Man-machine cooperative control method and system

Abstract

The invention provides a man-machine cooperative control method and system, and the method comprises the steps: collecting a plurality of electromyographic signals of limbs, and carrying out the filtering of the electromyographic signals, so as to remove the power frequency interference, and obtaining a denoised signal; adopting a sliding window method to extract time domain features and frequency domain features of the de-noised signal to form a multi-feature vector; inputting the multi-feature vector into an integrated learning model, and outputting a joint angle predicted value; performing real-time calculation by utilizing an estimation formula based on the joint angle predicted value to obtain a man-machine interaction torque; calculating a mode switching factor based on the man-machine interaction torque in the task period; acquiring an expected movement track and an actual movement track of the rehabilitation equipment, and calculating a track error; and on the basis of comparison results of the track error and the mode switching factor with a preset threshold value, the training mode of the rehabilitation equipment is switched, so that the misjudgment risk of a single signal source decision is remarkably reduced.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of human-computer interaction, and particularly relates to a human-computer collaborative control method and system. BACKGROUND

[0002] It is a mainstream direction to assist training by using intelligent devices such as exoskeleton robots. In this process, good human-computer collaboration is crucial - the device needs to accurately perceive the patient's movement intention and state, and provide natural and harmonious assistive force accordingly, forming close cooperation. This is the key to ensuring training safety and achieving good rehabilitation results.

[0003] However, the existing mainstream technology generally relies on a single signal source (such as only using electromyographic signals, joint angles or plantar pressure signals) for device control. Although this method simplifies the system, it has significant drawbacks: human movement intention expression is multi-dimensional and easily disturbed. Relying on a single signal is difficult to fully reflect complex intentions (for example, joint angles cannot distinguish between active force and passive swinging), and the signal is easily disabled by noise, attenuation or sensor failure. Once the key signal is disabled, the control system may respond incorrectly, causing human-computer collaboration to fail or even inducing safety risks, which seriously hinders the rehabilitation process. SUMMARY

[0004] To solve the above technical problems, the present application provides a human-computer collaborative control method and system to solve the technical problems in the prior art.

[0005] On the one hand, the present application provides the following technical solution, a human-computer collaborative control method, the method comprising:

[0006] Collecting a plurality of electromyographic signals of a limb, and filtering the electromyographic signals to remove power frequency interference to obtain denoised signals;

[0007] Extracting time domain features and frequency domain features of the denoised signals using a sliding window method to form a plurality of feature vectors; inputting the plurality of feature vectors into an ensemble learning model to output a joint angle prediction value;

[0008] Using an estimation formula to calculate a human-computer interaction torque in real time based on the joint angle prediction value;

[0009] Calculating a mode switching factor based on the human-computer interaction torque within a task cycle;

[0010] Obtaining a desired motion trajectory and an actual motion trajectory of a rehabilitation device, and calculating a trajectory error;

[0011] Switch the training mode of the rehabilitation device based on a comparison result of the trajectory error and the mode switching factor with a preset threshold value, wherein the training mode includes a robot dominant mode, a patient dominant mode, and a safety stop mode.

[0012] Compared with the prior art, the beneficial effects of the present application are: the present application constructs an closed-loop control chain of predicting joint angle based on electromyographic signal → calculating human-computer interaction torque → generating mode switching factor → fusing trajectory error decision, which first realizes an intelligent rehabilitation mode switching mechanism based on dual-mode criteria (patient force capacity + motion accuracy). Compared with single signal control method, this scheme significantly improves the safety of rehabilitation training, effectively prevents the risk of patient motion out of control by monitoring trajectory error in real time and triggering safety stop mode in priority; at the same time, the individualized training adaptation ability is enhanced, and the robot assistance strength is dynamically adjusted based on the quantitative evaluation of the patient's active force capacity in the task cycle, which accurately matches the needs of different rehabilitation stages. Under the premise of ensuring safety, the human-computer cooperation efficiency is optimized by maximizing the patient's active participation (patient dominant mode), which speeds up the process of neural function reconstruction. The innovation of the dual-mode criterion lies in the first time to fuse biomechanical properties (patient active force capacity) and kinematic properties (motion control accuracy) for collaborative decision-making. This design not only realizes active braking protection through motion accuracy monitoring, but also eliminates false triggering interference through force capacity evaluation, which significantly reduces the misjudgment risk of single signal source decision.

[0013] Further, the step of collecting a plurality of electromyographic signals of the limb and filtering the electromyographic signals to remove power frequency interference to obtain a denoising signal comprises:

[0014] The electromyographic signals of the rectus femoris muscle, the medial vastus muscle, the tibialis anterior muscle, the biceps femoris muscle, the medial gastrocnemius muscle and the soleus muscle of the lower limb are collected by a six-channel surface electromyographic sensor, and the sampling frequency is 1000 Hz;

[0015] The electromyographic signals are sequentially subjected to 30-250 Hz Butterworth band-pass filtering and 50 Hz notch filtering to obtain the denoising signal.

[0016] Further, the ensemble learning model is an Adaboost-LSTM-Attention model, comprising:

[0017] At least two layers of LSTM hidden layers, each of the LSTM hidden layers contains no less than 50 units;

[0018] An attention mechanism layer for assigning weights to the input feature vectors;

[0019] An Adaboost integration module for fusing the prediction results of multiple LSTM-Attention sub-models;

[0020] The super parameter optimization module adopts an improved goose optimization algorithm to minimize the root mean square error of the predicted joint angle and the actual joint angle.

[0021] Further, the step of calculating the mode switching factor based on the human-robot interaction torque in the task cycle comprises:

[0022] Based on the human-robot interaction torque, the average value of the projection component in the expected motion direction in the task cycle is calculated to obtain the mode switching factor, and the calculation formula is as follows:

[0023]

[0024] Wherein, denotes the mode switching factor in the jth task cycle, T task denotes the task cycle length, t j , t j-1 denote the starting / ending time of the jth cycle, respectively, denotes the projection component of the human-robot interaction torque in the expected direction, denotes the velocity direction vector of the expected motion trajectory, denotes the projection component of the human-robot interaction torque in the expected direction.

[0025] Further, the step of comparing the trajectory error with the mode switching factor with the preset threshold value comprises:

[0026] If the trajectory error is greater than the preset safety threshold value, enter the safety stop mode;

[0027] If the trajectory error is less than and the mode switching factor is less than the preset ability threshold value, enter the robot dominant mode;

[0028] If the trajectory error is less than and the mode switching factor is greater than the preset ability threshold value, enter the patient dominant mode.

[0029] Further, the estimation formula comprises:

[0030]

[0031] Wherein, τ h (t) denotes the human-robot interaction torque, M(θ(t)) denotes the inertia matrix, (θ(t)) denotes the predicted joint angle, denotes the joint angle acceleration, denotes the Coriolis force matrix, denotes the joint angular velocity, G(θ(t)) denotes the gravity term, K t denotes the motor torque constant, i mdenotes the real-time current of the motor, and denotes the Coulomb friction coefficient, denotes a sign function.

[0032] Further, a sliding window method is used to extract time domain features and frequency domain features of the denoised signal, and specifically includes:

[0033] Using a sliding window with a window length of 200 ms and a step length of 100 ms, the following features of each myoelectric signal channel are extracted:

[0034] Time domain features: average absolute value, root mean square value, zero-crossing rate, and slope sign change frequency;

[0035] Frequency domain features: median frequency and average power frequency;

[0036] All the extracted channel features are combined to form a 36-dimensional feature vector.

[0037] In a second aspect, the present application provides the following technical solution, a human-machine collaborative control system, the system comprises:

[0038] A denoising module is configured to collect a plurality of myoelectric signals of a limb, and filter and process the myoelectric signals to remove power frequency interference, thereby obtaining a denoised signal;

[0039] A prediction module is configured to extract time domain features and frequency domain features of the denoised signal using a sliding window method, thereby forming a multi-feature vector; and input the multi-feature vector into an ensemble learning model, and output a joint angle prediction value;

[0040] A calculation module is configured to calculate a human-machine interaction torque in real time based on the joint angle prediction value using an estimation formula;

[0041] A switching module is configured to calculate a mode switching factor based on the human-machine interaction torque within a task cycle;

[0042] An error module is configured to obtain a desired motion trajectory and an actual motion trajectory of a rehabilitation device, and calculate a trajectory error;

[0043] A comparison module is configured to switch a training mode of the rehabilitation device based on a comparison result of the trajectory error and the mode switching factor with a preset threshold, wherein the training mode includes a robot dominant mode, a patient dominant mode, and a safety stop mode.

[0044] In a third aspect, the present application provides the following technical solution, a computer comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor implements the human-machine collaborative control method as described above when executing the computer program.

[0045] In a fourth aspect, the present application provides a storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the man-machine cooperative control method described above. BRIEF DESCRIPTION OF DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiments or prior art description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0047] Figure 1 A flow chart of the man-machine cooperative control method provided for the first embodiment of the present application;

[0048] Figure 2 A structural block diagram of the man-machine cooperative control system provided for the second embodiment of the present application;

[0049] Figure 3 A hardware structure schematic diagram of the computer provided for the third embodiment of the present application.

[0050] The embodiments of the present application will be further described below with reference to the drawings. DETAILED DESCRIPTION

[0051] The embodiments of the present application will be described in detail below, and examples of the embodiments are shown in the drawings, wherein the same or similar reference numbers data are the same or similar elements or elements with the same or similar functions. The embodiments described below by referring to the drawings are exemplary and are intended to explain the embodiments of the present application, and cannot be understood as a limitation of the present application.

[0052] In the description of the embodiments of the present application, it should be understood that the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the embodiments of the present application and simplifying the description, and therefore cannot be understood as indicating or implying that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation of the present application.

[0053] In addition, the terms "first", "second", "third", etc. are used only for descriptive purposes and are not to be construed as indicating or implying relative importance or an ordered ranking of the indicated technical features. Thus, features defined with "first", "second" or "third" can explicitly or implicitly include one or more of such features. In the description of embodiments of the present application, the meaning of "a plurality" is two or more, unless explicitly specified otherwise.

[0054] Embodiment one

[0055] In the first embodiment of the present application, referring to Figure 1 A human-machine cooperative control method, comprising the following steps S01 to S06:

[0056] S01, collecting a plurality of myoelectric signals of a limb, and filtering the myoelectric signals to remove power frequency interference to obtain a denoising signal;

[0057] Specifically, the step of collecting a plurality of myoelectric signals of a limb and filtering the myoelectric signals to remove power frequency interference to obtain a denoising signal comprises:

[0058] Collecting the myoelectric signals of rectus femoris, vastus medialis, tibialis anterior, biceps femoris, medial gastrocnemius and soleus of lower limbs by six-channel surface myoelectric sensor, and the sampling frequency is 1000Hz;

[0059] Sequentially performing 30-250Hz Butterworth band-pass filtering and 50Hz notch filtering on the myoelectric signals to obtain the denoising signal.

[0060] In this embodiment, the myoelectric signals of key muscles (such as rectus femoris, vastus medialis, tibialis anterior, biceps femoris, medial gastrocnemius and soleus) of a target limb (such as lower limbs) are collected by a plurality of channels (preferably six channels) of surface myoelectric sensor, and the sampling frequency is preferably 1000Hz. The collected raw myoelectric signals contain noise such as power frequency interference, which need to be filtered. Specifically, 30-250Hz Butterworth band-pass filtering (to remove low-frequency motion artifacts and high-frequency noise) and 50Hz notch filtering (to specifically remove power frequency interference) are sequentially performed to obtain high-quality denoising signals.

[0061] Use Delsys Trigno wireless myoelectric sensor (six channels), paste on the lower limb muscles according to the positions shown in Figure 1

[0062] Channel 1: Rectus femoris (midpoint of the line connecting the anterior superior iliac spine to the superior edge of the patella);

[0063] Channel 2: Vastus medialis (5cm above the superior medial edge of the patella);

[0064] ​Channel 3: Tibialis anterior (1 / 3 of the line connecting the fibular head to the medial malleolus);

[0065] Channel 4: Biceps femoris (midpoint of the line connecting the ischial tuberosity to the fibular head);

[0066] Channel 5: Medial gastrocnemius (10 cm below the popliteal fossa, medial to the Achilles tendon);

[0067] Channel 6: Soleus (2 cm below the gastrocnemius muscle belly);

[0068] The original signal (electromyographic signal) is filtered by a 30-250Hz 4th order Butterworth bandpass filter (to remove motion artifacts) → 50Hz notch filter (to suppress power frequency interference) → output denoising signal (signal-to-noise ratio ≥ 45dB).

[0069] It can be understood that the combination of 30-250Hz bandpass filtering and 50Hz notch filtering effectively suppresses power frequency noise and low-frequency motion artifacts (such as limb tremor) in electromyographic signals, and the signal-to-noise ratio is improved; the six channels cover the key flexor and extensor muscle groups of the lower limbs (rectus femoris / medial vastus / tibialis anterior / biceps femoris / gastrocnemius / soleus), fully capturing the multi-joint coordination intention, and reducing the angle prediction error.

[0070] S02, using a sliding window method to extract the time domain features and frequency domain features of the denoising signal, forming a multi-feature vector; inputting the multi-feature vector into an ensemble learning model to output a joint angle prediction value;

[0071] Specifically, the ensemble learning model is an Adaboost-LSTM-Attention model, which includes:

[0072] At least two layers of LSTM hidden layers, each of which contains no less than 50 units;

[0073] An attention mechanism layer for assigning weights to the input feature vectors;

[0074] An Adaboost ensemble module for fusing the prediction results of multiple LSTM-Attention sub-models;

[0075] A hyperparameter optimization module that uses an improved goose optimization algorithm to minimize the root mean square error of the predicted joint angle and the actual joint angle.

[0076] More specifically, the time domain features and frequency domain features of the denoising signal are extracted using a sliding window method, which specifically includes:

[0077] Using a sliding window with a window length of 200ms and a step length of 100ms, the following features of each electromyographic signal channel are extracted:

[0078] Time domain features: mean absolute value, root mean square, zero crossing rate, and slope sign change number;

[0079] Frequency domain features: median frequency, and mean power frequency;

[0080] All the extracted channel features are combined to form a 36-dimensional feature vector.

[0081] In this embodiment, the sliding window method (preferably window length 200 ms, step length 100 ms) is used to process the denoised signal. In each sliding window, multiple time domain features (such as mean absolute value MAV, root mean square RMS, zero crossing rate ZC, and slope sign change number SSC) and frequency domain features (such as median frequency MF and mean power frequency MPF) of each electromyographic signal channel are calculated to form a multi-dimensional feature vector (e.g., 6 channels * 6 features = 36 dimensions) representing the current muscle activity state. The feature vector is input into a pre-trained ensemble learning model (preferably an Adaboost-LSTM-Attention model), which outputs a predicted value for the angle of the target joint (such as the knee joint or the ankle joint). The core innovation of this model is:

[0082] Basic structure: composed of a long short-term memory network (LSTM), containing at least two hidden layers, each with no less than 50 units, for capturing the time sequence dependence in electromyographic signal features.

[0083] Double-layer attention mechanism (Attention): the first layer acts on the input feature dimension (different muscle channels), automatically learning and focusing on the muscle features that are more critical to the current prediction task. The second layer acts on the time step dimension, automatically learning and focusing on the historical time information that is most important for prediction within the current sliding window. Adaboost ensemble and dynamic weights: multiple LSTM-Attention sub-models are integrated. The key innovation lies in the dynamic updating mechanism of the sub-model weights: the maximum weight α m is dynamically calculated and determined according to its prediction performance (root mean square error RMSE□) on an independent test set, with the formula:

[0084]

[0085] where α m is the integrated weight of the mth sub-model, and RMSE m is the prediction root mean square error of the mth LSTM-Attention sub-model on the independent test set. This mechanism gives higher weights to sub-models with smaller errors, improving overall accuracy and robustness.

[0086] Hyperparameter optimization: use improved goose optimization algorithm (IPOA) to automatically optimize the key hyperparameters of the model (such as the number of LSTM layers, units, learning rate, attention dimension, etc.), and the optimization goal is to minimize the root mean square error (RMSE) between the predicted joint angle and the actual joint angle.

[0087] S03, using an estimation formula to calculate the human-robot interaction torque in real time based on the predicted joint angle; the estimation formula includes:

[0088]

[0089] Where τ h (t) represents the human-robot interaction torque, M(θ(t)) represents the inertia matrix, (θ(t)) represents the predicted joint angle, represents the joint angular acceleration, represents the Coriolis force matrix, represents the joint angular velocity, G(θ(t)) represents the gravity term, K t represents the motor torque constant, i m represents the motor real-time current, μ represents the Coulomb friction coefficient, represents the sign function.

[0090] S04, based on the human-robot interaction torque in the task cycle to calculate the mode switching factor;

[0091] The step of calculating the mode switching factor based on the human-robot interaction torque in the task cycle includes:

[0092] Based on the human-robot interaction torque, calculate its average value in the task cycle in the projection component in the expected motion direction to obtain the mode switching factor, and the calculation formula is as follows:

[0093]

[0094] represents the mode switching factor in the jth task cycle, T task represents the task cycle length, t j , t j-1 respectively represent the start / end time of the jth cycle, represents the projection component of the human-robot interaction torque in the expected direction, represents the velocity direction vector of the expected motion trajectory, represents the projection component of the human-robot interaction torque in the expected direction.

[0095] It is worth noting that the torque vector is converted into a direction effectiveness index, the orthogonal direction invalid force interference is excluded, and the rehabilitation evaluation accuracy is improved.

[0096] S05, obtaining the expected motion trajectory and the actual motion trajectory of the rehabilitation equipment, and calculating the trajectory error.

[0097] S06, based on the comparison results of the trajectory error and the mode switching factor with the preset threshold, switching the training mode of the rehabilitation equipment, wherein the training mode includes a robot dominant mode, a patient dominant mode and a safety stop mode.

[0098] Specifically, the step of comparing the trajectory error and the mode switching factor with the preset threshold includes:

[0099] If the trajectory error is greater than the preset safety threshold, the safety stop mode is entered;

[0100] If the trajectory error is less than and the mode switching factor is less than the preset ability threshold, the robot dominant mode is entered;

[0101] If the trajectory error is less than and the mode switching factor is greater than the preset ability threshold, the patient dominant mode is entered.

[0102] In summary, a human-machine collaborative control method has the following effects:

[0103] The present application realizes the intelligent rehabilitation mode switching mechanism based on the dual-mode criterion (patient force ability + motion accuracy) for the first time by constructing the closed-loop control chain of the electromyographic signal predicting joint angle → calculating human-computer interaction torque → generating mode switching factor → fusion trajectory error decision. Compared with the single signal control method, the safety of the rehabilitation training is significantly improved, the risk of patient motion out of control is effectively prevented by monitoring the trajectory error in real time and triggering the safety stop mode preferentially, and the individualized training adaptation ability is enhanced. Based on the quantitative evaluation of the patient's active force ability in the task cycle, the robot assistance strength is dynamically adjusted to accurately match the needs of different rehabilitation stages. Under the premise of ensuring safety, the patient's active participation (patient dominant mode) is maximized, the human-machine cooperation efficiency is optimized, and the neural function reconstruction process is accelerated.

[0104] The innovation of the dual-mode criterion lies in the first time of fusion of biomechanical characteristics (patient active force ability) and kinematic characteristics (motion control accuracy) for collaborative decision-making. The design not only realizes active braking protection through motion accuracy monitoring, but also excludes false triggering interference through force ability evaluation, and the two complement each other to significantly reduce the misjudgment risk of single signal source decision.

[0105] Furthermore, by directly calculating the human-machine interaction torque using predicted joint angles, the problem of synchronization delay of multi-sensor signals was successfully solved, improving the control response speed. This strategy simultaneously eliminates the dependence on physical angle sensors, reducing system complexity and hardware costs. It also avoids the differential noise amplification effect of traditional sensor signals through the noise suppression characteristics of model predictions, improving torque estimation accuracy and forming a more stable and reliable human-machine interaction force closed loop.

[0106] Example 2

[0107] like Figure 2 As shown, a second embodiment of the present invention provides a human-machine collaborative control system, the system comprising:

[0108] The noise reduction module 10 is used to collect multiple electromyographic signals of the limb and filter the electromyographic signals to remove power frequency interference and obtain a noise-reduced signal.

[0109] Prediction module 20 is used to extract the time-domain and frequency-domain features of the denoised signal using the sliding window method to form a multi-feature vector; input the multi-feature vector into the ensemble learning model and output the joint angle prediction value;

[0110] The calculation module 30 is used to calculate the human-machine interaction torque in real time based on the predicted value of the joint angle using the estimation formula;

[0111] Switching module 40 is used to calculate the mode switching factor based on the human-computer interaction torque within the task cycle;

[0112] Error module 50 is used to acquire the expected and actual motion trajectories of the rehabilitation equipment and calculate the trajectory error;

[0113] The comparison module 60 is used to switch the training mode of the rehabilitation device based on the comparison results of the trajectory error and the mode switching factor with preset thresholds, wherein the training mode includes robot-led mode, patient-led mode and safety stop mode.

[0114] The human-machine collaborative control system provided in this embodiment of the invention has the same implementation principle and technical effects as the aforementioned method embodiment. For the sake of brevity, any parts not mentioned in the system embodiment can be referred to the corresponding content in the aforementioned method embodiment.

[0115] Example 3

[0116] like Figure 3As shown in the third embodiment of the present application, the embodiment of the present application provides the following technical scheme, a computer, comprising a memory 202, a processor 201 and a computer program stored in the memory 202 and capable of running on the processor 201, wherein the processor 201 implements the man-machine collaborative control method as described above when executing the computer program.

[0117] Specifically, the processor 201 can include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0118] The memory 202 can include a mass storage for data or instructions. By way of example, and without limitation, the memory 202 can include a hard disk drive (HDD), a floppy disk drive, a solid-state drive (SSD), a flash drive, a compact disc (CD) or DVD, a tape, a magnetic or optical or magneto-optical storage, or a Universal Serial Bus (USB) drive or a combination of two or more of these. The memory 202 can be removable and / or built-in (or fixed) as appropriate. The memory 202 can be internal or external as appropriate. In certain embodiments, the memory 202 is a nonvolatile memory. In certain embodiments, the memory 202 includes a Read-Only Memory (ROM) and a Random Access Memory (RAM). The ROM can be a mask-programmed ROM, a Programmable ROM (PROM), an Erasable Programmable ROM (EPROM), an Electrically Erasable Programmable ROM (EEPROM), an Electrically Alterable ROM (EAROM), or a FLASH memory, or a combination of two or more of these, as appropriate. The RAM can be a Static Random-Access Memory (SRAM) or a Dynamic Random-Access Memory (DRAM), which can be a Fast Page Mode Dynamic Random-Access Memory (FPMDRAM), an Extended Data Output Dynamic Random-Access Memory (EDODRAM), a Synchronous Dynamic Random-Access Memory (SDRAM), or the like, as appropriate.

[0119] The memory 202 can be used to store or buffer various data files needed for processing and / or communication, and possible computer program instructions executed by the processor 201.

[0120] The processor 201 realizes the above-mentioned human-computer collaborative control method by reading and executing the computer program instructions stored in the memory 202.

[0121] In some embodiments, the computer can further include a communication interface 203 and a bus 200. As shown, the processor 201, the memory 202, and the communication interface 203 are connected through the bus 200 and complete communication with each other. Figure 3

[0122] The communication interface 203 is used to realize the communication between various modules, devices, units and / or equipment in the embodiments of the present application. The communication interface 203 can also realize data communication with other components, such as external devices, image / data acquisition devices, databases, external storage, and image / data processing workstations, etc.

[0123] ​Bus 200 includes hardware, software, or both, to couple components of computer 100 to each other and to couple them both to other peripheral devices. Bus 200 includes, but is not limited to, at least one of the following: a data bus, an address bus, a control bus, an expansion bus, a local bus, and the like. By way of example, and not limitation, bus 200 can include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand (IB) interconnect, a Low Pin Count (LPC) bus, a memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or the like. Bus 200 can include one or more buses, where appropriate. Although this application describes and illustrates a particular bus, this application contemplates any suitable bus or interconnect.

[0124] Embodiment Four

[0125] In the fourth embodiment of the present application, in combination with the man-machine collaborative control method described above, the embodiment of the present application provides the following technical solution, a storage medium, the storage medium has a computer program stored thereon, the computer program is executed by a processor to realize the man-machine collaborative control method described above.

[0126] Those skilled in the art will appreciate that the logic and / or steps described in the flowcharts and / or otherwise described herein, such as in the description of the above embodiments, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor- containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions, or a combination thereof. For the purposes of this specification, a "computer-readable medium" can be any apparatus that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. The computer-readable medium can be, for example but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium include the following: an electrical connection (electronic) having one or more wires, a portable computer diskette (magnetic), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber (optical), and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium can even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, for example via an optical scanner, then compiled, interpreted, or otherwise processed, and stored in a computer memory in a form that can be later executed by a computer. In the above description of various embodiments of the present application, it will be recognized that the steps and / or functions of the described implementation can be rearranged or reordered, or other steps can be added or omitted, without departing from the scope and spirit of the present application. Accordingly, the above description should not be construed as limiting the present application.

[0127] More specific examples (a non-exhaustive list) of the computer-readable medium include the following: an electrical connection (electronic) having one or more wires, a portable computer diskette (magnetic), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber (optical), and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium can even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, for example via an optical scanner, then compiled, interpreted, or otherwise processed, and stored in a computer memory in a form that can be later executed by a computer.

[0128] It should be understood that aspects of the application can be implemented in hardware, software, firmware, or combinations thereof. In the above embodiments, various steps or methods can be implemented in software or firmware that is stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any of the following technologies known in the art, or combinations thereof, can be used: a discrete logic circuit having logic gates for implementing logic functions on data signals, an application specific integrated circuit having appropriate combinational logic gates, a programmable gate array (PGA), a field programmable gate array (FPGA), and / or the like.

[0129] The technical features of the above-described embodiments can be combined in any manner, and for brevity, not all possible combinations are described, but it is understood that any combination of the technical features is within the scope of the present specification.

[0130] The above-described embodiments are merely illustrative of several embodiments of the present application, which are described in more detail and in a specific and detailed manner, but should not be construed as limiting the scope of the patent. It should be noted that for those skilled in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. A human-machine collaborative control method, characterized in that, The method includes: Multiple electromyographic (EMG) signals from the limbs are collected, and the EMG signals are filtered to remove power frequency interference, resulting in a denoised signal. The time-domain and frequency-domain features of the denoised signal are extracted using the sliding window method to form a multi-feature vector; the multi-feature vector is then input into an ensemble learning model to output the predicted joint angle value. The human-computer interaction torque is calculated in real time based on the predicted joint angle using the estimation formula; The mode switching factor is calculated based on the human-computer interaction torque during the task cycle; Obtain the expected and actual motion trajectories of the rehabilitation equipment, and calculate the trajectory error; Based on the comparison results of the trajectory error and the mode switching factor with preset thresholds, the training mode of the rehabilitation equipment is switched, wherein the training mode includes robot-led mode, patient-led mode and safe stop mode.

2. The human-machine collaborative control method according to claim 1, characterized in that, The steps of acquiring multiple electromyographic (EMG) signals from the limb and filtering the EMG signals to remove power frequency interference and obtain a denoised signal include: Electromyographic signals of the rectus femoris, vastus medialis, tibialis anterior, biceps femoris, gastrocnemius medialis, and soleus muscles of the lower limb were collected using a six-channel surface electromyography sensor at a sampling frequency of 1000 Hz. The electromyographic signal was sequentially subjected to Butterworth bandpass filtering (30-250Hz) and notch filtering (50Hz) to obtain the denoised signal.

3. The human-machine collaborative control method according to claim 1, characterized in that, The ensemble learning model is the Adaboost-LSTM-Attention model, which includes: At least two LSTM hidden layers, each of which contains no less than 50 units; The attention mechanism layer is used to assign weights to the input feature vector; The Adaboost ensemble module is used to fuse the prediction results of multiple LSTM-Attention sub-models. The hyperparameter optimization module uses an improved goose optimization algorithm to minimize the root mean square error between the predicted joint angle and the actual joint angle.

4. The human-machine collaborative control method according to claim 1, characterized in that, The step of calculating the mode switching factor based on the human-computer interaction torque within the task cycle includes: Based on the human-computer interaction torque, the projection component of its average value in the desired motion direction during the task cycle is calculated to obtain the mode switching factor, where the calculation formula is as follows: in, Let T be the mode switching factor within the j-th task cycle. task The duration is represented by t. j , t j-1 Let these represent the start and end times of the j-th cycle, respectively. This represents the projected component of the human-computer interaction torque in the desired direction. It is represented as the velocity direction vector of the desired trajectory. It is represented as the projected component of the human-computer interaction torque in the desired direction.

5. The human-machine collaborative control method according to claim 1, characterized in that, The steps based on the comparison result between the trajectory error and the mode switching factor and the preset threshold include: If the trajectory error exceeds a preset safety threshold, then enter the safety stop mode; If the trajectory error is less than and the mode switching factor is less than a preset capability threshold, then the robot-dominated mode is entered. If the trajectory error is less than and the mode switching factor is greater than the preset capability threshold, then the patient-led mode is entered.

6. The human-machine collaborative control method according to claim 1, characterized in that, The estimation formula includes: Where, τ h (t) represents the human-machine interaction torque, M(θ(t)) represents the inertia matrix, and (θ(t)) represents the predicted joint angle. Expressed as joint angular acceleration, Represented as the Coriolis force matrix, Let G(θ(t)) represent the joint angular velocity, and K represent the gravitational term. t Expressed as the motor torque constant, i m The value is expressed as the real-time current of the motor, and μ represents the Coulomb friction coefficient. It is represented as a symbolic function.

7. The human-machine collaborative control method according to claim 1, characterized in that, The time-domain and frequency-domain features of the denoised signal are extracted using the sliding window method, specifically including: Using a sliding window with a window length of 200ms and a step size of 100ms, the following features were extracted for each electromyographic signal channel: Time-domain characteristics: mean absolute value, root mean square value, zero-crossing rate, and number of slope sign changes; Frequency domain characteristics: median frequency and average power frequency; The features extracted from all channels are combined to form a 36-dimensional feature vector.

8. A human-machine collaborative control system, characterized in that, The system includes: The noise reduction module is used to acquire multiple electromyographic signals from the limbs and filter the electromyographic signals to remove power frequency interference, thereby obtaining a noise-reduced signal. The prediction module is used to extract the time-domain and frequency-domain features of the denoised signal using the sliding window method to form a multi-feature vector; the multi-feature vector is input into the ensemble learning model to output the predicted joint angle value; The calculation module is used to calculate the human-computer interaction torque in real time based on the predicted value of the joint angle using the estimation formula; The switching module is used to calculate the mode switching factor based on the human-computer interaction torque within the task cycle; The error module is used to acquire the expected and actual motion trajectories of the rehabilitation equipment and calculate the trajectory error. The comparison module is used to switch the training mode of the rehabilitation device based on the comparison results of the trajectory error and the mode switching factor with preset thresholds, wherein the training mode includes robot-led mode, patient-led mode and safety stop mode.

9. A computer comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the human-machine collaborative control method as described in any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the human-machine collaborative control method as described in any one of claims 1 to 7.

Images