Feedforward-admittance control method suitable for rope-driven exoskeleton

Abstract

The invention relates to a feedforward-admittance control method suitable for a rope-driven exoskeleton, which can ensure the control performance of a control system of the rope-driven exoskeleton during non-rhythmic movement of a human body. According to the invention, the inertial measurement unit and the motor encoder can be utilized to realize modeling of man-machine coupling kinematics; modeling and identification of the man-machine coupling rigidity are realized by fusing the tension information of the driving rope measured by the tension sensor; designing a feed-forward item of the control method by utilizing kinematics and a rigidity model of man-machine coupling; designing an iterative learning strategy, and estimating the critical length of the driving rope; an admittance control algorithm is designed, a feedforward item is fused, a feedforward-admittance control method is formed, and accurate control over the rope-driven exoskeleton under the non-rhythmic motion of the human body is achieved. The problem that the control effect of the exoskeleton robot is reduced when the exoskeleton robot faces irregular motion modes of human bodies, such as urgent acceleration and deceleration motion, is solved, and the application scene range of the exoskeleton robot is expanded.

Description

Technical Field

[0001] The invention relates to the technical field of exoskeleton robots and control systems, and in particular to a feedforward-admittance control method suitable for a rope-driven exoskeleton. Background Art

[0002] As a type of exoskeleton robot, the rope-driven exoskeleton robot drives the pulley to contract the driving rope through the rotation of the motor, thereby generating tension on the driving rope, bringing a more supple and bionic linear assistance to the target human joints. It is an important component of assistive robots and high-end assistance equipment. However, in order to play the expected role in the human-computer interaction process, the precise control capability of the rope-driven exoskeleton robot is the key to determining the assistive functionality. It is also the prerequisite for providing the expected assistance to the movement of patients, workers, and special operations personnel. Therefore, the research on precise control methods for rope-driven exoskeletons is of great significance.

[0003] Unlike the control methods of independent working robots such as unmanned vehicles, exoskeleton robots need to be worn on the user at all times and move in coordination. Therefore, in addition to considering the adaptability to the environment, the control must also take into account the adaptability to human movement. In particular, considering the low bandwidth of the exoskeleton body caused by the flexible drive rope and some flexible materials, how to achieve high-precision and high-synchronization rope-driven exoskeleton robot control methods is difficult. Human movement types can be divided into rhythmic movements, such as regular walking, and non-rhythmic movements, such as sudden stops, sudden acceleration and deceleration, and the transition process from static to motion, and from motion to static. Among them, there are currently targeted technologies for the control methods of human rhythmic movements, such as iterative learning control, but there is a lack of exoskeleton robot control methods for non-rhythmic human movements. Strictly speaking, non-rhythmic movements are more representative in human daily life, and any movement that is not periodic or repeatable can be considered non-rhythmic. Therefore, even if a person walks on flat ground, but the speed, frequency, step length and other characteristics change between different gait cycles, then the walking on flat ground can be considered as non-rhythmic. Since non-rhythmic movements do not have obvious periodicity, the current exoskeleton control methods that use historical control cycle data, such as iterative learning control by control cycle, are no longer applicable, and open-loop curve tracking control is even more undesirable. Therefore, it is necessary to invent new solutions to solve the problem of precise control of rope-driven exoskeleton robots for non-rhythmic movements. Summary of the invention

[0004] In view of the shortcomings of the prior art, the present invention provides a feedforward-admittance control method suitable for a rope-driven exoskeleton. Without utilizing historical control information, the present invention can ensure the precise control performance of the rope-driven exoskeleton under non-rhythmic movements of the human body, and can effectively promote the practical application of exoskeleton robots in human daily life.

[0005] The technical solution adopted by the present invention to achieve the above-mentioned purpose is: a feedforward-admittance control method suitable for a rope-driven exoskeleton, which performs the following steps to realize modeling and control of a rope-driven exoskeleton in a non-rhythmic motion state, and the method comprises the following steps:

[0006] The kinematic modeling of the human-machine coupling body is realized by using the human joint angle and the driving rope length information; the stiffness modeling and identification of the human-machine coupling body is realized by using the driving rope length and tension information; the feedforward term of the robot control is constructed by using the kinematic and stiffness model of the human-machine coupling body;

[0007] The critical length of the driving rope in the feedforward term is estimated online by using an iterative learning strategy. The length change of the driving rope for the desired tension of the current driving rope is obtained from the perspective of feedback control by using admittance control. The length control amount of the driving rope is calculated by using the critical length and the length change of the driving rope, and the length control amount of the driving rope is converted into the speed control amount by using proportional-integral-damping control.

[0008] The feedback control term is superimposed on the speed control quantity derived from the feedforward term and input to the driver to achieve the final power assist control.

[0009] The human body joint angle information is measured by placing a rotary encoder at the human body joint position, or by using two inertial measurement units of adjacent limbs to measure and solve; the drive rope length information is obtained by online estimation using the motor encoder and rope wheel model information.

[0010] The kinematic model of the human-machine coupling body is composed of two parts: one part is the change in length A(t) of the driving rope caused by the rotational movement of the human joint; and the other part is the length value C(t) of the driving rope itself.

[0011] The tension information of the driving rope is measured by a tension sensor to measure the tension at the end of the driving rope. The stiffness identification process is carried out by an experimental method in which the subject wears the exoskeleton robot, keeps a fixed posture, and then synchronously measures the length and tension information of the driving rope to obtain the stiffness coefficient K of the exoskeleton body by fitting. exo .

[0012] The kinematics and stiffness models of the human-machine coupling body are used to construct the feedforward term, which is expressed as follows:

[0013] L(t)=A(t)+B(t)+C(t)

[0014] Where L(t) represents the overall feedforward term of the expected length of the drive rope, that is, the ideal change in the length of the drive rope obtained in advance based on prior knowledge; A(t) and C(t), as mentioned above, are feedforward terms related to the kinematic model of the human-machine coupling body; B(t) is a feedforward term related to the stiffness model of the human-machine coupling body.

[0015] The iterative learning strategy is used to estimate the critical length of the driving rope, where critical refers to the critical state of the driving rope being taut and relaxed; the symbol P is used here to represent the total length of the driving rope, that is, the sum of the exposed length and the length in the casing, and the estimation process is expressed as follows:

[0016]

[0017] in is the total length of the driving rope in the critical state; the superscript m indicates the mth gait cycle; t sMF It indicates the moment when the driving rope tension is the largest during the swing phase of the lower limbs of the human body; P mea Indicates the actual measured total length of the drive rope; F mea Indicates the actual measured driving rope tension; K coupled refers to the stiffness of the human-machine coupling body, which is the estimated exoskeleton stiffness K exo and human joint stiffness K joint Finally, since the length of the driving rope in the casing is a fixed value, assuming it is D, it can be calculated by To obtain the exposed critical length of the drive rope in the mth gait cycle, i.e. C(t).

[0018] The admittance control and the change in the length of the driving rope obtained by the admittance control for the current desired tension of the driving rope are expressed as follows:

[0019]

[0020] Where ΔP is the change in length of the drive rope; F des is the expected assist value; M, B, K are the three parameters of inertia, damping and stiffness of nanocontrol, which are used to determine the servo compliance of the control process; s represents the complex domain.

[0021] The process of converting the driving rope position into velocity through proportional-integral-damping control is as follows:

[0022]

[0023] Where V cmd is the converted speed quantity; K p , K i , K dare the coefficients of proportional, integral and damping respectively.

[0024] The method collects the change of joint angle through the inertia unit, collects the tension at the end of the rope through the tension sensor, and collects the change of the length of the driving rope through the motor encoder.

[0025] The present invention has the following beneficial effects and advantages:

[0026] The present invention uses the kinematics and stiffness modeling of the human-machine coupling body to obtain the feedforward term, and then obtains the feedback term through admittance control, and integrates the feedforward and feedback terms to achieve precise control of the rope-driven exoskeleton robot facing the non-rhythmic motion of the human body. This method does not use historical control information, but through a method combining feedforward and feedback, it can naturally shield the performance degradation problem caused by the non-periodicity of non-stationary motion to conventional control. This method can still ensure good power tracking control accuracy in the application tasks of human emergency stop, variable pace, and variable period. It has the advantages of strong adaptability and low environmental requirements, and is more suitable for application in the daily life of the human body to provide assistance for various movements and task modes of the human body. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a schematic diagram of kinematics and stiffness modeling of the human-machine coupling body in the present invention;

[0028] Figure 2 It is a diagram of the identification process of the stiffness coefficient in the present invention;

[0029] Figure 3 is a block diagram of the overall control method of the present invention;

[0030] Figure 4 It is a diagram of the components and tracking conditions of the control quantity in the present invention;

[0031] Figure 5 It is a tracking control effect diagram of the actual power-assisting curve in the present invention;

[0032] Figure 6 It is a bandwidth test result diagram of the control method in the present invention. DETAILED DESCRIPTION

[0033] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation method of the present invention is described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the invention, so the present invention is not limited by the specific implementation disclosed below.

[0034] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0035] The control method of the present invention can ensure the control performance of the control system of the rope-driven exoskeleton robot during non-rhythmic human motion without using historical control information. The invention can use the inertial measurement unit and the motor encoder to realize the modeling of human-machine coupling kinematics; by integrating the driving rope tension information measured by the tension sensor, the modeling and identification of the human-machine coupling stiffness can be realized; the kinematics and stiffness model of human-machine coupling is used to design and form the feedforward term of the control method; an iterative learning strategy is designed to iteratively estimate the critical length of the driving rope to compensate for the estimated deviation of the driving rope length caused by the movement of the exoskeleton in the body; an admittance control algorithm is designed to integrate the feedforward term to form a feedforward-admittance control method, and based on the driver speed control, the precise control of the rope-driven exoskeleton under non-rhythmic human motion is realized. The present invention solves the problem of decreased control effect of the exoskeleton robot when facing irregular human motion patterns, such as rapid acceleration and deceleration, expands the scope of application scenarios of the exoskeleton robot, and helps promote the daily application and development of the exoskeleton robot.

[0036] Taking the rope-driven exoskeleton robot for ankle joint motion assistance as an example, the robot provides plantar flexion assistance to the human ankle joint in the stance phase, and controls the driving rope to be in a relaxed state in the swing phase without providing assistance. The specific implementation method is described below. Figure 1 As shown in the figure, (L1+L2) represents the total length of the artificial tendon (the artificial tendon is defined as the distance between the fixing point of the sleeve of the driving rope and the fixing point of the driving rope); F represents the auxiliary pulling force provided by the robot via the driving rope; r represents the equivalent auxiliary force arm relative to the human joint; φ joint Indicates the human joint angle; K exo , K sheath , K joint , K cable and K end They represent the stiffness coefficients of the robot wearable part, the drive rope sleeve, the human joints, the drive rope and the fixed part of the drive rope end respectively; g(*) and f(*) represent the stiffness coupling functions respectively. The exoskeleton robot generates auxiliary force by contracting the drive rope and shortening the length of the artificial tendon. By modeling L1 and L2, we can know the corresponding relationship between the length of the artificial tendon and the tension, and thus obtain the corresponding relationship between the length of the drive rope and the tension, and realize feedforward control to improve the tracking control performance of the desired assistance curve. The length of L1 can be modeled as follows:

[0037]

[0038] Where t and t0 represent the current time and the starting time of the exoskeleton operation respectively; ΔL1 represents the offset, which is the sliding displacement of the exoskeleton wearable part under repeated stretching. Please note that the exoskeleton force F(t) here does not represent the parameter function about the time t, but represents the driving rope tension value at the time t. L2 can be modeled as:

[0039]

[0040] Define the joint angle φ joint The value is zero when the human body is standing upright. exo and K joint K sheath , K cable and K end It is 1-2 orders of magnitude smaller, so the total length L of the artificial tendon can be expressed as follows:

[0041]

[0042] Where K coupled is the human-machine coupling stiffness (K exo ×K joint / (K exo +K joint )). To solve this formula, we need to solve A(t), B(t) and C(t) separately.

[0043] ■A(t) represents the change in length of the artificial tendon caused by the movement of the human joint, where the value of the lever arm r can be obtained by manual measurement; and the angle φ joint Can be measured using two inertial measurement units or one encoder.

[0044] ■B(t) represents the change in length of the artificial tendon caused by the tension of the driving rope, which involves the stiffness of the wearable part of the exoskeleton robot and the human joint. The human joint stiffness K joint can be regarded as a constant; while the stiffness K of the robot's wearable part exo It can be obtained through experiments. Figure 2 As shown, K exo It is considered as a fixed value of 123.1N / cm. It can be seen from the figure that the main body of the exoskeleton will gradually slide down under repeated stretching. This phenomenon is particularly obvious in the first few gaits, and tends to be stable after 7-10 gaits. This shows that it is necessary to estimate the offset ΔL1 in C(t).

[0045] ■C(t) represents the critical length of the artificial tendon, that is, the critical state between tension and relaxation. Without considering ΔL1, this value is only related to the driving rope tension and the human joint angle. However, due to the existence of ΔL1, the C(t) value is variable even under fixed tension and fixed angle. Considering that ΔL1 changes slowly, it is assumed to be a constant value within a gait cycle, so the driving rope length P when the artificial tendon is in a critical state can be estimated. base As follows (note that in the specific embodiment, L represents the length of the artificial tendon, and P represents the total length of the drive rope):

[0046]

[0047] Where superscript m represents the mth gait cycle; subscript mea represents the measured value; P represents the total length of the drive rope; t SMF Indicates the moment of maximum tension of the driving rope during the swing phase. The data information of the maximum tension moment during the swing phase is selected for critical length estimation. On the one hand, it is to reduce the control computing load during the stance phase. On the other hand, because the tension of the driving rope during the swing phase is small but can reach a taut state, it can reduce the influence of the stiffness estimation error on the critical length estimation of the artificial tendon. Since the difference between the driving rope length and the artificial tendon length is a constant, C can be obtained at the same time.

[0048] During the swing phase, the drive rope is released by position feedback control, and the drive rope release command issued can be expressed as follows:

[0049]

[0050] in represents the driving rope length command issued by the controller in the swing phase of the mth gait cycle, and the driving rope will be fixed at this length until the end of the swing phase; F SW Indicates the expected maximum tension in the swing phase, which can be selected from 3N to 5N; F SMF Indicates the maximum tension of the swing phase actually measured; δ SW is a coefficient that determines the update rate of the length command of the drive rope in the swing phase. The larger the coefficient value, the more drastic the update amplitude. You can choose 0.6-0.8. It can be seen that the length command of the drive rope in the swing phase can be updated step by step, so that the maximum drive rope tension in the swing phase gradually converges to the set expected value F SW .

[0051] During the stance phase, in the rising phase of the desired assistance curve, when the desired force value is less than 5N, it is set to 5N to make the drive rope reach a taut state as soon as possible. Then use the estimated A, B, and C parameters, where A and B are used for feedforward speed control to assist tension tracking; C is used for feedback admittance control to assist in pre-tightening the drive rope. Admittance control uses tension deviation to generate a position correction ΔP as follows:

[0052]

[0053] Among them, M, B and K are the admittance parameters, which determine the dynamic compliance of the force control process. The desired length command of the drive rope can be expressed as follows:

[0054] P cmd (t) = max(P base -r·φ joint , P mea )+ΔP

[0055] The function max(*) indicates taking a larger value; P base -r·φ joint It is used to realize the pre-tightening operation of the drive rope. By adopting a position controller based on proportional-integral-damping injection, a position command P cmd (t) The corresponding speed command V cmd (t) are as follows:

[0056]

[0057] Where K p , K i , K d are the coefficients of proportion, integration and damping respectively; s represents the complex domain. Then, the feedforward speed control is introduced, and the final speed command V fin (t) are as follows:

[0058]

[0059] where dφ DF (t) / dt represents the joint angular velocity; F des (t) is the desired power curve function. The driver is set in speed mode, and the final command V fin It is sent to the driver to realize the assisted curve tracking control of the exoskeleton robot.

[0060] The overall block diagram of the control method is as follows: Figure 3As shown in the figure, for an exoskeleton robot control system, it is necessary to first use sensors, such as inertial measurement units, to collect information data about the user's body and the exoskeleton robot, and then generate a power assistance curve corresponding to the current scene based on the gait activity recognition results. The power assistance curve is sampled with the control frequency as the sampling frequency to obtain the expected power assistance point F des The method of the invention first obtains the speed command V for the current desired power point through admittance control and proportional-integral-damping control. cmd At the same time, the feedforward control will obtain the feedforward term dL / dt based on the current known information and the prior model, where L is the length of the artificial tendon mentioned above; the feedback term and the feedforward term are fused to generate the final command V fin , and then input into the driver to achieve tracking control of the desired power point; this process is a separate cycle process, usually with a frequency of 500Hz or 1000Hz, to achieve precise control of the entire power curve and complete the assistance of human body movement.

[0061] In order to verify the effectiveness of the above method, an experimental work was carried out using a rope-driven exoskeleton robot for assisting ankle plantar flexion as a platform. In the experiment, the admittance parameters M = 0.03, B = 0.01 and K = 1.20 were set; the equivalent assist arm r = 0.075m; the expected assistance curve was given as a piecewise Gaussian function curve, and the swing angle θ of the lower leg was taken as the independent variable. The expected curve can be expressed as follows:

[0062]

[0063] The parameters of the two Gaussian curves are selected as: A = 150N, μ = 12.434°, σ1 = 9.031°, σ2 = 5.775°. The two Gaussian function curves are connected at the peak moment to form a complete expected curve. The control process data are all recorded online into files in the PCM3365 industrial computer, and the subsequent processing of the data, such as gait cycle segmentation, drawing, error calculation, etc., are all completed in Matlab. It should be noted that this experiment is to verify whether the control performance of the proposed control method can be guaranteed when the control method does not use historical gait cycle information for non-rhythmic movements (the historical gait cycle information of non-rhythmic movements is not of reference significance).

[0064] Because non-rhythmic movements are not continuous, if the exoskeleton control verification experiment is only conducted for non-rhythmic activities, the amount of data that can be collected is very small. However, for the control method itself, the essential reason for distinguishing between rhythmic and non-rhythmic movements depends on whether the control method uses historical cycle data. If the rules of the control method itself are based on historical cycle data, then it is necessary to distinguish between rhythmic and non-rhythmic movements. Therefore, we can assume here: considering that the proposed feedforward-admittance control method does not use historical cycle data, as long as it is verified that the proposed control method can adapt to rhythmic movements, it can also be proved that the method has the ability to adapt to non-rhythmic movements (because as long as the historical gait cycle data is not used, for the controller, rhythmic and non-rhythmic movements are the same type of movement). Therefore, during the experiment, the subjects were required to walk on the treadmill at the most comfortable speed for the individual (4.8km / h), and the experimenters would not impose any movement obstacles on the subjects during the walking process.

[0065] By the final instruction V fin It can be seen from the composition formula that it can be composed of three parts: the feedback control amount generated by the admittance control (hereinafter referred to as FB), the feedforward amount generated by the joint motion (hereinafter referred to as FF_DF), and the feedforward amount generated by the expected power assist curve and the stiffness model (hereinafter referred to as FF_FO). Each part is as follows Figure 4 (above) (the error bars in the figure are the standard deviation of the data of 10 gait cycles). After the speed command curve is sent to the driver, the driver will control the motor to track the speed and form an actual speed curve, such as Figure 4 (below). It can be seen that compared with the instructions (FB) generated by feedback, the feedforward instructions (FF_DF, FF_FO) are advanced. For example, at about 50% of the gait cycle, the feedforward value (FF_DF) can increase the speed instruction in advance, corresponding to the movement of the human joints, and pre-stretch the steel wire, so that the exoskeleton can provide sufficient and fast auxiliary force to the human body in the pre-swing phase of the human body. The feedforward value (FF_FO) modifies the speed instruction in advance according to the rate of change of the expected curve during the assist curve tracking process to reduce the tracking delay. The curve corresponding to the speed instruction in the above figure is as follows Figure 5 As shown in the figure, even without using the data of historical gait cycles, accurate tracking and control of the power-assistance curve can be achieved by only using the kinematics and stiffness model of the human-machine coupling body. By collecting data from 10 consecutive gait cycles and calculating the performance of the control method, the root mean square error of the tracking control of the entire curve is: 0.309N±3.360N; the root mean square error of the tracking control of the peak force is: 1.307N±0.830N.

[0066] In addition to verifying the tracking control performance of the control method, the bandwidth of the control method was also evaluated. The experimental scenario for the bandwidth test of the control method is to instruct a subject to wear an exoskeleton robot on the right leg and keep the right leg in the pre-swing period. The control method then issues a sinusoidal assist command (20N at the trough and 120N at the peak). The frequency of the sinusoidal curve increases from 0.1Hz to 20Hz, and the incremental step is 0.1Hz, and the maintenance time at each frequency stage is 1s. Matlab is used to perform Fourier calculations on the collected expected force and actual force data, and to draw a Bode diagram. Figure 6 As shown, it can be seen that the bandwidth of this control method is 11.0 Hz without relying on historical periodic data, and the phase delay at this time is 150.3°, which can meet the application requirements of the rope-driven exoskeleton robot.

[0067] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should be regarded as within the scope of protection of the present invention.

Claims

1. A feedforward-admittance control method for a rope-driven exoskeleton, characterized in that: The following steps are performed to realize the modeling and control of the rope-driven exoskeleton in a non-rhythmic motion state. The method includes the following steps: The kinematic modeling of the human-machine coupling body is realized by using the human joint angle and the driving rope length information; the stiffness modeling and identification of the human-machine coupling body is realized by using the driving rope length and tension information; the feedforward term of the robot control is constructed by using the kinematic and stiffness model of the human-machine coupling body; The critical length of the driving rope in the feedforward term is estimated online by using an iterative learning strategy. The length change of the driving rope for the desired tension of the current driving rope is obtained from the perspective of feedback control by using admittance control. The length control amount of the driving rope is calculated by using the critical length and the length change of the driving rope, and the length control amount of the driving rope is converted into the speed control amount by using proportional-integral-damping control. The feedback control term is superimposed on the speed control quantity derived from the feedforward term and input to the driver to achieve the final power assist control.

2. A feedforward-admittance control method suitable for a rope-driven exoskeleton according to claim 1, characterized in that: The human body joint angle information is measured by placing a rotary encoder at the human body joint position, or by using two inertial measurement units of adjacent limbs to measure and solve; the drive rope length information is obtained by online estimation using the motor encoder and rope wheel model information.

3. A feedforward-admittance control method suitable for a rope-driven exoskeleton according to claim 1, characterized in that: The kinematic model of the human-machine coupling body is composed of two parts: one part is the change in length A(t) of the driving rope caused by the rotational movement of the human joint; and the other part is the length value C(t) of the driving rope itself.

4. A feedforward-admittance control method for a rope-driven exoskeleton according to claim 1, characterized in that: The tension information of the driving rope is measured by a tension sensor to measure the tension at the end of the driving rope. The stiffness identification process is carried out by an experimental method in which the subject wears the exoskeleton robot, keeps a fixed posture, and then synchronously measures the length and tension information of the driving rope to obtain the stiffness coefficient K of the exoskeleton body by fitting. exo .

5. A feedforward-admittance control method suitable for a rope-driven exoskeleton according to claim 1, characterized in that: The kinematics and stiffness models of the human-machine coupling body are used to construct the feedforward term, which is expressed as follows: L(t)=A(t)+B(t)+C(t) Where L(t) represents the overall feedforward term of the expected length of the drive rope, that is, the ideal change in the length of the drive rope obtained in advance based on prior knowledge; A(t) and C(t), as mentioned above, are feedforward terms related to the kinematic model of the human-machine coupling body; B(t) is a feedforward term related to the stiffness model of the human-machine coupling body.

6. A feedforward-admittance control method suitable for a rope-driven exoskeleton according to claim 1, characterized in that: The iterative learning strategy is used to estimate the critical length of the driving rope, where critical refers to the critical state of the driving rope being taut and relaxed; the symbol P is used here to represent the total length of the driving rope, that is, the sum of the exposed length and the length in the casing, and the estimation process is expressed as follows: in is the total length of the driving rope in the critical state; the superscript m indicates the mth gait cycle; t SMF It indicates the moment when the driving rope tension is the largest during the swing phase of the lower limbs of the human body; P mea Indicates the total length of the drive rope actually measured; F mea Indicates the actual measured driving rope tension; K coupled refers to the stiffness of the human-machine coupling body, which is the estimated exoskeleton stiffness K exo and human joint stiffness K joint Finally, since the length of the driving rope in the casing is a fixed value, assuming it is D, it can be calculated by To obtain the exposed critical length of the drive rope in the mth gait cycle, i.e. C(t).

7. A feedforward-admittance control method for a rope-driven exoskeleton according to claim 1, characterized in that: The admittance control and the change in the length of the driving rope obtained by the admittance control for the current desired tension of the driving rope are expressed as follows: Where ΔP is the change in length of the drive rope; F des is the expected assist value; M, B, K are the three parameters of inertia, damping and stiffness of nanocontrol, which are used to determine the servo compliance of the control process; s represents the complex domain.

8. A feedforward-admittance control method for a rope-driven exoskeleton according to claim 1, characterized in that: The process of converting the driving rope position into velocity through proportional-integral-damping control is as follows: Where V cmd is the converted speed quantity; K p , K i , K d are the coefficients of proportional, integral and damping respectively.

9. A feedforward-admittance control method for a rope-driven exoskeleton according to claim 1, characterized in that: The method collects the change of joint angle through the inertia unit, collects the tension at the end of the rope through the tension sensor, and collects the change of the length of the driving rope through the motor encoder.

Images