A flexible exoskeleton robot based on a series flexible elastic body drive unit and its control method

Abstract

The present invention discloses a flexible exoskeleton robot based on a series flexible elastic body drive unit, comprising a first fixing member and a second fixing member, wherein the first fixing member is provided with a driving assembly, and the driving assembly is dynamically connected to the second fixing member via a traction cable; the driving assembly comprises a motor, a winding pulley connected to the motor power, and a fixed pulley assembly, the traction cable is wound around the winding pulley and the fixed pulley assembly, and the two ends of the traction cable are respectively connected to the second fixing member via a flexible elastic tension sensor; the present invention also discloses a control method for the flexible exoskeleton robot. The robot of the present invention is soft, lightweight, and provides comfortable assistance. It can greatly improve the walking ability of the elderly and can also be used for rehabilitation training of people with muscle strength decline. It can also be used for walking assistance for normal people, essentially reducing the muscle strength and energy consumption of the human body in walking by reducing the degree of muscle activation.

Description

Technical Field

[0001] The present invention relates to the field of robotics, and in particular to a flexible exoskeleton robot based on a series flexible elastic body drive unit and a control method thereof. Background Art

[0002] Population aging is a common problem in many countries around the world. Elderly people suffer from reduced lower limb mobility due to diminished physical function. Furthermore, a significant number of people experience lower limb weakness and difficulty walking due to accidents or long-term physical labor requiring lower limbs. Therefore, an exoskeleton robot that can actively assist and enhance the wearer's walking ability has broad application prospects and demand. Currently available lower limb exoskeletons can be categorized by the drive structure and the degree of rigidity and flexibility of force transmission: rigid, flexible, and elastic with elastic elements in series.

[0003] Traditional rigid exoskeletons use rigid leg rods connected in parallel with the human body, and the motor and reducer apply torque to the hip joint through straps to provide assistance. This type of exoskeleton's assistance is highly dependent on its own rigid structure. When worn, the drive joint axis is easily offset relative to the human joint axis, thereby generating a large additional torque, which poses certain safety issues. At the same time, this type of exoskeleton is not sufficiently adaptable to the soft skin of the human body, and has fewer degrees of freedom, resulting in a poor wearing experience. At present, in order to solve the safety and comfort problems of traditional rigid exoskeletons, many inventions have been made to improve them, or directly target the field of flexible exoskeletons.

[0004] The patent specification with the announcement number CN111168648B discloses a four-degree-of-freedom hip joint exoskeleton walking-assisting robot based on flexible drive, including a four-degree-of-freedom joint freedom configuration module, a size adjustment module, a human-computer interaction module, a sensor system module and a belt plate. The size adjustment module includes a waist adjustment mechanism and a leg adjustment mechanism. The belt plates are symmetrically arranged at both ends of the waist adjustment mechanism. The four-degree-of-freedom joint freedom configuration module is movably connected to the side bottom of the belt plate. The human-computer interaction module includes a waist connection mechanism and a leg connection mechanism. The connecting mechanism is movably connected to the top side of the waist belt plate, the leg adjustment mechanism is connected to the bottom of the four-degree-of-freedom joint freedom configuration module, and the leg connecting mechanism is arranged at the bottom of the leg adjustment mechanism; although this solution can achieve a certain degree of flexibility and reduce impact by connecting elastic elements in series in the motor transmission chain, and can also increase the freedom, safety and comfort of the rigid exoskeleton by adding mechanisms, the weight and volume of this rigid exoskeleton are still its fatal weaknesses, which will cause a greater psychological burden when used by the elderly, and it has poor adaptability to the soft skin of the human body.

[0005] The patent specification with publication number CN110303478A discloses a flexible exoskeleton for assisting walking and its control method. The flexible exoskeleton is mainly composed of a control system, a detection system, a pneumatic flexible execution system and a tracheal assembly. The control system analyzes the user motion information and the pressure information of the flexible power-assisting execution system collected by the detection system, and recognizes and understands the lower limb movement intention based on the gait estimation model; based on the hip joint torque model, it calculates the corresponding instructions such as pneumatic switch, pressure and flow rate, executes the corresponding actions, and controls the negative pressure input and unloading process of the pneumatic flexible execution system in real time. The pneumatic flexible execution system converts the air pressure energy provided by the control system into mechanical energy that can achieve linear motion in real time, and provides the hip joint with the auxiliary torque required for flexion and extension in real time according to the user's walking posture, thereby achieving the purpose of assisted walking. This solution uses a motor, a reducer and a Bowden cable or a flexible rope as the driving force of a flexible exoskeleton. A tension sensor can be arranged at the traction end of the Bowden cable or the flexible rope, and the closed-loop force control algorithm can realize the underlying force tracking control. However, this ordinary tension sensor has a high stiffness, and the physical flexibility of the Bowden cable tube and the exposed Bowden cable itself is also limited. At the same time, the flexible rope will become tight and the stiffness will increase during traction. These will lead to the above-mentioned flexible exoskeleton's insufficient softness, tenderness and elasticity when fitting the skin, and poor adaptability to the force jump caused by the impact of walking on the ground, which to a certain extent affects the comfort of wearing and assistance. Summary of the Invention

[0006] One purpose of the present invention is to provide a flexible exoskeleton robot based on a series flexible elastic body drive unit, which is soft, light in weight, and comfortable in assisting. It can greatly improve the walking ability of the elderly, promote their daily exercise, and reduce the occurrence of common diseases in the elderly. It can also be used for rehabilitation training of people with muscle strength decline, reduce the workload of rehabilitation trainers, and alleviate the problem of tight human resources in hospitals to a certain extent. In addition, it can also be used for walking assistance for normal people, essentially reducing people's effort in walking by reducing the degree of muscle activation.

[0007] A flexible exoskeleton robot based on a series flexible elastic body drive unit comprises a first fixing member and a second fixing member, wherein the first fixing member is provided with a driving assembly, and the driving assembly is dynamically connected to the second fixing member via a traction cable;

[0008] The driving assembly includes a motor, a winding pulley connected to the motor power, and a fixed pulley group. The traction wire is wound around the winding pulley and the fixed pulley group. Both ends of the traction wire are connected to the second fixing member through a flexible elastic tension sensor.

[0009] In this solution, a series flexible elastic driving unit with integrated driving and sensing is designed. The driving unit itself is small in size, light in weight, and has good overall softness, thereby achieving the purpose of being comfortable to wear and able to adapt well to human movement.

[0010] In addition, the flexible elastic tensile sensor serves as both a tensile sensitive device that adapts to bending and twisting and an elastic transmission device for output force. It can use real-time stress and strain detection to eliminate the measurement error caused by the nonlinear force-position relationship of traditional elastic elements due to design and manufacturing, providing a hardware foundation for achieving more accurate force tracking control.

[0011] Preferably, the flexible elastic tension sensor adopts a flexible parallel plate capacitive strain sensor.

[0012] In this scheme, the principle of the flexible parallel plate capacitive strain sensor is that when it is subjected to tension, it will undergo elastic stretching, and the capacitance value will change to reflect its stress and strain. However, its capacitance value does not change when it is bent or twisted, and the measured tension value can be expressed as a function of the capacitance C.

[0013] F 拉 =f(C) Formula (5)

[0014] Preferably, the first fixing member and the second fixing member are straps.

[0015] In this solution, waist straps or leg straps can be selected according to the fixing position of the fixing member.

[0016] Another object of the present invention is to provide a control method for a flexible exoskeleton robot, which includes precise control of tension values, generation of desired tension values, and human-in-the-loop optimization of power assistance curves.

[0017] The control accuracy of traditional flexible exoskeletons, such as those using pneumatic artificial muscles, needs to be improved. However, this flexible exoskeleton robot can achieve higher control accuracy through closed-loop force control, and its power assistance effect is better than that of pneumatic artificial muscle flexible exoskeletons. At the same time, it reduces the degree of muscle activation of the human body through a more advanced human-in-the-loop optimization algorithm, objectively reducing people's effort in walking.

[0018] Preferably, the precise control of the tension value comprises the following steps:

[0019] (1) Feedback the measured value to the input terminal and compare it with the target value to obtain the deviation signal e(t).

[0020] e(t)=r(t)-c(t) Formula (1)

[0021] Among them, c(t) is the measured tension value, r(t) is the target tension value,

[0022] Use the proportional and differential quantities of the deviation signal to generate a portion of the control quantity u1(t).

[0023]

[0024] Among them, K P is the proportional coefficient, u P (t) is the proportional control quantity, K D is the differential coefficient; u D (t) is the differential control quantity; increasing K P It can reduce the steady-state error, but may increase the overshoot and increase K D It can reduce overshoot and help improve system response speed, but it may make the system overly sensitive to disturbances;

[0025] (2) Design the feedforward term according to some characteristics of the controlled object, and the feedforward control quantity u2(t) is designed as shown in formula (3).

[0026] u2(t)=K FW r(t) Formula (3)

[0027] Among them, K FW is the feedforward compensation coefficient;

[0028] (3) The total control quantity is the composite of the control quantity u1(t) and the feedforward control quantity u2(t), as shown in formula (4).

[0029] u(t)=u1(t)+u2(t) Formula (4).

[0030] Preferably, the generation of the expected tension value comprises the following steps:

[0031] (1) Use landmark events to segment the continuous walking gait and obtain the gait phase of the walking person at a certain moment through the gait phase recognition algorithm;

[0032] (2) setting a parameterized power assistance curve in which the power assistance value changes with the gait phase;

[0033] (3) Obtain the expected pulling force value from the power assistance curve according to the gait phase.

[0034] Further preferably, the gait phase recognition algorithm adopts an adaptive oscillator, a phase angle oscillator or a neural network.

[0035] Preferably, the human-in-the-loop optimization of the power assist curve is specifically as follows: measuring the degree of muscle activation of the human body by an electromyographic sensor, and using a black box optimization algorithm to obtain a set of parameter values ​​that can reduce the human body muscle activation the most.

[0036] Beneficial effects of the present invention:

[0037] (1) A series-connected quasi-direct drive motor, a pull wire, and a flexible elastic tension sensor are used as a driving unit to design a series-connected flexible elastic body that integrates drive and sensing. The overall weight is light, the volume is small, the softness is high, the degree of freedom is large, it can adapt well to the movement of the human body, it is very comfortable to wear, and the psychological burden on the user is small.

[0038] (2) The part of the series elastic body that transmits tension is composed of a flexible rope and a flexible elastic tension sensor. It is flexible and elastic and has a soft texture. It can fit the skin well. The tension is consistent with the direction of human muscle strength. It has good physical flexibility and is well adaptable to ground impact.

[0039] (3) The series flexible elastic body itself is driven and sensed in one. It is not only elastic and can adapt to bending and twisting, but also can use real-time stress and strain detection to eliminate the measurement error caused by the force-position nonlinear relationship of traditional elastic elements due to design and manufacturing, and can achieve more accurate force tracking control. At the same time, the most advanced human-in-the-loop optimization strategy is used to optimize the parameterized power assist curve in the human loop, striving to minimize the degree of muscle activation of the human body and objectively reduce the effort of the human body in walking. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is a schematic structural diagram of a series-connected flexible elastic body drive unit according to the present invention;

[0041] Figure 2 is a schematic diagram of a flexible elastic tension sensor;

[0042] Figure 3 A schematic diagram of a human body wearing the robot of the present invention;

[0043] Figure 4 Schematic diagram for knee joint assistance;

[0044] Figure 5 Schematic diagram for ankle joint assistance;

[0045] Figure 6 Schematic diagram of combined power assistance for the hip, knee, and ankle joints;

[0046] Figure 7 This is a control flow chart of the method of the present invention;

[0047] Figure 8 This is the block diagram of the adaptive oscillator model;

[0048] Figure 9 This is a schematic diagram of the parameterized boost curve;

[0049] Figure 10 Description of the Bayesian optimization algorithm. DETAILED DESCRIPTION

[0050] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0051] like Figure 1 As shown, a flexible exoskeleton robot based on a series flexible elastic body drive unit includes a first fixing part 1 and a second fixing part 2. The first fixing part 1 is provided with a driving component 3, and the driving component 3 is dynamically connected to the second fixing part 2 through a traction wire 4.

[0052] The driving assembly 3 specifically includes a quasi-direct drive motor 31, a winding pulley 32 connected to the power of the quasi-direct drive motor 31, and a fixed pulley group 33. The traction wire 4 is wound around the winding pulley 32 and the fixed pulley group 33. The two ends of the traction wire 4 are respectively connected to the second fixing member 2 through a flexible elastic tension sensor 5. Among them, the quasi-direct drive motor 31 is equipped with a reducer 34. The flexible elastic tension sensor 5 adopts a flexible parallel plate capacitive strain sensor. The structure of the flexible parallel plate capacitive strain sensor is as follows: Figure 2 shown.

[0053] In this embodiment, the first fixing member 1 and the second fixing member 2 can be selected as a waist strap or a leg strap according to the fixing position, and the number of the driving components 3 and the second fixing member 2 can also be increased as needed; taking the realization of hip joint extension assistance and flexion assistance as an example, the driving process is described, as shown in FIG. Figure 3 As shown, the first fixing member 1 is a waist strap, and the second fixing member 2 is two thigh straps. At the same time, two sets of driving components 3 are provided on the waist strap to drive the two thigh straps respectively. In addition, the connection points between the flexible elastic tension sensor 5 and the thigh straps are located at the front and back sides of the thigh.

[0054] The driving process is as follows:

[0055] The quasi-direct drive motor 31 drives the winding pulley 32 to rotate forward and reverse through the reducer 34. When the winding pulley 32 rotates clockwise, the winch pulls the wire 4, the wire is tightened on the winding-in side and relaxed on the winding-out side, and the tension is transmitted through the fixed pulley group 33, so that the flexible tension sensor 5 on the back of the thigh generates tension, and the tension is transmitted to the human thigh through the thigh strap, thereby achieving hip joint extension assistance; when the winding pulley 32 rotates counterclockwise, contrary to before, it will cause the flexible tension sensor 5 on the front of the thigh to generate tension, thereby achieving hip joint flexion assistance.

[0056] The number and position of the straps can be adjusted according to the different parts of the body that need assistance.

[0057] like Figure 4 As shown, the aforementioned waist strap is adjusted to two thigh straps, which are fixed to the two thighs respectively. A driving component is fixed on each thigh strap. The original thigh strap is moved to the calf, and finally knee joint assistance can be achieved.

[0058] like Figure 5 As shown, the waist strap and thigh strap are eliminated and two calf straps are used. A driving assembly is fixed on each calf strap, and the distal ends of the two flexible elastic sensors are respectively fixed to the instep and heel of the shoe, which can ultimately achieve ankle joint assistance.

[0059] like Figure 6 As shown, the drive assembly is still fixed on the waist strap. The length of the flexible elastic tensile sensor itself can be lengthened due to manufacturing conditions. A longer flexible elastic tensile sensor is taken, which passes through the front of the thigh strap and the back of the calf strap, and the distal end is fixed at the heel position. When stretched, it can assist the hip flexion, knee flexion and ankle plantar flexion. The flexible elastic tensile sensor on the back of the thigh can assist the hip extension, and finally achieve combined assistance for the hip, knee and ankle joints.

[0060] A control method for a flexible exoskeleton robot. In order to achieve precise control, a control component and a sensor component are added to the above-mentioned flexible exoskeleton robot; the control component specifically includes a control circuit board and a battery, and is encapsulated in a first fixing member 1; the sensor component includes but is not limited to an IMU inertial measurement unit, an electromyographic sensor, etc., and the IMU inertial measurement unit and the electromyographic sensor are arranged on corresponding parts of the human body.

[0061] The specific control process of the method of the present invention is as follows Figure 7 As shown, the method specifically includes precise control of the pulling force value, generation of the expected pulling force value, and human-in-the-loop optimization of the power assist curve.

[0062] Precise control of tension value:

[0063] The most basic requirement for controlling a flexible exoskeleton robot is to achieve precise control of the tension acting on the human body. That is, a desired tension value is provided to the controller as an input signal. The controller can generate a current control signal for the motor. The motor rotates to drive the pull wire, and the pull wire accurately applies the desired tension value to the human body through a flexible tension sensor.

[0064] Since the required force value during human motion is constantly changing, the controller is required to accurately track the rapidly changing input signal, that is, the controller is required to have a high response speed and a small tracking error. A major feature of the present invention is the integration of transmission and perception, that is, the flexible elastic tension sensor can measure the tension acting on the human body in real time while transmitting the tension, providing a hardware foundation for closed-loop feedback control. The measured value is fed back to the input end and compared with the target value to obtain the deviation signal e(t).

[0065] e(t)=r(t)-c(t) Formula (1)

[0066] Among them, c(t) is the measured tension value, r(t) is the target tension value,

[0067] Use the proportional and differential quantities of the deviation signal to generate a portion of the control quantity u1(t).

[0068]

[0069] Among them, K P is the proportional coefficient, u P (t) is the proportional control quantity, K D is the differential coefficient; u D (t) is the differential control quantity;

[0070] PD control does not need to know the characteristics of the controlled object, and only needs to adjust the coefficients to obtain a better control effect. PD control has the characteristics of advanced control, can sense the changes in input in advance, has a fast response speed, and can keep the closed-loop pole away from the imaginary axis of the complex plane, thereby improving the stability of the system. However, it cannot eliminate the steady-state error. Therefore, the present invention introduces a feedforward link on this basis. The feedforward term can be designed according to some characteristics of the controlled object. The feedforward control quantity u2(t) is designed as shown in formula (3).

[0071] u2(t)=K FW r(t) Formula (3)

[0072] Among them, K FW It is the feedforward compensation coefficient, which can be obtained by analyzing the controlled object. The feedforward link can further improve the response speed of the system, while reducing the pressure of the feedback link and reducing the steady-state error that cannot be eliminated by the PD link.

[0073] The final total control quantity, i.e., the motor current control signal, is a composite of the PD control quantity u1(t) and the feedforward control quantity u2(t), as shown in formula (4).

[0074] u(t)=u1(t)+u2(t) Formula (4).

[0075] In summary, the composite control strategy used in the present invention can well achieve fast, stable and accurate tension value control, providing a basis for subsequent control.

[0076] Generation of expected tension value:

[0077] After achieving precise tension tracking control, the next control task is to control the present invention to provide a suitable assist value during human walking, which requires consideration of the assist timing and assist value.

[0078] First, consider the issue of assist timing. Since human walking has obvious periodicity, the present invention uses landmark events such as the maximum hip flexion angle to segment the gait of continuous walking. The gait phase of each gait cycle increases from 0% to 100%. In order to obtain the gait phase of a walking human body at a certain moment, a gait phase recognition algorithm is required. Typical algorithms used in the present invention include but are not limited to adaptive oscillators, phase angle oscillators, neural networks, etc. This embodiment is introduced by taking the adaptive oscillator as an example. Its model block diagram is as follows: Figure 8 As shown, the model consists of a series of oscillators with ω(t) as the fundamental frequency. The oscillator with the i-th fundamental frequency in the model includes amplitude α(t), frequency iω(t) and phase The three state parameters are adaptively learned from the characteristics of the periodic or quasi-periodic input signal. The adaptive oscillator model has the ability to synchronize the input signal. The specific formula is shown in formula group (6).

[0079]

[0080] in, Represents the input signal θ r (t)Signal output synchronously with the oscillator The error between is used to learn the characteristics of the periodic input signal, V ω ,V η Represents the learning parameters in the model, which determines the speed at which the model synchronizes the periodic signal.

[0081] The inertial measurement unit (IMU) is used to measure the kinematic information of the human body during walking, including hip joint angle, angular velocity, and angular acceleration. One of the periodic signals, such as the hip joint angle, is used as the input of the adaptive oscillator. The adaptive oscillator learns the periodic input signal to obtain an estimated angular frequency value ω of the periodic signal. By multiplying the angular frequency value by time, a gait phase of 0-2π can be obtained. The gait phase value required for the next step of control can be obtained by converting it proportionally to 0%-100%.

[0082] After calculating the gait phase of the human body according to the above gait phase recognition algorithm, the problem of the power assist value at this moment needs to be solved. The present invention takes the maximum hip flexion angle as the starting point of the gait cycle and pre-sets a curve of power assist value changing with gait phase, which is called power assist curve. In order to facilitate the subsequent optimization of the power assist curve in the loop, the power assist curve adopts parameterized design, such as Figure 9 As shown, there are four shape control parameters, namely rise time Peak time Fall time and peak torque τE-p The first half of the curve is the torque value for assisting hip extension, and the second half of the curve is the assist value for assisting hip flexion. The two have the same shape but opposite assist values. The assist curve is a piecewise smooth quadratic function curve. A single assist curve can be uniquely determined based on the four parameters. The specific determination method is to calculate the piecewise function describing the curve based on the four parameters. The description of the piecewise function of the first 50% of the curve is shown in formulas (7)-(10), where It represents the gait phase. The function shape of the latter 50% is the same as that of the first half, except that the function value is opposite, so it will not be repeated here.

[0083]

[0084]

[0085]

[0086]

[0087] Human-in-the-loop optimization of the power curve:

[0088] The ultimate goal of overall control is to assist walking, reduce muscle activation, and objectively reduce walking effort. Therefore, muscle activation can be used as an indicator of the quality of the power assistance curve. When a person wearing an exoskeleton walks, changing the power assistance curve parameters changes the effectiveness of the assistance, which in turn affects muscle activation. The process of finding the set of power assistance curve parameters that minimizes muscle activation is called human-in-the-loop optimization. Muscle activation is measured using electromyography (EMG). Commonly used black-box optimization algorithms include Bayesian optimization and covariance matrix adaptive evolutionary strategies.

[0089] The human-in-the-loop optimization process of this embodiment uses the Bayesian optimization algorithm, and its algorithm description is as follows: Figure 10 As shown, the basic process is to perform Gaussian process regression on a black-box function f(x) to obtain a fitting model for the function, as well as its fitted mean and variance. The acquisition function is then used to obtain the next sampling point, and the fitted function model is updated. After a certain number of iterations, the independent variable value corresponding to the maximum value of the function among all explored points is the optimal parameter. For the current problem, the input of the black-box function f(x) is the parameter value in the power assistance curve, and the output of the function is the inverse of the human electromyographic signal value when the parameter is used for power assistance. This is a black-box function with a high evaluation cost. By performing Bayesian optimization on it, we can obtain the set of parameter values ​​that minimizes muscle activation in the human body with fewer evaluations, thereby objectively achieving power assistance for the human body.

[0090] Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A control method for a flexible exoskeleton robot based on a series flexible elastic body drive unit, characterized in that: The flexible exoskeleton robot includes a first fixing member and a second fixing member, wherein the first fixing member is provided with a driving assembly, and the driving assembly is dynamically connected to the second fixing member via a traction cable; The driving assembly includes a motor, a winding pulley connected to the motor power, and a fixed pulley group, the traction wire is wound around the winding pulley and the fixed pulley group, and both ends of the traction wire are connected to the second fixing member through a flexible elastic tension sensor; The control method includes: precise control of the pulling force value, generation of the desired pulling force value, and human-in-the-loop optimization of the power assist curve; The precise control of the tension value includes the following steps: (1) Feedback the measured value to the input terminal and compare it with the target value to obtain the deviation signal e(t). e(t)=r(t)-c(t) Formula (1) Among them, c(t) is the measured tension value, r(t) is the target tension value, Use the proportional and differential quantities of the deviation signal to generate a portion of the control quantity u1(t). Among them, K P is the proportional coefficient, u P (t) is the proportional control quantity, K D is the differential coefficient; u D (t) is the differential control quantity; (2) Design the feedforward term according to some characteristics of the controlled object, and the feedforward control quantity u2(t) is designed as shown in formula (3). u2(t)=K FW r(t) formula (3) Among them, K FW is the feedforward compensation coefficient; (3) The total control quantity is the composite of the control quantity u1(t) and the feedforward control quantity u2(t), as shown in formula (4). u(t)=u1(t)+u2(t) Formula (4) The generation of the expected tension value comprises the following steps: (1) Use landmark events to segment the continuous walking gait and obtain the gait phase of the walking person at a certain moment through the gait phase recognition algorithm; (2) Setting a parameterized power assistance curve in which the power assistance value changes with the gait phase, where Represents the gait phase, and the parameterized power assistance curve expression is as follows: in, represents the rise time, Indicates the peak time, represents the next time, τ E-p represents the peak torque; (3) Obtaining the expected pulling force value from the power assistance curve according to the gait phase; The human-in-the-loop optimization of the power assist curve is specifically as follows: measuring the degree of muscle activation of a person through an electromyographic sensor, and using a black box optimization algorithm to obtain a set of parameter values ​​that can reduce the human muscle activation the most.

2. The control method of the flexible exoskeleton robot based on the series flexible elastic body drive unit according to claim 1 is characterized in that: The flexible elastic tension sensor adopts a flexible parallel plate capacitive strain sensor.

3. The control method of the flexible exoskeleton robot based on the series flexible elastic body drive unit according to claim 1 is characterized in that: The first fixing member and the second fixing member are straps.

4. The control method of the flexible exoskeleton robot based on the series flexible elastic body drive unit according to claim 1 is characterized in that: The gait phase recognition algorithm adopts an adaptive oscillator, a phase angle oscillator or a neural network.

Images