Man-machine coordination motion control method for predicting active walking speed of convalescent patient

Abstract

The invention belongs to the technical field of wheeled rehabilitation robot control, and particularly relates to a man-machine coordination motion control method for predicting the active walking speed of a rehabilitative person. The intelligence of the rehabilitation robot and the safety of man-machine coordination movement are improved. The method comprises the following steps: S1, acquiring data information related to walking of a convalescent person, and processing the data information by applying a feature extraction algorithm to generate a feature vector; predicting the walking speed of the convalescent patient by using the feature vector in combination with a time sequence prediction model; s2, designing a time-aware Boosting strategy, dynamically adjusting weight information of walking speed prediction, and establishing an active walking speed prediction model for the convalescent patient; s3, constructing a man-machine motion tracking error system by using the predicted value of the walking speed of the convalescent and the rehabilitation robot system dynamics model; and S4, based on the tracking error system, designing a self-adaptive learning tracking control method, so that the rehabilitation robot can follow the active walking speed of the rehabilitation person and stably track a predetermined training track.

Description

Technical Field

[0001] The invention belongs to the technical field of wheeled rehabilitation robot control, and in particular relates to a human-machine coordinated motion control method for predicting active walking speed of a rehabilitator. Background Art

[0002] Due to the occurrence of traffic accidents and the accelerated aging of the population, the number of patients with walking disorders has increased year by year. However, there is a shortage of professional rehabilitation personnel, which directly leads to the difficulty for patients with walking disorders to obtain timely and effective exercise training. Over time, the walking function of patients gradually deteriorates or even disappears, which greatly affects their ability to achieve daily independent living.

[0003] With the widespread use of rehabilitation walking robots in rehabilitation centers, nursing homes and other places, the problem of shortage of rehabilitation personnel has been effectively alleviated. However, in the actual application process, when the trainees' walking ability and balance ability gradually improve, they often have the desire to participate in active training. However, current research cannot achieve the accurate prediction of the trainees' active walking speed by rehabilitation robots in the active training mode, which makes it difficult to coordinate the movement speed of humans and machines, and thus poses a threat to the safety of trainees.

[0004] In recent years, although some progress has been made in the research on tracking control methods of rehabilitation robots, most of the research focuses on the movement of rehabilitation robots at a speed pre-specified by the doctor. However, when the trainee has the need for active walking speed training, the robot's prediction of the trainee's active walking speed is not fully considered, nor is the tracking control problem of human-machine speed coordinated movement in the active training mode effectively solved.

[0005] In view of this, studying the prediction method of the active walking speed of the rehabilitated person and realizing human-machine coordinated motion are of vital importance to improving the intelligence and safety of rehabilitation robots. Based on a new perspective, the present invention innovatively proposes a prediction method of the active walking speed of the trainee and an adaptive learning tracking control method for human-machine coordinated motion, providing new technical support for human-machine coordinated motion in the active training mode. Summary of the invention

[0006] The present invention aims at the defects of the prior art and provides a human-machine coordinated motion control method for predicting the active walking speed of a rehabilitated person, which improves the intelligence of the rehabilitation robot and the safety of the human-machine coordinated motion.

[0007] To achieve the above-mentioned purpose, the present invention adopts the following technical scheme: a human-machine coordinated motion control method for predicting the active walking speed of a rehabilitated person, comprising:

[0008] S1. Acquire data information related to the rehabilitation person's walking, and apply a feature extraction algorithm to process the data information to generate a feature vector; and use the feature vector in combination with a time series prediction model to predict the rehabilitation person's walking speed;

[0009] S2. Design a time-aware Boosting strategy to dynamically adjust the weight information of walking speed prediction and establish a model for predicting the active walking speed of the rehabilitation patients.

[0010] S3, using the predicted walking speed of the rehabilitation patient and the dynamics model of the rehabilitation robot system, a human-machine motion tracking error system is constructed;

[0011] S4. Based on the tracking error system, an adaptive learning tracking control method is designed so that the rehabilitation robot can follow the active walking speed of the rehabilitator and stably track the predetermined training trajectory.

[0012] Further, in S1, the application feature extraction algorithm processes the data information to generate a feature vector based on formula (1):

[0013]

[0014] Among them, V t represents the generated feature vector, φ(·) represents the nonlinear activation function, W γ represents the dilated convolution weight, Y t-γ Represents the input data information characteristics that affect the walking speed of the rehabilitation patient, OS t-γ represents the external interference environment information characteristics of the human-machine system, b represents the bias parameter, ρ represents the adjustment parameter of the input data information characteristics, and the symbol ⊙ represents element multiplication;

[0015] In S1, the use of the feature vector combined with the time series prediction model to predict the walking speed of the rehabilitated person is based on formula (2):

[0016] v t =LSTM(F0,v t-1 , ψ t ☉V t )

[0017] Among them, LSTM(·) represents the long short-term memory network function, v t and v t-1 They represent the walking speed prediction output at the current moment and the previous moment respectively, F0 represents the weight matrix, ψ t Represents a transformation matrix.

[0018] Furthermore, S2 is specifically divided into:

[0019] S2.1. The time-aware boosting strategy is based on adjusting the decay speed parameter to adapt to the difference between the actual walking speed and the predicted walking speed of the rehabilitation patient:

[0020]

[0021] Among them, λ represents the adjustment f(t k ,ò k ) parameter of the decay rate, k represents the difference between the actual walking speed of the recovered person and the k-th predicted walking speed;

[0022] S2.2, according to k The dynamic weight adjustment strategy for walking speed prediction is designed as follows:

[0023]

[0024] Among them, β represents the adjustment parameter of dynamic weight, Q k represents the dynamic weight of the k-th prediction;

[0025] S2.3. Determine the number of base models for time-aware boosting for walking speed prediction:

[0026]

[0027] Where N represents the number of base models, T D represents the predicted time of walking speed, T d represents the walking speed prediction time of each base model, c represents the adjustment parameter of the number of base models;

[0028] S2.4. Construct a prediction model for active walking speed of the rehabilitation patients:

[0029]

[0030] Where v(t) represents the predicted walking speed, σ(·) represents the activation function, and α k Represents the weight coefficient of the kth basis model.

[0031] Furthermore, S3 specifically includes:

[0032] S3.1. The dynamic model of the rehabilitation robot is based on formula (7):

[0033]

[0034] in

[0035]

[0036] Wherein, M represents the mass of the rehabilitation robot, m represents the mass of the trainer, M1 and M2 represent the coefficient matrices, X(t) represents the motion trajectory of the rehabilitation robot in the three directions of x-axis, y-axis and rotation angle, u(t) is the control input force of the four wheels of the rehabilitation robot, r0 represents the distance from the center of gravity to the center of the rehabilitation robot, I0 represents the moment of inertia of the rehabilitation robot, represents the moment of inertia of the rehabilitation patient, θ represents the angle between the horizontal axis and the line connecting the center of the robot and the center of the first wheel, that is, θ=θ1. From the structure of the rehabilitation robot, we can know that θ3=θ+π, l μ represents the distance from the center of gravity of the system to the center of each wheel, Represents the x′ axis and the l corresponding to each wheel μ The angle between them, μ = 1, 2, 3, 4;

[0037] S3.2, when the rehabilitation robot tracks the motion and learns it for the ith time, formula (7) becomes:

[0038]

[0039] Where i∈Z + Indicates the number of learning times;

[0040] S3.3, Separate the uncertain human-machine system motion environment information in the coefficient matrix M1 in formula (8), and record M1 = M α +M β , where M α It is composed of the mass and moment of inertia of the rehabilitation robot, M β Indicates that the human-machine system is uncertain about the motion environment, and

[0041]

[0042] S3.4, therefore, formula (8) is transformed into the following form:

[0043]

[0044] in, And σ(t) is bounded;

[0045] S3.5, let x 1,i (t) = X i (t), The system state equation is obtained as:

[0046]

[0047] S3.6, let the actual motion trajectory of the rehabilitation robot in the i-th learning be x 1,i (t) and the training trajectory specified by the doctor is Xd (t), the actual speed of the recovered person is x 2,i (t) and the predicted active walking speed is v(t), so the learning rate θ(t) is introduced to design the trajectory tracking error e 1,i (t) and velocity tracking error e 2,i (t) are:

[0048]

[0049] set up represents the estimated value of the learning rate θ(t), and d means Deviation compensation for θ(t); θ * Represents the specified learning rate, let Represents the learning rate error, then

[0050]

[0051] S3.7, Combining equation (10) and equation (11), we get the human-machine tracking motion error system as equation (13):

[0052]

[0053] Furthermore, S4 specifically includes:

[0054] For the human-machine tracking motion error system, that is, formula (13), the i-th controller with adaptive learning rate is designed as follows:

[0055]

[0056] in, represents the generalized inverse matrix of B(θ), Indicates e 2,i The generalized inverse matrix of (t), represents the estimated value of σ(t) in the i-th iteration learning, and the estimated error μ represents an adjustable parameter, and μ>|d|;

[0057] The Lyapunov function designed for the i-th learning is as follows

[0058]

[0059] Taking the derivative of equation (17) and substituting equation (14) and equation (16) into equation (17) yields

[0060]

[0061] Define the function L i (t) are as follows:

[0062]

[0063] in

[0064]

[0065] make

[0066]

[0067] in

[0068]

[0069] Let V i (0) = 0, so we can get

[0070]

[0071] According to formula (23), we know that L i (t) is a decreasing function;

[0072] When i = 0, the derivative of equation (19) yields

[0073]

[0074] From formula (18), we can know

[0075]

[0076] So we know that L0(t) is differentiable and its derivative is less than a bounded value, so L0(t) is continuous and bounded on t∈[0,T], where T represents the time of one learning. i (t) bounded;

[0077] From formula (21), we can get

[0078]

[0079] According to equations (23) and (26), we can get

[0080]

[0081] get

[0082]

[0083] According to the convergence of the series, as the number of learning times increases, we can get And the learning rate error The rehabilitation robot realizes trajectory tracking and active walking speed tracking of the rehabilitator, and coordinated movement is realized in the active training mode of the human-machine system.

[0084] Furthermore, based on the STM32F411 series MCU, the output PWM signal is provided to the motor drive module, so that the rehabilitation robot can help the rehabilitator coordinate exercise training at the predicted walking speed. The specific implementation method is as follows:

[0085] S101, use STM32F411 series microcontroller as the main controller;

[0086] S102, the main controller receives a feedback signal from the motor speed measurement module and connects the output to the motor drive module;

[0087] S103, the motor drive module is connected to the DC motor;

[0088] S104, the power supply system supplies power to the main controller, the motor speed measurement module, the motor drive module and the DC motor;

[0089] S105, the main controller reads the feedback signal of the motor encoder and compares it with the predicted walking speed to calculate an error signal;

[0090] S106. Based on the error signal, the main controller calculates the required motor control amount according to a predetermined human-machine coordinated motion control method, and outputs the control amount to the motor drive module so that the motor rotates to drive the wheels to maintain balance and move in a specified manner.

[0091] Furthermore, the STM32F411 series single chip microcomputer outputs a PWM signal to the motor drive module to adjust the speed and direction of the motor.

[0092] Furthermore, the motor speed measurement module includes an encoder for providing real-time feedback information of the motor speed.

[0093] Furthermore, the human-machine coordinated motion control method adjusts the motor control amount based on the predicted walking speed of the rehabilitated person to achieve gait coordination in rehabilitation training.

[0094] Compared with the prior art, the present invention has beneficial effects.

[0095] The present invention establishes a prediction model for the active walking speed of the rehabilitated person; uses the active walking speed of the rehabilitated person and the dynamics model of the rehabilitation robot system to construct a human-machine motion tracking error system, and proposes an adaptive learning tracking control method, providing a new technology for human-machine coordinated motion in the active training mode. BRIEF DESCRIPTION OF THE DRAWINGS

[0096] The present invention is further described below in conjunction with the accompanying drawings and specific implementation methods. The protection scope of the present invention is not limited to the following description.

[0097] Figure 1This is a flow chart of the human-machine coordinated motion control method of the present invention;

[0098] Figure 2 is the coordinate diagram of the system of the present invention;

[0099] Figure 3 It is the minimum system of STM32F411 single-chip microcomputer of the present invention;

[0100] Figure 4 This is the peripheral circuit of the motor drive module of the present invention. DETAILED DESCRIPTION

[0101] In order to make the purpose, technical scheme and beneficial effects of the embodiments of the present invention clearer, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments.

[0102] like Figure 1-4 As shown in the embodiment: the human-machine coordinated motion control method for predicting the active walking speed of the rehabilitated person includes:

[0103] Step 1: Using the data information that affects the walking speed of the rehabilitation patients, the TCN method is used to extract the features of the data information, and then the LSTM method is combined to predict the walking speed of the rehabilitation patients;

[0104] Step 2: Design a time-aware Boosting strategy, dynamically adjust the weight information of walking speed prediction, and establish a model for predicting the active walking speed of the rehabilitated person;

[0105] Step 3: Using the active walking speed of the rehabilitation person and the dynamics model of the rehabilitation robot system, and introducing the learning rate, a human-machine motion tracking error system was constructed;

[0106] Step 4: Based on the tracking error system, an adaptive learning tracking control method is proposed to coordinate the rehabilitation robot with the active walking speed of the rehabilitator and stably track the training trajectory specified by the doctor.

[0107] Among them, in step 1, feature extraction and modeling of walking speed prediction. First, collect data information that affects the walking speed of the rehabilitated person, including but not limited to the rehabilitated person's step frequency, walking time, walking speed, physical condition, etc. Then use the TCN (Temporal Convolutional Network) method to extract features from these data. It can capture long-term dependencies in time series. Through TCN, the present invention obtains a set of feature vectors that can represent the characteristics of the rehabilitated person's walking speed. Then use the LSTM (Long Short-Term Memory) model combined with the above feature vectors to predict the walking speed of the rehabilitated person. LSTM is good at learning dependencies within long time intervals, so it is very suitable for predicting tasks with time continuity such as walking speed. The prediction process is given by formula (2)v t =LSTM(F0,v t-1 ,Ψ t ⊙V t )definition.

[0108] In step 2, the time-aware Boosting strategy is designed:

[0109] In order to improve the prediction accuracy, this scheme designs a time-aware Boosting strategy to dynamically adjust the weight information of walking speed prediction. This strategy takes into account the change of prediction error over time and realizes dynamic weight adjustment through formulas (3) to (6).

[0110]

[0111] Q k+1 =Q k ×(1+βò k ) (4)

[0112]

[0113] In particular, the number of base models is determined based on the time span of walking speed prediction and the time required for each prediction. This helps to build a more accurate active walking speed prediction model for rehabilitated persons, thereby better reflecting the actual situation of rehabilitated persons.

[0114] Step 3: Build a human-machine motion tracking error system:

[0115] In this stage, the results of the patient's active walking speed prediction are combined with the dynamic model of the rehabilitation robot system. According to formulas (7) to (9): The present invention describes the motion characteristics of the rehabilitation robot and separates the uncertain human-machine system motion environment information in the coefficient matrix. In addition, a learning rate is introduced to adjust the speed at which the rehabilitation robot follows the rehabilitator to ensure the synchronization between the two. Through formulas (10) to (13), the present invention establishes a human-machine motion tracking error system that can evaluate the motion difference between the rehabilitation robot and the rehabilitator.

[0116]

[0117] Step 4: Adaptive learning tracking control method:

[0118] Finally, for the previously established human-machine motion tracking error system, the present invention proposes an adaptive learning tracking control method. This method aims to enable the rehabilitation robot to not only follow the active walking speed of the rehabilitated person, but also to stably track the target trajectory set by the doctor for rehabilitation training. To this end, the present invention designs a controller with an adaptive learning rate (Formula 14-16), and proves the stability of the system through the Lyapunov function (Formula 17). As the number of learning times increases, as shown in Formulas (23) to (28), the rehabilitation robot gradually reduces the error between the target trajectory and finally realizes an efficient and stable rehabilitation training process. The specific formula is:

[0119]

[0120]

[0121] In summary, this embodiment provides a complete methodological framework for realizing the prediction of the active walking speed of the rehabilitated person and the coordinated motion control with the rehabilitation robot, providing technical support for personalized rehabilitation training.

[0122] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "preferred embodiments", "specific implementation", or "preferred implementation" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0123] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, it should be understood by those skilled in the art that the technical solutions described in the aforementioned embodiments may still be modified, or some or all of the technical features therein may be replaced by equivalents. Therefore, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope defined by the claims of the present invention.

Claims

1. A human-machine coordinated motion control method for predicting active walking speed of a rehabilitated person, characterized by: include: S1. Acquire data information related to the rehabilitation person's walking, and apply a feature extraction algorithm to process the data information to generate a feature vector; use the feature vector in combination with a time series prediction model to predict the rehabilitation person's walking speed; S2. Design a time-aware Boosting strategy to dynamically adjust the weight information of walking speed prediction and establish a model for predicting the active walking speed of the rehabilitation patients. S3, using the predicted walking speed of the rehabilitation patient and the dynamics model of the rehabilitation robot system, a human-machine motion tracking error system is constructed; S4. Based on the tracking error system, an adaptive learning tracking control method is designed so that the rehabilitation robot can follow the active walking speed of the rehabilitator and stably track the predetermined training trajectory.

2. The human-machine coordinated motion control method for predicting active walking speed of a rehabilitated person according to claim 1, characterized in that: In S1, the application feature extraction algorithm processes the data information to generate a feature vector based on formula (1): Among them, V t represents the generated feature vector, φ(·) represents the nonlinear activation function, W γ represents the dilated convolution weight, Y t-γ Represents the input data information characteristics that affect the walking speed of the rehabilitation patient, OS t-γ represents the external interference environment information characteristics of the human-machine system, b represents the bias parameter, ρ represents the adjustment parameter of the input data information characteristics, and the symbol ⊙ represents element multiplication; In S1, the use of the feature vector combined with the time series prediction model to predict the walking speed of the rehabilitated person is based on formula (2): v t =LSTM(F0,v t-1 ,ψ t ☉V t ) Among them, LSTM(·) represents the long short-term memory network function, v t and v t-1 They represent the walking speed prediction output at the current moment and the previous moment, respectively, F0 represents the weight matrix, ψ t Represents a transformation matrix.

3. The human-machine coordinated motion control method for predicting active walking speed of a rehabilitated person according to claim 1, characterized in that: S2 is specifically divided into: S2.

1. The time-aware boosting strategy is based on adjusting the decay speed parameter to adapt to the difference between the actual walking speed and the predicted walking speed of the rehabilitation patient: Among them, λ represents the adjustment f(t k ,ò k ) parameter of the decay rate, k represents the difference between the actual walking speed of the recovered person and the k-th predicted walking speed; S2.2, according to k The dynamic weight adjustment strategy for walking speed prediction is designed as follows: Q k+1 =Q k ×(1+βò k ) (4) Among them, β represents the adjustment parameter of dynamic weight, Q k represents the dynamic weight of the k-th prediction; S2.

3. Determine the number of base models for time-aware boosting for walking speed prediction: Where N represents the number of base models, T D represents the predicted time of walking speed, T d represents the walking speed prediction time of each base model, c represents the adjustment parameter of the number of base models; S2.

4. Construct a prediction model for active walking speed of rehabilitated patients: Where v(t) represents the predicted walking speed, σ(·) represents the activation function, and α k Represents the weight coefficient of the kth basis model.

4. The human-machine coordinated motion control method for predicting active walking speed of a rehabilitated person according to claim 1, characterized in that: S3 specifically includes: S3.

1. The dynamic model of the rehabilitation robot is based on formula (7): in Wherein, M represents the mass of the rehabilitation robot, m represents the mass of the trainer, M1 and M2 represent the coefficient matrices, X(t) represents the motion trajectory of the rehabilitation robot in the three directions of x-axis, y-axis and rotation angle, u(t) is the control input force of the four wheels of the rehabilitation robot, r0 represents the distance from the center of gravity to the center of the rehabilitation robot, I0 represents the moment of inertia of the rehabilitation robot, represents the moment of inertia of the rehabilitation patient, θ represents the angle between the horizontal axis and the line connecting the center of the robot and the center of the first wheel, that is, θ=θ1. From the structure of the rehabilitation robot, we can know that θ3=θ+π, l μ represents the distance from the center of gravity of the system to the center of each wheel, Represents the x′ axis and the l corresponding to each wheel μ The angle between them, μ = 1, 2, 3, 4; S3.2, when the rehabilitation robot tracks the motion and learns it for the ith time, formula (7) becomes: Where i∈Z + Indicates the number of learning times; S3.3, Separate the uncertain human-machine system motion environment information in the coefficient matrix M1 in formula (8), and record M1 = M α +M β , where M α It is composed of the mass and moment of inertia of the rehabilitation robot, M β Indicates that the human-machine system is uncertain about the motion environment, and S3.4, therefore, formula (8) is transformed into the following form: in, And σ(t) is bounded; S3.5, let x 1,i (t) = X i (t), The system state equation is obtained as: S3.6, let the actual motion trajectory of the rehabilitation robot in the i-th learning be x 1,i (t) and the training trajectory specified by the doctor is X d (t), the actual speed of the recovered person is x 2,i (t) and the predicted active walking speed is v(t), thus introducing the learning rate Design trajectory tracking error e 1,i (t) and velocity tracking error e 2,i (t) are: set up Represents the learning rate The estimated value of d means right Deviation compensation; Represents the specified learning rate, let Represents the learning rate error, then S3.7, Combining equation (10) and equation (11), we get the human-machine tracking motion error system as equation (13):

5. The human-machine coordinated motion control method for predicting active walking speed of a rehabilitated person according to claim 1, characterized in that: S4 specifically includes: For the human-machine tracking motion error system, that is, formula (13), the i-th controller with adaptive learning rate is designed as follows: in, represents the generalized inverse matrix of B(θ), Indicates e 2,i The generalized inverse matrix of (t), represents the estimated value of σ(t) in the i-th iteration learning, and the estimated error μ represents an adjustable parameter, and μ>|d|; The Lyapunov function designed for the i-th learning is as follows Taking the derivative of equation (17) and substituting equation (14) and equation (16) into equation (17) yields Define the function L i (t) are as follows: in make in Let V i (0) = 0, so we can get According to formula (23), we know that L i (t) is a decreasing function; When i = 0, the derivative of equation (19) yields From formula (18), we can know So we know that L0(t) is differentiable and its derivative is less than a bounded value, so L0(t) is continuous and bounded on t∈[0,T], where T represents the time of one learning. i (t) bounded; From formula (21), we can get According to equations (23) and (26), we can get get According to the convergence of the series, as the number of learning times increases, we can get And the learning rate error The rehabilitation robot realizes trajectory tracking and active walking speed tracking of the rehabilitator, and coordinated movement is realized in the active training mode of the human-machine system.

6. The human-machine coordinated motion control method for predicting active walking speed of a rehabilitated person according to claim 1, characterized in that: Based on the STM32F411 series MCU, the output PWM signal is provided to the motor drive module, so that the rehabilitation robot can help the rehabilitation person coordinate exercise training at a predicted walking speed. The specific implementation method is as follows: S101, use STM32F411 series microcontroller as the main controller; S102, the main controller receives a feedback signal from the motor speed measurement module and connects the output to the motor drive module; S103, the motor drive module is connected to the DC motor; S104, the power supply system supplies power to the main controller, the motor speed measurement module, the motor drive module and the DC motor; S105, the main controller reads the feedback signal of the motor encoder and compares it with the predicted walking speed to calculate an error signal; S106. Based on the error signal, the main controller calculates the required motor control amount according to a predetermined human-machine coordinated motion control method, and outputs the control amount to the motor drive module so that the motor rotates to drive the wheels to maintain balance and move in a specified manner.

7. The human-machine coordinated motion control method for predicting active walking speed of a rehabilitated person according to claim 6, characterized in that: The STM32F411 series single chip microcomputer outputs a PWM signal to the motor drive module to adjust the speed and direction of the motor.

8. The human-machine coordinated motion control method for predicting active walking speed of a rehabilitated person according to claim 6, characterized in that: The motor speed measurement module includes an encoder for providing real-time feedback information of the motor speed.

9. The human-machine coordinated motion control method for predicting active walking speed of a rehabilitated person according to claim 6, characterized in that: The human-machine coordinated motion control method adjusts the motor control amount based on the predicted walking speed of the rehabilitated person to achieve gait coordination in rehabilitation training.

Images