracy, but it lacks naturalness in lower limb motor control and is difficult to effectively activate mirror neurons, resulting in poor recovery of lower limb motor function and easy visual fatigue for the target.

[0027] The MI brain-computer interface paradigm is relatively abstract, making it difficult for subjects to implement and stably extract the limb movement features of the target object. Furthermore, the accuracy of left and right lower limb classification based solely on motor imagery features is low, which can easily lead to the target object losing confidence in the recovery of lower limb motor function. This causes the target object to focus on distinguishing the start or stop of walking, resulting in a mismatch between the exoskeleton's movement and the target object's expectations.

[0028] Furthermore, the SSVEP and MI brain-computer interface paradigms can be combined to train the lower limbs of target subjects. However, these two paradigms are difficult to adapt and match, increasing the task burden on the target subjects. Moreover, the brain-computer interface paradigm based on the combination of SSVEP and MI can only be performed suspended in the air or in a restricted space. In addition, the external device stimulation pattern based on the SSVEP and MI paradigm is fixed, making it easy for the target subjects to shift from active participants to passive executors, thus making it difficult to maintain high levels of cortical activation throughout the lower limb movement training process, resulting in poor experience and rehabilitation outcomes for the target subjects.

[0029] In view of this, embodiments of the present disclosure provide a lower limb hybrid exoskeleton training system and virtual reality device, which can better combine SSVEP and MI, improve the stability and accuracy of extracting the lower limb movement features of the target object, reduce the conditional restrictions of walking training, enhance the participation of the target object, and thus effectively improve the target object's experience and rehabilitation effect.

[0030] Embodiments of this disclosure provide a lower limb hybrid exoskeleton training system. Figure 1 A schematic diagram of a lower limb hybrid exoskeleton training system according to an embodiment of the present disclosure is shown. Figure 1As shown, the lower limb hybrid exoskeleton training system includes: a display 100, an exoskeleton 200, a data acquisition unit 300, and a controller 400. It should be understood that the controller 400 can be electrically connected to at least one of the display 100, the exoskeleton 200, or the data acquisition unit 300, or can also be communicatively connected. For example, communication can be established via Transmission Control Protocol / Internet Protocol (TCP / IP) or Bluetooth, etc., without limitation herein.

[0031] According to embodiments of this disclosure, the exoskeleton 200 includes a first wearable portion 201 and a second wearable portion 202, which are respectively worn on different lower limbs of a target object. For example, the first lower limb may be the target object's left leg, and the second lower limb may be the target object's right leg. The first wearable portion 201 may be a wearable portion for wearing on the target object's left leg, and the second wearable portion 202 may be a wearable portion for wearing on the target object's right leg.

[0032] According to embodiments of this disclosure, the collector 300 is used to collect signals from the sensorimotor region and occipital lobe region of the target brain 500.

[0033] According to embodiments of this disclosure, the controller 400 can be a terminal device such as a host computer. The controller 400 can be used to, in a first stage after receiving a start command, control the display 100 to display baseline acquired images and read baseline EEG signals from the sensorimotor area from the acquisition unit 300 during the first stage; in a second stage following the first stage, control the display 100 to display a first image from a first step-walking image set in a first display area of ​​the display 100, and alternately switch multiple first images at a first switching frequency, thereby repeatedly displaying the walking motion corresponding to the first lower limb at a first walking frequency; simultaneously, control the display 100 to display a second image from a second step-walking image set in a second display area of ​​the display 100, and alternately switch multiple second images at a second switching frequency, thereby repeatedly displaying the second step-walking motion at a second walking frequency. The system displays walking movements corresponding to the second lower limb; and reads the sensorimotor brainwave signals (e.g., MI signals) from the sensorimotor area during the second stage from the collector 300. When the sensorimotor brainwave signals indicate that the target object has the intention to move, the system extracts features from the occipital lobe brainwave signals (e.g., SSVEP signals) based on the first switching frequency, the second switching frequency, the first walking frequency, and the second walking frequency. Based on the extracted occipital lobe brainwave features, the system determines the target wearable part from the first wearable part 201 and the second wearable part 202. Based on the sensorimotor brainwave signals and the baseline brainwave signals, the system drives the target wearable part to move the target lower limb of the target object wearing the target wearable part.

[0034] According to embodiments of this disclosure, the display 100 can be a display for a virtual reality (VR) device. For example, a VR device can be VR glasses. The exoskeleton 200 is a wearable mechanical device used to assist in the movement of a target's limbs. The data acquisition device 300 is a multimodal data acquisition device built based on the wearable multimodal research platform (NeuroHUB).

[0035] According to embodiments of this disclosure, the data acquisition device 300 includes a 64-lead EEG cap, a host computer used as a data acquisition terminal, an event synchronization box, and a router, wherein the lead positions of the data acquisition device 300 conform to the 10-20 lead standard. The event synchronization box and the data acquisition terminal are both on the same network. The acquisition device 300 has 36 acquisition channels with an acquisition frequency of 1000 Hz. The specific locations of the acquisition channels corresponding to the brain regions may include FCz, FC1, FC2, FC3, FC4, FC5, FC6, FT7, FT8, Cz, C1, C2, C3, C4, C5, C6, T7, T8, CP1, CP2, CP3, CP4, CP5, CP6, TP7, TP8, POz, PO3, PO4, PO5, PO6, PO7, PO8, Oz, O1, and O2.

[0036] The following example illustrates the training process of the lower limb movement of a target subject using the Lower Limb Hybrid Exoskeleton 200 Training System.

[0037] According to an embodiment of this disclosure, in the first stage after receiving a start command, the controller 400 controls the display 100 to display a baseline acquisition image according to the start command. This induces the sensorimotor region of the target object to generate a baseline EEG signal. In one embodiment of this disclosure, the baseline EEG signal can serve as a reference signal for the sensorimotor region EEG signal. The sensorimotor region EEG signal can be used to detect the activation level of the motor brain region of the target object. Subsequently, the controller 400 can read the baseline EEG signal of the sensorimotor region from the acquisition unit 300 during the first stage. Further, the baseline EEG signal can be used as a reference signal to detect the activation level of the target object's lower limbs. Specifically, based on the baseline EEG signal and the sensorimotor region EEG signal, an activation index value for the sensorimotor region of the target object's lower limbs can be determined. This activation index value can be used to determine the lower limb activation state of the target object. For example, if the activation index value is less than a predetermined index value, it can be determined that the target object's lower limbs are not effectively activated; thus, the target object's lower limbs can be activated by means of electrical stimulation or the like. If the activation index value is greater than or equal to the predetermined index value, it can be determined that the target object's lower limbs are effectively activated. In this way, the target wearable device can be controlled to assist the target object's lower limbs in movement.

[0038] Figure 2 A schematic diagram illustrating a walking motion according to an embodiment of the present disclosure is shown. Figure 2 As shown, the controller can display a walking scene animation (i.e., the first and second images mentioned above) on the display. The first image is switched at a first switching frequency to demonstrate the changes in the walking motion of the first lower limb to the target object. Simultaneously, the second image is switched at a second switching frequency to demonstrate the changes in the walking motion of the second lower limb to the target object. In one embodiment, the first display area and the second display area can be different areas located on the same display (e.g., the left and right areas of the display divided by a vertical center line, respectively). The first switching frequency can be the switching frequency between different first images, and the second switching frequency can be the switching frequency between different second images.

[0039] According to embodiments of this disclosure, the occipital lobe region is located in the primary visual cortex of the brain, responsible for receiving visual signals from the retina. Therefore, by extracting features from the occipital lobe EEG signals in the second stage, the characteristics of the visual signals of the first and second images displayed on the monitor in the brain can be obtained, i.e., occipital lobe EEG features. Based on these occipital lobe EEG features, it can be determined whether the target object is focusing on the first or second image. Thus, if it is determined that the target object is focusing on the first image, the first lower limb can be identified as the target lower limb. If it is determined that the target object is focusing on the second image, the second lower limb can be identified as the target lower limb. Based on this, if the first lower limb is the target lower limb, the first wearable part can be identified as the target wearable part. If the second lower limb is the target lower limb, the second wearable part can be identified as the target wearable part.

[0040] According to embodiments of this disclosure, the aforementioned sensorimotor region is a region in the cerebral cortex responsible for integrating somatic sensory information and coordinating motor functions. By simultaneously acquiring sensorimotor EEG signals and occipital lobe EEG signals, it can be determined whether the target object intends to move its lower limbs after viewing the first and second images displayed on the monitor. Thus, if the sensorimotor EEG signals indicate that the target object desires to move its lower limbs, the corresponding target wearable device can be controlled to move the target lower limb based on the target lower limb indicated by the occipital lobe EEG signals.

[0041] According to embodiments of this disclosure, by switching a first image at a first switching frequency to repeatedly display the walking motion of the first lower limb at a first walking frequency, and switching a second image at a second switching frequency to repeatedly display the walking motion of the second lower limb at a second walking frequency, the occipital lobe of the target brain can be stimulated to induce occipital lobe EEG signals indicating the lower limb the target subject wishes to use for walking. Based on the occipital lobe EEG characteristics of the occipital lobe EEG signals, the target lower limb wished to be used for walking can be accurately determined from the first and second wearable devices. Thus, the target wearable device worn on the target lower limb can be driven to assist the target subject in movement based on the sensorimotor EEG signals and baseline EEG signals. Based on this, the lower limb the target subject wishes to move can be matched with the wearable device controlled by the exoskeleton, enhancing the user experience and reducing the cognitive burden of driving the lower limb. Furthermore, since the target wearable device can be accurately driven to move the target lower limb, targeted lower limb training can be performed on the target subject, improving training effectiveness and enhancing the user experience.

[0042] According to embodiments of this disclosure, the collector may include a first buffer and a second buffer. The first buffer is used to store sensorimotor EEG signals collected from the sensorimotor region at different times in the form of different first data points, and the second buffer is used to store occipital lobe EEG signals collected from the occipital lobe region at different times in the form of different second data points.

[0043] According to embodiments of this disclosure, the first buffer and the second buffer can be circular buffers, specifically 20-second circular buffers, but not limited to this. The first buffer and the second buffer can receive the first data point and the second data point in real time, respectively. A write pointer can be set to indicate the position of the last write. For at least one of the first buffer or the second buffer, based on the existing data points, if the amount of data at the pointer position plus the amount of data written this time is not greater than the length of the corresponding buffer, a new data point is added after the address corresponding to the pointer to overwrite the original zero matrix, and the write pointer position is updated; if it is greater than the length of the corresponding buffer, the buffer is first filled, and the remaining data points are then used to overwrite the buffer from the beginning, and the pointer position is updated. Based on this, a counting function can be set to return the current data update amount each time it is called. Furthermore, a read function can be set to lock the buffer with a thread lock each time it is read, read the current data in the buffer, release the buffer, and return the data in the buffer. In addition, a reset function can be set, which, when the function is called, resets the buffer to a zero matrix with the shape of channel × number of sampled data points, and resets the data update amount and the write pointer to 0. For data reading, a read pointer can be set to indicate the last read position. Furthermore, a windowing function can be set, taking into account the data points in the buffer, window length, step size, read pointer address, and read type. When read type 1, it returns the first data point of the specified window length starting from the current pointer position for MI classification (i.e., determining whether the target object has a motion intention); when read type 2, it returns all the second data points from the initial pointer position to the end of the current window for SSMVEP classification (i.e., determining the target lower limb that the target object expects to move).

[0044] According to embodiments of this disclosure, the controller is further configured to, while controlling the display to show the baseline acquisition image for a predetermined duration, determine that the first stage for acquiring baseline EEG signals has been completed, and determine that a second stage following the first stage can proceed. The predetermined duration can be set to 1 second, etc., as needed. Simultaneously, a stage start tag can be sent to the acquisition unit, so that the acquisition unit uses the start time of receiving the stage start tag as the start time for acquiring the first and second data points. Based on this, the order of the addresses of the first and second buffers can be associated with the acquisition times of the first and second data points, respectively. Therefore, the address corresponding to the start time can be used as the starting address; in other words, the first buffer address corresponding to the sending time of the stage start tag in the first buffer can be used as the starting address.

[0045] According to embodiments of this disclosure, the amount of data at a first data point at an address following the starting address of the first buffer can be monitored based on the starting address to determine the amount of data updates at addresses corresponding to multiple times after the transmission time in the first buffer.

[0046] According to embodiments of this disclosure, when the data update amount is greater than a predetermined data amount determined based on a predetermined duration and a predetermined sampling rate, an initial read address for the first buffer is determined based on the starting address, the visual pathway transit time, and the predetermined sampling rate. The aforementioned visual pathway transit time (e.g., 0.14 seconds) represents the time it takes for the visual signal from the target object's retina to reach the aforementioned occipital lobe brain region. For example, the product of the visual pathway transit time and the predetermined sampling rate can be used as the address change, and the read address is obtained based on the starting address and the address change. It is understood that after the target object sees the first or second image, the visual signal will be transmitted to the brain region. This transmission duration is the visual pathway transit time. During this process, the collector may collect invalid signals before the visual signal reaches the brain region. Based on this, by re-determining the read address in the above manner to read the first data point from the first buffer, the validity of the read first data point can be guaranteed. Furthermore, state detection can be performed on the first data point read from the first buffer based on the initial read address, and if the detection result indicates that the target object is in a state of motor imagination, multiple target second data points can be read from the second buffer based on the initial read address. Based on multiple target second data points, occipital lobe EEG features were extracted.

[0047] According to embodiments of this disclosure, the first and second buffer areas are disposed in the data acquisition unit. The first buffer area is a buffer region for storing data (referred to as first data points) of sensorimotor EEG signals, and the second buffer area is a buffer region for storing data (referred to as second data points) of occipital lobe EEG signals. The aforementioned stage start tag is used to provide the data acquisition unit with a start signal to begin monitoring the data update amount of sensorimotor EEG signals.

[0048] According to embodiments of this disclosure, the aforementioned starting address represents the starting address from which the data collector collects the first data point, and this starting address can be used as the starting address for monitoring the data update amount. The sampling rate represents the frequency at which the data collector performs sampling.

[0049] According to embodiments of this disclosure, when a phase initiation tag is sent to the first cache address (starting address) of the first cache area, the data acquisition unit begins monitoring the data update amount of the sensorimotor EEG signal. If the data update amount is greater than a predetermined amount, it indicates that the amount of sensorimotor EEG signal data stored in the first cache area is sufficient to support detecting whether the target object wants to perform lower limb movements while viewing the first and second images displayed on the monitor. Based on this, the data acquisition unit can begin reading the sensorimotor EEG signal from the initial read address of the first cache area. The data acquisition unit can determine the data update amount by calling a statistical function of the data receiving class. The predetermined data amount is the product of a predetermined duration and a predetermined sampling rate.

[0050] According to embodiments of this disclosure, the collector performs state detection on the read sensorimotor EEG signals. When the target object is detected to have a walking intention, the data of the target occipital lobe EEG signal (target second data point) can be read from the second buffer, thereby obtaining the occipital lobe EEG features of the target object corresponding to the current motor imagination state.

[0051] According to embodiments of this disclosure, the acquisition device can read EEG signals from the sensorimotor cortex through acquisition channels at locations such as FCz, FC1, FC2, FC3, FC4, FC5, FC6, FT7, FT8, Cz, C1, C2, C3, C4, C5, C6, T7, T8, CP1, CP2, CP3, CP4, CP5, CP6, TP7, and TP8. The frequency band settings can be 4 Hz~8 Hz, 8 Hz~13 Hz, 13 Hz~30 Hz, etc. After performing Mutual Information-based Best Individual Feature (MIBIF) selection and normalization on the sensorimotor EEG signals, the data can be fed into a pre-trained Support Vector Machine (SVM) model for classification to determine whether the target object is in a state of motor imagery. In this embodiment of the disclosure, a Butterworth fourth-order filter can be used to perform bandpass filtering of the first data point from 1 to 40 Hz, downsample to 512 Hz, and perform mean-variance normalization to achieve the above-mentioned normalization operation on the first data point.

[0052] According to embodiments of this disclosure, the acquisition device can read the target occipital lobe EEG signal through acquisition channels at positions POz, PO3, PO4, PO5, PO6, PO7, PO8, Oz, O1, and O2, with frequency band settings of 1Hz~8Hz, 8Hz~13Hz, or 13Hz~20Hz, etc. Subsequently, feature extraction can be performed on the target occipital lobe EEG signal based on the CCA-TRCA joint algorithm to obtain the occipital lobe EEG features of the target object corresponding to the current motor imagery state. For example, the aforementioned Butterworth fourth-order filter can be used to normalize the target second data point, followed by feature extraction from the normalized target second data point; details are omitted here.

[0053] According to embodiments of this disclosure, by setting a data update amount greater than a predetermined data amount determined based on a predetermined duration and a predetermined sampling rate, the first buffer can store as many valid first data points as possible, thereby fully arousing the target object's movement intention. Simultaneously, the second buffer can store as many valid second data points as possible, improving the stability of occipital lobe EEG feature extraction and thus improving the accuracy of determining the target object's state.

[0054] According to embodiments of this disclosure, the addresses of the first buffer and the second buffer are associated with each other. For example, the read address of the first buffer may correspond to the read address of the second buffer. The controller is further configured to, for the kth time window, start from the kth read address and end at the kth last address corresponding to the last moment of the kth time window, read the kth group of first data points from the first buffer and perform state detection to obtain the kth detection result; k is an integer greater than 1; wherein, when k=2, the kth read address is determined based on the initial read address.

[0055] According to embodiments of this disclosure, when one data point is collected per second, the address can be shifted one position to the right accordingly. Thus, within each time window, the address can shift multiple positions over time, and each address can store a corresponding data point. Based on this, the corresponding address change can be determined according to the first time window, and based on this address change and the initial read address, the last address of the first time window is determined as the second read address of the second time window, and this serves as the starting point for reading the first data point collected in the second time window. Subsequent time windows follow the same principle and will not be elaborated further. Further, if the k-th detection result indicates that the target object is in a motion-imagined state, multiple target second data points are read from the second buffer based on the initial read address and the k-th last address. For example, if the target object is detected to be in a motion-imagined state within three consecutive time windows, the target second data points can be read from the second buffer based on the first read address of the first time window and the third last address of the third time window within those three consecutive time windows. The purpose of this is to fully acquire the first data point collected within the same time period and all target second data points corresponding to that first data point. In other words, it involves acquiring the target second data points corresponding to the lower limbs that are expected to be driven when the target object is expected to move, so as to accurately and timely drive the target wearable device and improve the user experience of the target object.

[0056] According to embodiments of this disclosure, the aforementioned time window has a fixed window length, and corresponding sensorimotor EEG signals can be acquired within each time window. The aforementioned state detection is performed within each time window. When the target object is detected to be in a motor imagery state, the occipital lobe EEG signals in the second buffer can be read to extract occipital lobe EEG features.

[0057] According to embodiments of this disclosure, after detecting that the target object is in a state of motor imagery, occipital lobe EEG signals are read from the second buffer. These occipital lobe EEG signals are generated when the target object has already expressed a motor intention. Thus, based on the above method, occipital lobe EEG signals closely related to the target object's state of motor imagery can be extracted with sufficient accuracy, thereby improving the accuracy of occipital lobe EEG feature extraction.

[0058] According to an embodiment of this disclosure, the controller is further configured to update the k-th read address to the (k+1)-th read address based on a predetermined step size when the k-th detection result indicates that the target object is not in a motion-imagined state, so as to read data from the first buffer starting from the (k+1)-th read address to perform state detection; and drive the target wearable unit according to a predetermined strategy when k+1=n and the (k+1)-th detection result indicates that the target object is not in a motion-imagined state; where n is an integer greater than k.

[0059] According to an embodiment of this disclosure, the data detection within a time window is illustrated below. First, the maximum number of detections is set to 3. If the current time window number is 1, then starting from reading the first data point, the EEG signal of the sensorimotor area is read with a window length of 4 seconds and a step size of 1 second.

[0060] When the aforementioned data update volume reaches the window length of 4 seconds, the data reading thread is invoked to read the EEG signals from the sensorimotor cortex within the time window, performing a motor imagery state detection on the target object. The detection results are sent to the controller via a synchronization box for subsequent analysis, and the reading pointer is moved from the previous reading address to the starting point of the next time window. The interval between each time window is the product of a predetermined step size and a predetermined sampling rate.

[0061] According to embodiments of this disclosure, when the target object is detected to be in a state of motion visualization, the lower limb hybrid exoskeleton training system switches the state to a feedback state. Figure 3a A schematic diagram illustrating a successful EEG activation result notification according to an embodiment of this disclosure is shown. Figure 3a As shown, the display shows (for example, the display's feedback display component may show): EEG activation successful, currently activating the right side (currently activating the right leg); or, Figure 3b This illustration schematically shows another successful EEG activation result notification according to an embodiment of the present disclosure. Figure 3b As shown, the display shows: EEG activation successful, currently activating the left side (currently activating the left leg). The content displayed by the feedback display component is the activation status information, i.e., the text indicating the activation status of the EEG signal.

[0062] According to embodiments of this disclosure, if the target object is not identified as being in a motion-imagination state within any of the three time windows, the lower limb hybrid exoskeleton training system will also switch to the feedback state. To avoid the target object standing still for a long time and failing to receive timely and effective training, the controller can still activate the exoskeleton to drive the target object to move. Figure 3c A schematic diagram illustrating a brainwave activation result failure prompt according to an embodiment of the present disclosure is shown. Figure 3cAs shown, the monitor displays: EEG activation failed, system startup on the right.

[0063] According to embodiments of this disclosure, the product of the step size and the predetermined sampling rate can precisely control the sliding of the reading address to the starting point of the next time window, reducing omissions in reading EEG signals from the sensorimotor area and achieving continuity in data reading. By controlling the reading address according to the product of the predetermined step size and the predetermined sampling rate, the data within each time window can have clear start and end addresses on the time axis, thereby improving the accuracy and effectiveness of data reading.

[0064] According to embodiments of this disclosure, the target wearable unit can also be used to electrically stimulate the target lower limb. The processor is further configured to: calculate a baseline power spectrum based on the baseline EEG signal when the k-th detection result indicates the target object is in a state of motor imagery; and read H groups of target first data points from a first buffer based on H groups of target addresses following the k-th last address (e.g., these can be consecutive H groups of target addresses, which may be adjacent to the k-th last address, but may not be adjacent; this disclosure does not limit this). The H groups of target addresses can be consecutive H groups of target addresses following the k-th last address, and each group of target addresses may include at least one target address. Correspondingly, each group of target first data points may include at least one target data point. Each group of target addresses may correspond to a predetermined time window following the k-th time window, which will not be elaborated here. Subsequently, H target power spectra can be calculated based on the H groups of target first data points. For the h-th target power spectrum among H target power spectra, the activation value of the h-th lower limb brain region of the target subject is calculated based on the baseline power spectrum and the h-th target power spectrum. Then, the h-th activation index value is calculated based on the h-th lower limb brain region activation value and the target activation value. h is a positive integer, and H is an integer greater than or equal to h. Thus, H activation index values ​​are obtained. Based on the H activation index values, the activation status information of the display in the third display area is controlled to indicate the lower limb activation status to the target subject. Furthermore, if all H activation index values ​​are less than a predetermined index value, the electrical stimulation intensity of the target wearable part is increased according to the stimulation adjustment step size to provide enhanced stimulation to the target subject within a predetermined time period, and the electrical stimulation intensity is restored after the predetermined time period. The predetermined time period here can be a period after 0.5 seconds or 1 second, etc., and will not be elaborated further here.

[0065] According to embodiments of this disclosure, the calculation process of activation index values ​​is illustrated below. When training a target subject, the lower limb hybrid exoskeleton training system converts the baseline EEG signal from the time domain to the frequency domain, and then analyzes the energy distribution of the baseline EEG signal at different frequencies to obtain the aforementioned baseline power spectrum.

[0066] According to an embodiment of this disclosure, since the target first data point corresponds to the sensorimotor EEG signal within the predetermined time window, the sensorimotor EEG signal can be converted from the time domain to the frequency domain, and the energy distribution of the sensorimotor EEG signal at different frequencies can be analyzed to obtain the target power spectrum.

[0067] According to embodiments of this disclosure, for each target first data point, based on the aforementioned baseline power spectrum and target power spectrum, the calculation formula for the activation value of the lower limb brain region of the target object is as follows:

[0068] (1)

[0069] in, Represents the baseline power spectrum. This represents the target power spectrum calculated based on the first data point of the target. This represents the activation value of the lower limb brain region of the target subject. t represents time.

[0070] According to embodiments of this disclosure, a higher activation value in the lower limb brain region of the target subject indicates a greater degree of activation in that region (e.g., sensorimotor area). This activation level is represented by a target activation value, which is initially set to 0. If the activation value in the lower limb brain region of the target subject exceeds the target activation value, the activation value is updated. The formula for calculating the activation index value is as follows:

[0071] (2)

[0072] in, Indicates the target activation value. This indicates the activation value of the lower limb brain region of the target subject. This indicates the activation index value.

[0073] Figure 4a A schematic diagram illustrating activation status information according to an embodiment of this disclosure is shown. Figure 4a As shown, when the activation index value is lower than the predetermined index value, the controller sends the first activation status feedback information to the display, and the display then shows in the third display area: "Activation status information: The current status is low, please pay attention." The predetermined index value can be 0.75.

[0074] According to embodiments of this disclosure, when multiple consecutive activation index values ​​are less than a predetermined index value, for example, when five consecutive activation index values ​​are less than 0.75, the electrical stimulation intensity of the target wearable part is increased according to the stimulation adjustment step size, for example, by increasing the stimulation adjustment step size by 1 unit. During a predetermined time period, the target object is given enhanced stimulation by increasing the electrical stimulation intensity of the target wearable part, and the electrical stimulation intensity is restored after the predetermined time period, thereby assisting the target object in focusing their attention. The predetermined time period can be 5 seconds, and the stimulation adjustment step size is less than the highest tolerable electrical stimulation threshold.

[0075] Figure 4b A schematic diagram illustrating activation status information according to an embodiment of this disclosure is shown. Figure 4b As shown, when the activation index value is in the range of 0.75~1, the controller sends a second activation status feedback message to the display, and the display shows in the third display area: the activation status information is good, please continue.

[0076] According to embodiments of this disclosure, by using activation index values ​​to reflect the lower limb activation state of the target subject, and by providing enhanced stimulation to the target subject to assist in focusing attention, the neural plasticity of the target subject can be accelerated, thereby improving the efficiency of lower limb motor training. Furthermore, restoring the electrical stimulation intensity after a predetermined period can reduce overstimulation of the target subject's lower limbs, improving the safety of the target subject during lower limb motor training.

[0077] According to embodiments of this disclosure, the CCA-TRCA joint algorithm consists of two parts: Filter Bank Canonical Correlation Analysis (FBCCA) and Filter Bank Task-Related Component Analysis (FBTRCA). The CCA class (i.e., canonical correlation analysis class) template signal in the CCA part can be generated by the controller. For example, the controller can calculate harmonics (e.g., 1x or 2x) for the first switching frequency and the second switching frequency respectively, obtaining the first harmonic frequency and the second harmonic frequency. It should be understood that the first harmonic frequency may include 1x and 2x harmonics of the first switching frequency. The second harmonic frequency may include 1x and 2x harmonics of the second switching frequency. Based on the first switching frequency and the first step frequency, a first target frequency is obtained. Specifically, the first switching frequency and the first step frequency can be added to obtain a first upper limit frequency, and the first switching frequency and the first step frequency can be subtracted to obtain a first lower limit frequency. Based on this, the first upper limit frequency and the first lower limit frequency can be determined as the first target frequency. Subsequently, a second target frequency can be obtained based on the second switching frequency and the second walking frequency. Specifically, the second switching frequency and the second walking frequency can be added together to obtain a second upper limit frequency, and the second switching frequency and the second walking frequency can be subtracted to obtain a second lower limit frequency. Based on this, the second upper limit frequency and the second lower limit frequency can be determined as the second target frequency. Based on the first harmonic frequency, the second harmonic frequency, the first target frequency, and the second target frequency, a CCA-type template signal is generated.

[0078] According to embodiments of this disclosure, a first 1x orthogonal template pair can be generated based on a harmonic of the first switching frequency (e.g., the orthogonal template pair is a template pair constructed based on sine and cosine, and so on), and a first 2x orthogonal template pair can be generated based on a harmonic of the first switching frequency. A first upper limit orthogonal template pair is generated based on a first upper limit frequency. A first lower limit orthogonal template pair is generated based on a first lower limit frequency. A first lower limit orthogonal template pair is generated based on the first 1x orthogonal template pair, the first 2x orthogonal template pair, the first upper limit orthogonal template pair, and the first lower limit orthogonal template pair, a CCA-class first template matrix is ​​constructed.

[0079] According to embodiments of this disclosure, a second 1x orthogonal template pair can be generated based on a harmonic frequency equal to one time the second switching frequency, and a second 2x orthogonal template pair can be generated based on a harmonic frequency equal to two times the second switching frequency. A second upper limit orthogonal template pair can be generated based on a second upper limit frequency. A second lower limit orthogonal template pair can be generated based on a second lower limit frequency. A second lower limit orthogonal template pair can be generated based on a second lower limit frequency. Based on the second 1x orthogonal template pair, the second 2x orthogonal template pair, the second upper limit orthogonal template pair, and the second lower limit orthogonal template pair, a second template matrix of CCA class is constructed. Subsequently, the first correlation of CCA class between the first template matrix of CCA class and multiple target second data points, and the second correlation of CCA class between the second template matrix of CCA class and multiple target second data points can be calculated respectively. Based on the first correlation of CCA class and the second correlation of CCA class, the first part of the occipital lobe EEG features is obtained.

[0080] According to embodiments of this disclosure, the occipital lobe EEG algorithm for the FBTRCA portion can be trained offline to obtain the first template signal of the TRCA class (i.e., task-related component analysis class) for the first image and the second template signal of the TCRA class for the second image. For example, signals from the occipital lobe can be collected while the target object is gazing at the first image, and then the TCRA algorithm can be used to extract the first template signal of the TCRA class from the signal. Similarly, signals from the occipital lobe can be collected while the target object is gazing at the second image, and then the TCRA algorithm can be used to extract the second template signal of the TCRA class from the signal. Subsequently, the first correlation of the TCRA class between the first template signal of the TCRA class and multiple target second data points can be calculated, the second correlation of the TCRA class between the second template signal of the TCRA class and multiple target second data points can be calculated, and the second part of the occipital lobe EEG features can be obtained based on the first and second correlations of the TCRA class.

[0081] Based on this, the occipital lobe EEG features from the first part and the second part can be weighted and summed according to predetermined weights to obtain accurate occipital lobe EEG features. For example, the first correlation of CCA class and the first correlation of TCRA class can be weighted and summed according to predetermined weights to obtain a first summation result, and the second correlation of CCA class and the second correlation of TCRA class can be weighted and summed to obtain a second summation result. The occipital lobe EEG features can then be obtained based on the first and second summation results. In this way, the image of interest to the target object can be determined as either the first image or the second image based on these occipital lobe EEG features, thereby determining whether the target lower limb is the first or the second lower limb.

[0082] According to embodiments of this disclosure, the controller is further configured to, when reading from the first cache or the second cache, invoke a thread lock to lock the read first cache or the second cache to stop the data writing process of the first cache or the second cache; and when not reading from at least one region of the first cache or the second cache, unlock the at least one region that has not been read.

[0083] According to embodiments of this disclosure, the aforementioned thread lock is a core synchronization mechanism used in multithreaded programming to control concurrent access to shared resources. By locking the first and second buffers with the thread lock, when the collector reads data from the first or second buffer, the writing process to the first or second buffer being read is stopped.

[0084] According to embodiments of this disclosure, locking the first and second caches with thread locks can reduce storage conflicts caused by writing and reading data in the first or second cache, thereby improving the accuracy of reading data from the first or second cache.

[0085] According to embodiments of this disclosure, the display is deployed on a virtual reality device. The controller is also configured to, in a first phase, control the display to show a baseline acquisition image from a fixed viewpoint; the baseline acquisition image is rendered based on a static environment image and attention-guided markers; and in a second phase, control the display to dynamically show an environment image from a non-fixed viewpoint based on the walking state of the target object.

[0086] According to embodiments of this disclosure, both the baseline image acquired from a fixed perspective and the environmental images from non-fixed perspectives can be constructed using Unity software. The continuously displayed environmental images can serve as non-fixed perspective environmental images to form a dynamic, step-by-step scene animation. Figure 5 A schematic diagram illustrating a baseline acquisition image according to an embodiment of the present disclosure is shown. Figure 5 As shown, the baseline image is a static image of the environment. In the center is an attention guide icon, which can be a cross-shaped image. On either side of the cross-shaped image are images corresponding to the left lower limb (the first image can be shown here later) and the right lower limb (the second image can be shown here later). Next, the second stage can proceed, acquiring EEG signals from the sensorimotor area and the occipital lobe, and determining the target wearable device based on these signals.

[0087] Subsequently, the system state of the lower limb hybrid exoskeleton training system is recorded as the GAIT state. The exoskeleton is activated, and the target subject begins lower limb movement training. Figure 6 The illustration schematically depicts an environmental image from a non-fixed viewpoint according to an embodiment of the present disclosure. For example... Figure 6As shown, when the training mode is suspended mode, the non-fixed-view environment image moves backward along a preset path according to the exoskeleton speed. When the training mode is flat-ground walking mode, the non-fixed-view environment image changes according to the position of the target object.

[0088] According to embodiments of this disclosure, displaying attention-guided markers in baseline images acquired from a fixed perspective can enhance the target subject's attention, enabling them to quickly enter a state of lower limb movement training. Environmental images from a non-fixed perspective can provide the target subject with an intuitive first-person view, making the simulation of lower limb movement training closer to real life, improving the realism of the training, and consequently enhancing the target subject's experience.

[0089] According to embodiments of this disclosure, in the first stage, attention guide marks are located in the fourth display area of ​​the display; when the display shows a baseline acquisition image, the display area of ​​the display other than the fourth display area is covered with a transparent black mask so that the attention guide marks guide the attention of the target object; in the second stage, when the display shows a first image and a second image, the display area of ​​the display other than the first display area and the second display area is covered with a black mask so that the first image and the second image attract the attention of the target object; when the target wearable part is determined, the display shows an environmental image with the black mask removed and a feedback display component.

[0090] According to embodiments of this disclosure, the first and second images described above can be rendered using Unity software. The Unity software includes a scene component, a stimulus component, and a feedback display component. The display refresh rate can be 60 Hz. The scene component includes the target object's walking path and a park trail, with the park trail including coin targets to increase the fun for the target object during training. The stimulus component includes the aforementioned black mask and the image sequence (the first and second images). The aforementioned black mask can darken the baseline acquired image to highlight the visual stimulus to the target object. The feedback display component includes an activation feedback window and EEG signal activation status text (e.g., "EEG activation successful, currently activating the right side," as described above) in a third display area.

[0091] According to embodiments of this disclosure, when the lower limb hybrid exoskeleton training system is in a waiting state, a static image of the starting position of the lower limb movement training scene is displayed on the monitor. At this time, the target object's perspective is locked and does not change with head movement; the black mask and feedback display components are not visible. When the lower limb hybrid exoskeleton training system is in a starting state, it indicates the start of the current training round; the black mask becomes visible and covers the target object's field of vision.

[0092] According to embodiments of this disclosure, the use of a black mask can further enhance the target's attention, enabling them to quickly enter a state of lower limb exercise training. When the target is about to begin lower limb exercise training, they can be prompted to prepare, allowing them to better enter a training state and thus improving the smoothness of subsequent lower limb exercise training, further enhancing the target's experience.

[0093] Furthermore, when the target subject's lower limb movement training time reaches the preset time, the controller sends a stop signal to the exoskeleton, which returns to its initial position, and the system status of the lower limb hybrid exoskeleton training system switches to a waiting state. A certain rest period is taken between two lower limb movement training rounds, and the above process is repeated until all lower limb movement training rounds are completed. At this point, the system status of the lower limb hybrid exoskeleton training system switches to an exit state, meaning the system stops operating. If the target subject encounters an emergency during training, they can press the emergency stop button on the exoskeleton to end the training. This design improves the safety of the target subject during training.

[0094] Figure 7 The diagram illustrates a training pattern for lower limb movement of the target subject. For example... Figure 7 As shown, the lower limb training scenario can induce the target's electroencephalogram (EEG) state, which is then displayed as feedback within the lower limb movement training scenario. During lower limb movement training, the target's EEG state drives lower limb movement, and the lower limb movement state, in turn, influences the target's EEG state. The lower limb movement training scenario and the target's lower limb movement state remain synchronized.

[0095] The following example illustrates the lower limb hybrid exoskeleton training system. Figure 8 A schematic diagram of the lower limb hybrid exoskeleton training system is shown. Figure 8 As shown, the lower limb hybrid exoskeleton training system includes an EEG acquisition module 700 based on a data acquisition device, a stimulation module 800 based on a display, a data analysis module 900 based on a controller, and an exoskeleton module 1000 based on the aforementioned exoskeleton. The stimulation module 800 can display the target object's lower limb training scene. The target object's lower limb training scene includes an initial scene, a training start prompt scene, a stimulation playback scene, a feedback result scene, and a stepping scene.

[0096] Embodiments of this disclosure also provide a virtual reality device, including the aforementioned lower limb hybrid exoskeleton training system. The lower limb hybrid exoskeleton training system can be deployed on the virtual reality device. This lower limb hybrid exoskeleton training system can be any of the lower limb hybrid exoskeleton training systems described above.

[0097] Those skilled in the art will understand that the features described in the various embodiments of this disclosure can be combined and / or combined in various ways, even if such combinations or combinations are not explicitly described in this disclosure. In particular, the features described in the various embodiments of this disclosure can be combined and / or combined in various ways without departing from the spirit and teachings of this disclosure. All such combinations and / or combinations fall within the scope of this disclosure.

[0098] The embodiments of this disclosure have been described above. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of this disclosure. Although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. Various substitutions and modifications can be made by those skilled in the art without departing from the scope of this disclosure, and all such substitutions and modifications should fall within the scope of this disclosure.

Claims

1. A lower limb hybrid exoskeleton training system, characterized in that, include: monitor; An exoskeleton includes a first wearable part and a second wearable part, which are respectively worn on different lower limbs of a target object; A data acquisition device is used to acquire signals from the brain of a target object; the brain of the target object includes sensorimotor areas and the occipital lobe. Controller, used for: In the first phase after receiving the start command, the control display shows the baseline acquisition image and reads the baseline EEG signal of the sensorimotor area in the first phase from the acquisition device. In the second phase following the first phase The display is controlled to display the first image in the first step image set in the first display area of ​​the display, and to switch multiple first images in turn at a first switching frequency, thereby repeatedly displaying the walking action corresponding to the first lower limb at a first walking frequency. At the same time, the display is controlled to display the second image in the second step image set in the second display area of ​​the display, and to switch multiple second images in turn at a second switching frequency, thereby repeatedly displaying the walking action corresponding to the second lower limb at a second walking frequency. and, The sensorimotor brainwave signals are read from the sensorimotor area during the second stage. When the sensorimotor brainwave signals indicate that the target object has the intention to move, features are extracted from the occipital lobe brainwave signals based on the first switching frequency, the second switching frequency, the first walking frequency, and the second walking frequency. Based on the extracted occipital lobe brainwave features, the target wearable part is determined from the first wearable part and the second wearable part. Based on the sensorimotor brainwave signals and the baseline brainwave signals, the target wearable part is driven to move the target lower limb of the target object wearing the target wearable part. The acquisition unit includes a first buffer and a second buffer. The first buffer is used to store the EEG signals of the sensorimotor region collected from the sensorimotor region at different times in the form of different first data points. The second buffer is used to store the EEG signals of the occipital lobe region collected from the occipital lobe region at different times in the form of different second data points. The controller is also used to determine the initial read address for the first buffer based on the starting address, the visual pathway conduction time, and the predetermined sampling rate when the data update amount is greater than the predetermined data amount determined based on the predetermined duration and the predetermined sampling rate.

2. The lower limb hybrid exoskeleton training system according to claim 1, characterized in that, The controller is also used for: When the display shows the baseline acquisition image for a predetermined duration, it is determined to enter the second stage. A stage start tag is sent to the acquisition device, and the first cache address corresponding to the sending time of the stage start tag in the first cache area is used as the starting address. The data update amount of the address corresponding to multiple times after the sending time in the first cache area is monitored. Based on the initial read address, the state of the first data point read from the first buffer is detected, and if the detection result indicates that the target object is in a motion imagination state, multiple target second data points are read from the second buffer based on the initial read address. Based on the multiple target second data points, the occipital lobe EEG features are extracted.

3. The lower limb hybrid exoskeleton training system according to claim 2, characterized in that, The addresses of the first cache and the second cache are related to each other; The controller is also used for: For the k-th time window, starting from the k-th read address and ending at the k-th last address corresponding to the last moment of the k-th time window, the k-th group of first data points is read from the first buffer and a state detection is performed to obtain the k-th detection result; k is an integer greater than 1; wherein, when k=2, the k-th read address is determined based on the initial read address; When the k-th detection result indicates that the target object is in a motion-imagined state, the multiple target second data points are read from the second buffer based on the initial read address and the k-th last address.

4. The lower limb hybrid exoskeleton training system according to claim 3, characterized in that, The controller is also used for: If the k-th detection result indicates that the target object is not in a state of motion, the k-th read address is updated to the (k+1)-th read address based on a predetermined step size, so that data is read from the first buffer starting from the (k+1)-th read address to perform state detection; When k+1=n and the k+1th detection result indicates that the target object is not in a state of motion imagination, the target wearable part is driven according to a predetermined strategy; n is an integer greater than k.

5. The lower limb hybrid exoskeleton training system according to claim 3 or 4, characterized in that, The target wearable device is also used to electrically stimulate the target lower limb; The processor is also configured to, when the k-th detection result characterizes the target object as being in a state of motion imagination: Based on the baseline EEG signal, calculate the baseline power spectrum; Based on the H group target addresses after the kth last address, read the first data points of the H group targets from the first buffer, and calculate the power spectra of the H targets based on the first data points of the H group targets; For the h-th target power spectrum among the H target power spectra, the activation value of the h-th lower limb brain region of the target object is calculated based on the baseline power spectrum and the h-th target power spectrum, and the h-th activation index value is calculated based on the h-th lower limb brain region activation value and the target activation value; h is a positive integer, and H is an integer greater than or equal to h; Based on H activation index values, the activation status information of the display in the third display area is controlled to indicate the lower limb activation status to the target object; If the H activation index values ​​are less than the predetermined index values, the electrical stimulation intensity of the target wearable part is increased according to the stimulation adjustment step size to provide enhanced stimulation to the target object within a predetermined time period, and the electrical stimulation intensity is restored after the predetermined time period.

6. The lower limb hybrid exoskeleton training system according to any one of claims 2 to 4, characterized in that, The controller is also used for: Calculate the multiplication factor for the first switching frequency and the second switching frequency respectively to obtain the first multiplication frequency and the second multiplication frequency; Based on the first switching frequency and the first walking frequency, the first target frequency is obtained; Based on the second switching frequency and the second walking frequency, the second target frequency is obtained; A template signal is generated based on the first harmonic frequency, the second harmonic frequency, the first target frequency, and the second target frequency; Based on the correlation between the data points of the template signal and the plurality of target second data points, feature extraction is performed on the plurality of target second data points to obtain the occipital lobe EEG features.

7. The lower limb hybrid exoskeleton training system according to any one of claims 2 to 4, characterized in that, The controller is also used for: When reading from the first cache or the second cache, a thread lock is invoked to lock the first cache or the second cache being read, thereby stopping the data writing process of the first cache or the second cache. If at least one region of the first cache or the second cache has not been read, the unread region is unlocked.

8. The lower limb hybrid exoskeleton training system according to any one of claims 1 to 4, characterized in that, The display is deployed on a virtual reality device; The controller is also used for: During the first phase, the display is controlled to show the baseline acquisition image from a fixed viewing angle; the baseline acquisition image is rendered based on a static environment image and attention guidance markers; During the second stage, based on the walking state of the target wearable device, the display is controlled to dynamically display environmental images from a non-fixed viewing angle.

9. The lower limb hybrid exoskeleton training system according to claim 8, characterized in that, During the first phase, the attention guidance mark is located in the fourth display area of ​​the display; when the display shows the baseline acquisition image, the display area of ​​the display other than the fourth display area is covered with a transparent black mask so that the attention guidance mark guides the attention of the target object. During the second stage, when the display shows the first image and the second image, the display area of ​​the display other than the first display area and the second display area is covered with the black mask so that the first image and the second image attract the attention of the target object; when the target wearing part is determined, the display shows the environmental image with the black mask removed.

10. A virtual reality device, characterized in that, Includes the lower limb hybrid exoskeleton training system as described in any one of claims 1 to 9.

Images