Multi-joint layered cooperative intelligent control method for power-assisted exoskeleton
By employing a multi-joint hierarchical collaborative intelligent control method, and utilizing an adaptive oscillator model and sensor data to optimize the assist torque of the hip and knee joints, the problem of uneven assistance in existing exoskeletons is solved, resulting in a smoother and more effective user experience.
Patent Information
- Application Number
- CN202410824173.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-25
- Publication Date
- 2025-12-26
AI Technical Summary
Existing lower limb assistive exoskeleton control methods result in uneven assistance, affecting normal gait and leading to a poor user experience.
A multi-joint hierarchical collaborative intelligent control method is adopted. By acquiring the real-time hip joint angle, an adaptive oscillator model is established, and an assist curve is constructed to show the relationship between the lower limb joint assist torque and gait phase. The actual torque is obtained by combining an inertial measurement unit and a force sensor, and the assist torque of the hip and knee joints is optimized to achieve coordinated assistance of the hip and knee joints.
It improves the effectiveness and smoothness of exoskeleton assistance, enhances the user experience, and avoids problems such as inaccurate gait phase segmentation and inconsistent assistance.
Smart Images
Figure CN121209601A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of robotic exoskeleton technology, and in particular to a multi-joint layered collaborative intelligent control method for assistive exoskeletons. Background Technology
[0002] When using assistive exoskeletons, a control algorithm needs to be designed to provide assistance to the human body without affecting the user's normal posture. Taking lower limb assistive exoskeletons as an example, they are often used in scenarios such as weight-bearing walking and weight-bearing uphill climbing. By setting motors in the hip and knee joints to provide assistance, they help the human body overcome movement obstacles, bear loads, and relieve lower limb fatigue.
[0003] Current lower limb assistive exoskeleton control methods are typically single-joint assist modes. They first use sensors to detect the user's current movement trend, and then provide assistance through hip and knee joint motors respectively, with the hip and knee joint assistance operating independently and without correlation. Since human walking is a process involving the coordinated effort of multiple joints, the force exertion follows a smooth curve.
[0004] Therefore, current methods for controlling lower limb assistive exoskeletons can lead to problems such as uneven assistance, disruption of normal gait, and poor user experience. Summary of the Invention
[0005] Based on the above analysis, the present invention aims to provide a multi-joint layered collaborative intelligent control method for assistive exoskeletons, in order to solve the problems of uneven assistance, impact on normal gait, and poor user experience caused by existing lower limb assistive exoskeleton control methods.
[0006] This invention provides a multi-joint layered collaborative intelligent control method for assistive exoskeletons, the method comprising the following steps:
[0007] The real-time hip joint angle is obtained, and an adaptive oscillator model is established using a sine curve. The parameters of the adaptive oscillator model are updated based on the real-time hip joint angle to obtain an updated adaptive oscillator model. The real-time gait phase is obtained based on the updated adaptive oscillator model. The real-time gait phase is segmented to obtain the gait phase of several first sampling points.
[0008] The system acquires the user's height, weight, lower limb joint center, and gait frequency to construct an assist curve relating the lower limb joint assist torque to the gait phase; the assist curve includes a hip joint assist curve and a knee joint assist curve.
[0009] The actual torque of the lower limb joint corresponding to each first sampling point is obtained. The gait phase corresponding to each first sampling point is input into the assist curve to obtain the lower limb joint assist torque. If the error between the lower limb joint torque and the actual torque is greater than the threshold, the lower limb joint assist torque is updated and an optimized assist curve is obtained. Lower limb joint assistance is realized based on the optimized assist curve.
[0010] Furthermore, the step of acquiring the user's height, weight, lower limb joint center, and gait frequency to construct the assist curve relating the lower limb joint assist torque to the gait phase includes:
[0011] Obtain the user's height and weight to determine the mass of their thigh and calf.
[0012] Obtain the user's thigh and calf lengths by measuring the center of the hip, knee, and ankle joints.
[0013] The moment of inertia of the thigh is obtained based on the mass and length of the thigh, and the moment of inertia of the lower leg is obtained based on the mass and length of the lower leg.
[0014] The corresponding hip and knee angles are obtained based on the gait phase of the first sampling point.
[0015] The hip joint assist torque corresponding to the first sampling point is obtained based on the thigh mass, thigh length, thigh rotational inertia, and hip joint angle. The knee joint assist torque corresponding to the first sampling point is obtained based on the lower leg mass, lower leg length, lower leg rotational inertia, and knee joint angle. The gait phase period is obtained based on the stride frequency. The hip joint assist torque and gait phase corresponding to the first sampling point are used to obtain the hip joint assist curve through Gaussian fitting. The knee joint assist torque and gait phase corresponding to the first sampling point are used to obtain the knee joint assist curve through Gaussian fitting.
[0016] Furthermore, the hip / knee joint assist torque is obtained using the following formula:
[0017]
[0018] Where, α k For the angular acceleration of the hip / knee joint, θ k The angle between the hip and knee joints is given by g, and the acceleration due to gravity is given by I. k Let m be the moment of inertia of the thigh / lower leg. k For thigh / calf mass, L k Thigh / calf length.
[0019] Furthermore, the actual torques of the hip and knee joints corresponding to each first sampling point are obtained through inertial measurement unit sensors and force sensors. The gait phase of each first sampling point is input into the hip joint assist curve to obtain the hip joint assist torque, and the gait phase of each first sampling point is input into the knee joint assist curve to obtain the knee joint assist torque.
[0020] Furthermore, inertial measurement unit (IMU) sensors are placed on the front of the thigh and calf, and force sensors are placed on the front and back of the thigh and calf. The IMU sensors acquire hip and knee joint angles, and the force sensors acquire the interaction forces between the thigh / calf and the lower limb exoskeleton. For each first sampling point: the corresponding hip and knee joint angles are obtained; the hip joint angular velocity is obtained based on the hip joint angle; the knee joint angular velocity is obtained based on the knee joint angle; hip flexion or extension is determined based on the hip joint angular velocity; and knee flexion or extension is determined based on the knee joint angular velocity. When hip flexion occurs, the force sensor on the front of the thigh... The hip joint lever arm is obtained by measuring the distance between the placement of the sensor and the center of the hip joint; when extending the hip, the hip joint lever arm is obtained by measuring the distance between the placement of the posterior thigh force sensor and the center of the hip joint; when flexing the knee, the knee joint lever arm is obtained by measuring the distance between the placement of the anterior lower leg force sensor and the center of the knee joint; when extending the knee, the knee joint lever arm is obtained by measuring the distance between the placement of the posterior lower leg force sensor and the center of the knee joint; the actual hip joint torque is obtained by measuring the interaction force between the hip joint lever arm and the thigh and lower limb exoskeleton, and the actual knee joint torque is obtained by measuring the interaction force between the knee joint lever arm and the lower leg and lower limb exoskeleton.
[0021] Furthermore, for each first sampling point: the hip joint torque error is obtained by the difference between the actual hip joint torque and the hip joint assist torque at that first sampling point. If the hip joint torque error is greater than a threshold, the actual hip joint torque is used to replace the hip joint assist torque, and the hip joint assist torque is updated. Similarly, the knee joint torque error is obtained by the difference between the actual knee joint torque and the knee joint assist torque at that first sampling point. If the knee joint torque error is greater than a threshold, the actual knee joint torque is used to replace the knee joint assist torque, and the knee joint assist torque is updated.
[0022] Furthermore, an optimized hip joint assist curve is obtained by fitting the updated hip joint assist torque to the hip joint assist curve using a Gaussian function; an optimized knee joint assist curve is obtained by fitting the updated knee joint assist torque to the knee joint assist curve using a Gaussian function.
[0023] Furthermore, the fitted hip joint angle is obtained based on the adaptive oscillator model; the hip joint angle fitting error is obtained based on the real-time hip joint angle and the fitted hip joint angle; if the hip joint angle fitting error does not reach the threshold, the parameters of the adaptive oscillator model are updated until the hip joint angle fitting error reaches the threshold. When the hip joint angle fitting error reaches the threshold, the updated adaptive oscillator model is obtained.
[0024] If the hip joint angle fitting error does not reach the threshold, the parameters of the adaptive oscillator model are updated based on the learning rate to obtain the updated fitted hip joint angle, until the difference between the real-time hip joint angle and the updated fitted hip joint angle is less than the threshold.
[0025] Furthermore, the adaptive oscillator model based on the sine curve is shown below:
[0026] X(t)=A(t)*sin(ω(t)*t+C(t))+B(t),
[0027] Where C(t) is the phase bias of the adaptive oscillator model, ω(t) is the frequency of the adaptive oscillator model, A(t) is the amplitude of the adaptive oscillator model, B(t) is the amplitude bias of the adaptive oscillator model, and X(t) is the learning objective of the adaptive oscillator model, i.e., fitting the hip joint angle.
[0028] Furthermore, the real-time gait phase is obtained using the following method:
[0029] φ(t)=ω(t)*t+C(t).
[0030] Compared with the prior art, the present invention can achieve at least one of the following beneficial effects:
[0031] 1. This invention achieves coordinated assistance of the hip and knee joints by associating the hip joint angle and its torque with the knee joint angle and its torque in the same step phase, thereby improving the effectiveness and smoothness of the exoskeleton assistance and enhancing the user experience.
[0032] 2. This invention obtains the real-time gait phase through an adaptive oscillator model. It eliminates the need to divide a gait cycle during a user's walking process into several different stages based on the support phase and the swing phase. Instead, it obtains a continuous gait phase through the adaptive oscillator model, thereby avoiding the problems of inaccurate gait phase division and inconsistent assistance caused by assisting based on discrete gait phases.
[0033] 3. This invention obtains real-time hip and knee joint assist torques through hip and knee joint assist curves and real-time gait phase. Since the assist curves are pre-designed, there is no need to identify the user's movement trend through sensors before providing assistance. Instead, it directly provides real-time hip and knee joint assist torques based on the user's gait cycle, which improves the effectiveness and smoothness of exoskeleton assistance and enhances the user experience.
[0034] In this invention, the above-described technical solutions can be combined with each other to achieve more preferred combinations. Other features and advantages of this invention will be set forth in the following description, and some advantages may become apparent from the description or be learned by practicing the invention. The objects and other advantages of this invention can be realized and obtained from what is particularly pointed out in the description and drawings. Attached Figure Description
[0035] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts.
[0036] Figure 1 This is a flowchart of a multi-joint layered collaborative intelligent control method for assistive exoskeletons according to an embodiment of the present invention;
[0037] Figure 2 This is a schematic diagram comparing the hip joint angle change with a sinusoidal signal according to an embodiment of the present invention;
[0038] Figure 3 This is a schematic diagram of a lower limb assistive exoskeleton according to an embodiment of the present invention. Detailed Implementation
[0039] Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, which form part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not intended to limit the scope of the present invention.
[0040] One specific embodiment of the present invention discloses a multi-joint layered collaborative intelligent control method for assistive exoskeletons. For example... Figure 1 As shown, the method includes the following steps:
[0041] Step S1: Obtain the real-time hip joint angle and establish an adaptive oscillator model using a sine curve; update the parameters of the adaptive oscillator model based on the real-time hip joint angle to obtain the updated adaptive oscillator model; obtain the real-time gait phase based on the updated adaptive oscillator model; segment the real-time gait phase to obtain the gait phase of several first sampling points.
[0042] Step S2: Obtain the user's height, weight, lower limb joint center, and cadence to construct an assist curve relating the lower limb joint assist torque to the gait phase; the assist curve includes a hip joint assist curve and a knee joint assist curve;
[0043] Step S3: Obtain the actual torque of the lower limb joint corresponding to each first sampling point, input the gait phase corresponding to each first sampling point into the assist curve to obtain the lower limb joint assist torque, if the error between the lower limb joint torque and the actual torque is greater than the threshold, update the lower limb joint assist torque and obtain the optimized assist curve, and realize lower limb joint assistance based on the optimized assist curve.
[0044] Specifically, in step S1, the fitted hip joint angle is obtained based on the adaptive oscillator model; the hip joint angle fitting error is obtained based on the real-time hip joint angle and the fitted hip joint angle; if the hip joint angle fitting error does not reach the threshold, the parameters of the adaptive oscillator model are updated until the hip joint angle fitting error reaches the threshold; when the hip joint angle fitting error reaches the threshold, the updated adaptive oscillator model is obtained.
[0045] Furthermore, if the hip joint angle fitting error does not reach the threshold, the parameters of the adaptive oscillator model are updated based on the learning rate to obtain the updated fitted hip joint angle, until the difference between the real-time hip joint angle and the updated fitted hip joint angle is less than the threshold.
[0046] Specifically, the unilateral gait during walking can be viewed as an alternating process of swinging and supporting. During the swinging phase, the hip joint flexes and the knee joint flexes; while during the supporting phase, the hip joint extends and the knee joint extends. Real-time hip joint angles are acquired using flexible angle sensors, such as... Figure 2 As shown, the hip joint angle changes sinusoidally, with the hip transition point being the change from hip flexion to hip extension or vice versa. Similarly, the knee joint angle also changes sinusoidally, with the knee transition point occurring slightly earlier than the hip transition point. Therefore, the period of change for both the hip and knee joint angles is 0-2π.
[0047] Furthermore, the adaptive oscillator model based on the sine curve is shown below:
[0048] X(t)=A(t)*sin(ω(t)*t+C(t))+B(t),
[0049] Where C(t) is the phase bias of the adaptive oscillator model, ω(t) is the frequency of the adaptive oscillator model, A(t) is the amplitude of the adaptive oscillator model, B(t) is the amplitude bias of the adaptive oscillator model, and X(t) is the learning objective of the adaptive oscillator model, i.e., fitting the hip joint angle.
[0050] Furthermore, the hip joint angle fitting error is the difference between the real-time hip joint angle and the fitted hip joint angle, which is obtained using the following formula:
[0051] F(t)=x(t)-A(t)*sin(ω(t)*t+C(t))+B(t),
[0052] Where F(t) is the hip joint angle fitting error, and x(t) is the real-time hip joint angle.
[0053] Furthermore, the learning rate is obtained using the following method:
[0054]
[0055] Where v is the lateral learning parameter, r is the longitudinal learning parameter, F(t) is the difference between the real-time hip joint angle and the fitted hip joint angle, and φ(t) is the real-time gait phase.
[0056] Furthermore, the real-time gait phase is obtained using the following method:
[0057] φ(t)=ω(t)*t+C(t).
[0058] Understandably, the adaptive oscillator model can obtain the real-time fitted hip joint angle, and the parameters of the adaptive oscillator model are updated based on the hip joint angle fitting error. During parameter updates, the learning rate determines the step size for parameter updates. When the hip joint angle fitting error is less than a threshold, the real-time gait phase is calculated using ω(t)*t+C(t). The real-time gait phase includes the corresponding hip joint angle and knee joint angle.
[0059] Understandably, this invention obtains the real-time gait phase through an adaptive oscillator model. Instead of dividing a gait cycle during a user's walking process into several different stages based on the support phase and the swing phase, it obtains a continuous gait phase through the adaptive oscillator model. This avoids the problems of inaccurate gait phase division and inconsistent assistance caused by assisting based on discrete gait phases.
[0060] Furthermore, the step of segmenting the real-time gait phase to obtain several first sampling points includes:
[0061] The real-time gait phase is divided into gait cycles of 0-2π and the gait phase of one gait cycle is obtained.
[0062] During this gait cycle Sixteen first sampling points are set at equal intervals, and the gait phase corresponding to each first sampling point is obtained.
[0063] Specifically, in step S2, obtaining the user's height, weight, lower limb joint center, and gait frequency to construct an assist curve relating the lower limb joint assist torque to the gait phase includes:
[0064] Obtain the user's height and weight to determine the mass of their thigh and calf.
[0065] Obtain the user's thigh and calf lengths by measuring the center of the hip, knee, and ankle joints.
[0066] The moment of inertia of the thigh is obtained based on the mass and length of the thigh, and the moment of inertia of the lower leg is obtained based on the mass and length of the lower leg.
[0067] The corresponding hip and knee angles are obtained based on the gait phase of the first sampling point.
[0068] The hip joint assist torque corresponding to the first sampling point is obtained based on the thigh mass, thigh length, thigh rotational inertia, and hip joint angle. The knee joint assist torque corresponding to the first sampling point is obtained based on the lower leg mass, lower leg length, lower leg rotational inertia, and knee joint angle. The gait phase period is obtained based on the stride frequency. The hip joint assist torque and gait phase corresponding to the first sampling point are used to obtain the hip joint assist curve through Gaussian fitting. The knee joint assist torque and gait phase corresponding to the first sampling point are used to obtain the knee joint assist curve through Gaussian fitting.
[0069] Specifically, the mass of a user's thigh or calf can be obtained using the formula M = a1·H + a2·W - a3, where H is the user's height, W is the user's weight, and a1, a2, and a3 are coefficients. a1, a2, and a3 are obtained through statistical data; the values of a1, a2, and a3 differ when calculating thigh mass from those when calculating calf mass. The user's thigh length is the distance from the center of the hip joint to the center of the knee joint, and the calf length is the distance from the center of the knee joint to the center of the ankle joint.
[0070] Furthermore, the moment of inertia of the thigh / lower leg is obtained using the following formula:
[0071]
[0072] Among them, I k Let m be the moment of inertia of the thigh / lower leg. k For thigh / calf mass, L k Thigh / calf length.
[0073] Furthermore, lower limb assistive exoskeletons such as Figure 3 As shown, the hip / knee joint assist torque is obtained through the following formula:
[0074]
[0075] Where, α k For the angular acceleration of the hip / knee joint, θ k denoted as the hip / knee joint angle, and g as the acceleration due to gravity.
[0076] Furthermore, the user's step frequency is acquired through an inertial measurement unit sensor, and the length of the gait phase period is obtained based on the step frequency.
[0077] Specifically, cadence is the number of steps taken per unit of time (1 minute), through... The length of the gait phase period can be obtained, where f is the stride frequency. The gait phase at a certain moment includes the hip joint angle and its direction at that moment. Based on the hip joint angle at that moment, the corresponding hip joint assist torque can be obtained. Therefore, the gait phase and its corresponding hip joint assist torque corresponding to all first sampling points can be obtained. Thus, the hip joint assist curve, i.e., the curve showing the relationship between gait phase and hip joint assist torque, can be obtained by fitting a Gaussian function. Similarly, the gait phase at a certain moment also includes the knee joint angle and its direction at that moment. Based on the knee joint angle at that moment, the corresponding knee joint assist torque can be obtained. Therefore, the gait phase and its corresponding knee joint assist torque corresponding to all first sampling points can be obtained. Thus, the knee joint assist curve, i.e., the curve showing the relationship between gait phase and knee joint assist torque, can be obtained by fitting a Gaussian function.
[0078] Furthermore, Gaussian fitting is performed using the following method:
[0079]
[0080] Where t is the independent variable, A i B i C i Let n be the parameter of the i-th Gaussian function, and n be the number of Gaussian functions.
[0081] Understandably, since the assist torque used to construct the assist curve is a theoretical value, the resulting assist curve needs to be optimized based on real-time gait phase and actual assist conditions. Because the changes in hip and knee joint angles follow a sinusoidal pattern with a period of 0-2π, the hip and knee assist curves also exhibit a sinusoidal pattern with a period of 0-2π.
[0082] Specifically, in step S3, the actual torques of the hip and knee joints corresponding to each first sampling point are obtained through the inertial measurement unit sensor and the force sensor. The gait phase of each first sampling point is input into the hip joint assist curve to obtain the hip joint assist torque, and the gait phase of each first sampling point is input into the knee joint assist curve to obtain the knee joint assist torque.
[0083] Specifically, inertial measurement unit (IMU) sensors are placed on the front of the thigh and calf, and force sensors are placed on the front and back of the thigh and calf. The IMU sensors are used to obtain the hip joint angle and knee joint angle, and the force sensors are used to obtain the interaction force between the thigh and calf and the lower limb exoskeleton. For each first sampling point: obtain its corresponding hip joint angle and knee joint angle; obtain the hip joint angular velocity based on the hip joint angle and the knee joint angular velocity based on the knee joint angle; determine hip flexion or extension based on the hip joint angular velocity and knee flexion or extension based on the knee joint angular velocity; when hip flexion occurs, obtain the hip joint lever arm by the distance between the placement position of the front thigh force sensor and the center of the hip joint; when hip extension occurs, obtain the hip joint lever arm by the distance between the placement position of the posterior thigh force sensor and the center of the hip joint; when knee flexion occurs, obtain the knee joint lever arm by the distance between the placement position of the anterior lower leg force sensor and the center of the knee joint; when knee extension occurs, obtain the knee joint lever arm by the distance between the placement position of the posterior lower leg force sensor and the center of the knee joint; obtain the actual hip joint torque by the interaction force between the hip joint lever arm and the thigh and lower limb exoskeleton, and obtain the actual knee joint torque by the interaction force between the knee joint lever arm and the lower leg and lower limb exoskeleton.
[0084] Specifically, the hip joint angular velocity is obtained by differentiating the hip joint angle and the knee joint angular velocity by differentiating the knee joint angle. When the hip joint angular velocity is greater than 0, it indicates hip flexion, and when it is less than 0, it indicates hip extension. Similarly, knee flexion or extension can be determined based on the knee joint angular velocity.
[0085] Furthermore, for each first sampling point: the hip joint torque error is obtained by the difference between the actual hip joint torque and the hip joint assist torque at that first sampling point. If the hip joint torque error is greater than a threshold, the actual hip joint torque is used to replace the hip joint assist torque, and the hip joint assist torque is updated. Similarly, the knee joint torque error is obtained by the difference between the actual knee joint torque and the knee joint assist torque at that first sampling point. If the knee joint torque error is greater than a threshold, the actual knee joint torque is used to replace the knee joint assist torque, and the knee joint assist torque is updated.
[0086] Furthermore, an optimized hip joint assist curve is obtained by fitting the updated hip joint assist torque to the hip joint assist curve using a Gaussian function; similarly, an optimized knee joint assist curve is obtained by fitting the updated knee joint assist torque to the knee joint assist curve using a Gaussian function.
[0087] Specifically, several second sampling points are set, and the hip joint assist curve within a gait cycle is sampled based on the second sampling points to obtain the hip joint assist torque at the second sampling points. The hip joint assist torque at the second sampling points and the updated hip joint assist torque are fitted with a Gaussian function to obtain the optimized hip joint assist curve.
[0088] Specifically, the number of the second sampling points is greater than the number of the first sampling points; if the gait phase corresponding to the second sampling point is the same as the gait phase corresponding to the first sampling point, and the first sampling point has an updated hip joint assist torque, then Gaussian fitting is performed based on the updated hip joint assist torque.
[0089] Furthermore, the real-time hip joint assist torque is obtained based on the optimized hip joint assist curve, and hip joint assist is achieved based on the real-time hip joint assist torque; similarly, the real-time knee joint assist torque is obtained based on the optimized knee joint assist curve, and knee joint assist is achieved based on the real-time knee joint assist torque.
[0090] Understandably, this invention achieves coordinated hip and knee joint assistance by associating hip joint angles and torques with the same gait phase, thereby improving the effectiveness and smoothness of exoskeleton assistance and enhancing the user experience. This invention obtains real-time hip and knee joint assistance torques through hip and knee joint assistance curves and real-time gait phase. Since the assistance curves are pre-designed, there is no need to first identify the user's movement trend through sensors before providing assistance; instead, it directly provides real-time hip and knee joint assistance torques based on the user's gait cycle, improving the effectiveness and smoothness of exoskeleton assistance and enhancing the user experience.
[0091] Compared with existing technologies, the beneficial effects of the multi-joint layered collaborative intelligent control method for assistive exoskeletons provided by this invention are as follows:
[0092] 1. This invention achieves coordinated assistance of the hip and knee joints by associating the hip joint angle and its torque with the knee joint angle and its torque in the same step phase, thereby improving the effectiveness and smoothness of the exoskeleton assistance and enhancing the user experience.
[0093] 2. This invention obtains the real-time gait phase through an adaptive oscillator model. It eliminates the need to divide a gait cycle during a user's walking process into several different stages based on the support phase and the swing phase. Instead, it obtains a continuous gait phase through the adaptive oscillator model, thereby avoiding the problems of inaccurate gait phase division and inconsistent assistance caused by assisting based on discrete gait phases.
[0094] 3. This invention obtains real-time hip and knee joint assist torques through hip and knee joint assist curves and real-time gait phase. Since the assist curves are pre-designed, there is no need to identify the user's movement trend through sensors before providing assistance. Instead, it directly provides real-time hip and knee joint assist torques based on the user's gait cycle, which improves the effectiveness and smoothness of exoskeleton assistance and enhances the user experience.
[0095] Those skilled in the art will understand that all or part of the processes of the methods described in the above embodiments can be implemented by a computer program instructing related hardware, and the program can be stored in a computer-readable storage medium. The computer-readable storage medium may be a disk, optical disk, read-only memory, or random access memory, etc.
[0096] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.
Claims
1. A multi-joint layered cooperative intelligent control method for assisting an exoskeleton, characterized by, The method comprises the following steps: Obtaining real-time hip joint angle, and establishing an adaptive oscillator model by using a sinusoidal curve; updating parameters of the adaptive oscillator model based on the real-time hip joint angle to obtain an updated adaptive oscillator model, and obtaining real-time gait phase based on the updated adaptive oscillator model; and segmenting the real-time gait phase to obtain gait phase of a plurality of first sampling points. Obtaining height, weight, lower limb joint center and step frequency of a user to construct a power curve of relationship between lower limb joint power torque and gait phase; the power curve comprises a hip joint power curve and a knee joint power curve. Obtaining actual torque of the lower limb joint corresponding to each first sampling point, inputting gait phase of each first sampling point into the power curve to obtain lower limb joint power torque, updating the lower limb joint power torque if the error between the actual torque and the lower limb joint torque is greater than a threshold value, and obtaining an optimized power curve, and realizing lower limb joint power based on the optimized power curve.
2. The multi-joint layered cooperative intelligent control method for a powered exoskeleton according to claim 1, wherein, The method of obtaining height, weight, lower limb joint center and step frequency of a user to construct a power curve of relationship between lower limb joint power torque and gait phase comprises: Obtaining height and weight of a user to obtain thigh and calf mass; Obtaining hip joint center, knee joint center and ankle joint center of a user to obtain thigh and calf length; Obtaining thigh moment of inertia based on thigh mass and length, and obtaining calf moment of inertia based on calf mass and length; Obtaining corresponding hip joint angle and knee joint angle based on gait phase of the first sampling point; Obtaining hip joint power torque corresponding to the first sampling point based on thigh mass, thigh length, thigh moment of inertia and hip joint angle, obtaining knee joint power torque corresponding to the first sampling point based on calf mass, calf length, calf moment of inertia and knee joint angle, obtaining period of gait phase based on step frequency, obtaining hip joint power curve by Gaussian fitting based on hip joint power torque corresponding to the first sampling point and gait phase, and obtaining knee joint power curve by Gaussian fitting based on knee joint power torque corresponding to the first sampling point and gait phase.
3. The multi-joint layered cooperative intelligent control method for a powered exoskeleton according to claim 2, wherein, The hip / knee joint power torque is obtained by the following formula: where α k is the hip / knee angular acceleration, θ k is the hip / knee angle, g is the gravitational acceleration, I k is the thigh / calf moment of inertia, m k is the thigh / calf mass, and L k is the thigh / calf length.
4. The multi-joint layered cooperative intelligent control method for a powered exoskeleton according to claim 1, wherein, Obtaining actual torque of the hip joint and the knee joint corresponding to each first sampling point by an inertial measurement unit sensor and a force sensor, inputting gait phase of each first sampling point into the hip joint power curve to obtain hip joint power torque, and inputting gait phase of each first sampling point into the knee joint power curve to obtain knee joint power torque.
5. The multi-joint layered cooperative intelligent control method for a powered exoskeleton according to claim 4, wherein, The inertial measurement unit sensor is placed on the front side of the thigh and the calf, and the force sensor is placed on the front side and the back side of the thigh and the calf, the hip joint angle and the knee joint angle are obtained through the inertial measurement unit sensor, and the interaction force between the thigh and the lower limb exoskeleton and the interaction force between the calf and the lower limb exoskeleton are obtained through the force sensor; for each first sampling point: the corresponding hip joint angle and knee joint angle are obtained, the hip joint angular velocity is obtained based on the hip joint angle, the knee joint angular velocity is obtained based on the knee joint angle, the hip flexion or the hip extension is judged according to the hip joint angular velocity, and the knee flexion or the knee extension is judged according to the knee joint angular velocity; when the hip is flexed, the hip joint force arm is obtained through the distance between the placement position of the thigh front side force sensor and the center of the hip joint; when the hip is extended, the hip joint force arm is obtained through the distance between the placement position of the thigh back side force sensor and the center of the hip joint; when the knee is flexed, the knee joint force arm is obtained through the distance between the placement position of the calf front side force sensor and the center of the knee joint; when the knee is extended, the knee joint force arm is obtained through the distance between the placement position of the calf back side force sensor and the center of the knee joint; the actual hip joint torque is obtained through the hip joint force arm and the interaction force between the thigh and the lower limb exoskeleton, and the actual knee joint torque is obtained through the knee joint force arm and the interaction force between the calf and the lower limb exoskeleton.
6. The multi-joint layered cooperative intelligent control method for a powered exoskeleton according to claim 4, wherein, For each first sampling point: the hip joint torque error of the first sampling point is obtained through the difference between the actual hip joint torque and the hip joint assist torque, if the hip joint torque error is greater than a threshold value, the actual hip joint torque is used to replace the hip joint assist torque, and the hip joint assist torque is updated; similarly, the knee joint torque error of the first sampling point is obtained through the difference between the actual knee joint torque and the knee joint assist torque, if the knee joint torque error is greater than a threshold value, the actual knee joint torque is used to replace the knee joint assist torque, and the knee joint assist torque is updated.
7. The multi-joint layered cooperative intelligent control method for a powered exoskeleton according to claim 6, wherein, The optimized hip joint assist curve is obtained through Gaussian function fitting based on the hip joint assist curve and the updated hip joint assist torque; the optimized knee joint assist curve is obtained through Gaussian function fitting based on the knee joint assist curve and the updated knee joint assist torque.
8. The multi-joint layered cooperative intelligent control method for a powered exoskeleton according to claim 1, wherein, The fitted hip joint angle is obtained based on the adaptive oscillator model; the hip joint angle fitting error is obtained based on the real-time hip joint angle and the fitted hip joint angle; if the hip joint angle fitting error does not reach a threshold value, the parameters of the adaptive oscillator model are updated until the hip joint angle fitting error reaches the threshold value, and the updated adaptive oscillator model is obtained when the hip joint angle fitting error reaches the threshold value; If the hip joint angle fitting error does not reach the threshold value, the updated fitted hip joint angle is obtained by updating the parameters of the adaptive oscillator model based on the learning rate until the difference between the real-time hip joint angle and the updated fitted hip joint angle is less than a threshold value.
9. The multi-joint layered cooperative intelligent control method for a powered exoskeleton according to claim 1, wherein, The adaptive oscillator model based on a sinusoidal curve is as follows: Wherein, C(t) is the phase bias of the adaptive oscillator model, ω(t) is the frequency of the adaptive oscillator model, A(t) is the amplitude of the adaptive oscillator model, B(t) is the amplitude bias of the adaptive oscillator model, X(t) is the learning target of the adaptive oscillator model, that is, fitting the hip joint angle.
10. The multi-joint layered cooperative intelligent control method for a powered exoskeleton according to claim 9, wherein, The real-time gait phase is obtained by the following method: φ(t) = ω(t) * t + C(t).