Human-computer interaction control method for intelligent lower limb exoskeleton with body

Abstract

The invention discloses a human-computer interaction control method for an intelligent lower limb exoskeleton with a body, and relates to the technical field of human-computer interaction, and the method comprises the following steps: S1, constructing a human exoskeleton coupling dynamic model; s2, performing joint calculation on the human exoskeleton coupling dynamic model through a kinematics equation to obtain calculation data; s3, inputting the human exoskeleton coupling dynamic model, the resolving data and the impedance evaluation index into an evaluator to obtain a real-time coupling adaptability score; s4, switching a corresponding man-machine interaction control strategy based on the real-time coupling adaptability score; and S5, minimizing the reward function based on a reinforcement learning algorithm so as to regulate and control the control parameters of the switched man-machine interaction control strategy. The assisting effect and wearing comfort of the lower limb exoskeleton can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of human-computer interaction, and more particularly to a somatic intelligent lower extremity exoskeleton human-computer interaction control method. BACKGROUND

[0002] At present, lower extremity exoskeletons have realized preliminary application in the fields of rehabilitation, walking aid, industry and military through multi-sensor fusion and various control strategies, but in actual use, the control method thereof is mostly dependent on preset trajectories, fixed impedance parameters or static gait patterns, and lacks adaptability to individual differences of users and dynamic changes of motion states. However, since most of the existing control strategies do not deeply consider the bidirectional coupling interaction mechanism between human and exoskeleton, the exoskeletons have a lagged response, poor gait adaptability and poor human-computer interaction in complex environments or variable tasks, thereby seriously limiting the assistance effect and wearing comfort of the exoskeletons. Therefore, how to provide a somatic intelligent lower extremity exoskeleton human-computer interaction control method which can improve the assistance effect and wearing comfort of the lower extremity exoskeleton is a problem to be solved by those skilled in the art. SUMMARY

[0003] Therefore, the purpose of the present application is to provide a somatic intelligent lower extremity exoskeleton human-computer interaction control method.

[0004] In order to achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0005] A somatic intelligent lower extremity exoskeleton human-computer interaction control method, comprising the following steps:

[0006] S1: constructing a human-exoskeleton coupling dynamics model;

[0007] S2: jointly solving the human-exoskeleton coupling dynamics model through a kinematics equation to obtain solving data;

[0008] S3: inputting the human-exoskeleton coupling dynamics model, the solving data and an impedance evaluation index into an evaluator to obtain a real-time coupling adaptability score;

[0009] S4: switching a corresponding human-computer interaction control strategy based on the real-time coupling adaptability score;

[0010] S5: minimizing a reward function based on a reinforcement learning algorithm to regulate control parameters of the switched human-computer interaction control strategy.

[0011] Preferably, S1 specifically comprises the following steps:

[0012] S11: constructing a human lower extremity multi-joint dynamics model by using a multi-rigid-body system theory and human lower extremity biomechanical parameters;

[0013] A lower limb exoskeleton dynamics model is constructed based on exoskeleton structure parameters, driver characteristics and link constraints; wherein the lower limb exoskeleton dynamics model integrates a dynamic response model and a joint friction model; the dynamic response model is used to simulate the response lag characteristics of the actuator; and the joint friction model is used to simulate the friction of the joint;

[0014] A nonlinear contact model between the foot bottom and the ground is constructed based on the Hertz contact model; wherein the nonlinear contact model is used to input the contact force data between the foot bottom and the ground collected by the foot bottom pressure sensor in real time;

[0015] S12: coupling the human lower limb multi-joint dynamics model, the lower limb exoskeleton dynamics model and the nonlinear contact model to obtain the human-exoskeleton coupled dynamics model.

[0016] Preferably, the calculation data includes the relative torque of the coupled joint, the transmission power, the interference resistance and the dynamic interaction force.

[0017] Preferably, S3 specifically comprises the following steps:

[0018] S31: constructing a joint impedance model;

[0019] S32: inputting the human-exoskeleton coupled dynamics model, the calculation data and the impedance evaluation index into an evaluator to obtain a real-time coupling adaptability score; wherein the impedance evaluation index includes the equivalent stiffness parameter, the damping parameter, the hysteresis parameter and the coordination degree parameter in the joint impedance model; the impedance evaluation index also includes the electromyographic activation degree collected by experiments, the phase difference and the torque deviation between human-machine motion; wherein the torque deviation is the deviation of the expected joint torque and the actual joint torque.

[0020] Preferably, S4 specifically comprises the following steps:

[0021] Based on the real-time coupling adaptability score, a corresponding cooperation state level is obtained;

[0022] Based on the cooperation state level, the corresponding human-machine interaction control strategy is switched to.

[0023] Preferably, the calculation formula of the reward function is:

[0024] R = αR1 + βR2 + γR3;

[0025] In the formula, R represents the reward function; α, β and γ represent reward weight indexes; R1 represents the electromyographic power index; R2 represents the stability index; and R3 represents the cooperation index.

[0026] Preferably, the calculation formula of the electromyographic power index is:

[0027]

[0028] wherein, represents the root mean square of the electromyogram at the i th moment; i = 1, 2,..., N, and N is a positive integer.

[0029] Preferably, the calculation formula of the stability index is:

[0030]

[0031] wherein, T i represents the time when the heel contacts the ground at the i th moment; T i-1 represents the time when the heel contacts the ground at the (i-1) th moment; ||T i -T i-1 ||represents the Euclidean norm of T i and T i-1 ; i = 1, 2,..., N, and N is a positive integer.

[0032] Preferably, the calculation formula of the synergy index is:

[0033]

[0034] wherein, τ i represents the human-robot interaction torque at the i th moment; τ i-1 represents the human-robot interaction torque at the (i-1) th moment; ||τ i -τ i-1 ||represents the Euclidean norm of τ i and τ i-1 ; i = 1, 2,..., N, and N is a positive integer.

[0035] Preferably, S5 specifically comprises the following steps:

[0036] continuously changing the value of the control parameter to obtain the reward function value of the control parameter at different values;

[0037] the control parameter value corresponding to the minimum reward function value is taken as the final control parameter value.

[0038] According to the above technical solution, compared with the prior art, the present application provides a somatic intelligent lower extremity exoskeleton human-robot interaction control method, which can improve the assistance effect and wearing comfort of the lower extremity exoskeleton. BRIEF DESCRIPTION OF DRAWINGS

[0039] In order to make the technical solutions in the embodiments of the present application or the prior art clearer, the accompanying drawings needed in the embodiments or prior art description will be briefly introduced. Obviously, the accompanying drawings in the following description only aim to explain part of the embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort on the basis of the provided drawings.

[0040] Figure 1 A flow chart of a body-possessed intelligent lower extremity exoskeleton human-computer interaction control method provided by the present application. DETAILED DESCRIPTION

[0041] The technical solutions in the embodiments of the present application will be described clearly and completely in the following with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without any creative effort belong to the protection scope of the present application.

[0042] As shown in Figure 1 The embodiments of the present application disclose a body-possessed intelligent lower extremity exoskeleton human-computer interaction control method, comprising the following steps:

[0043] S1: constructing a human-exoskeleton coupling dynamics model;

[0044] In one or more embodiments, S1 specifically comprises the following steps:

[0045] S11: constructing a human lower extremity multi-joint dynamics model by using multi-rigid-body system theory and human lower extremity biomechanical parameters;

[0046] It can be understood that the human lower extremity biomechanical parameters include human lower extremity height, human lower extremity mass, human lower extremity inertia and lower extremity joint range of motion.

[0047] The multi-joints in the human lower extremity multi-joint dynamics model include hip joint, knee joint and ankle joint.

[0048] In one or more embodiments, S1 further comprises the following steps:

[0049] Parameter identification is performed on the human lower extremity biomechanical parameters by using experimental data (motion capture data, surface electromyography signal, plantar pressure distribution data) to improve the accuracy of the human lower extremity multi-joint dynamics model.

[0050] constructing a lower limb exoskeleton dynamics model based on exoskeleton structure parameters, driver characteristics and link constraints; wherein the lower limb exoskeleton dynamics model integrates a dynamic response model and a joint friction model; the dynamic response model is used to simulate the response lag characteristics of the actuator; the joint friction model is used to simulate the friction of the joint;

[0051] constructing a nonlinear contact model between the foot bottom and the ground based on the Hertz contact model; wherein the nonlinear contact model is used to input the contact force data between the foot bottom and the ground collected by the foot bottom pressure sensor in real time;

[0052] S12: coupling the human lower limb multi-joint dynamics model, the lower limb exoskeleton dynamics model and the nonlinear contact model to obtain the human-exoskeleton coupled dynamics model.

[0053] S2: jointly solving the human-exoskeleton coupled dynamics model through kinematic equations to obtain solving data;

[0054] In one or more embodiments, the solving data includes the relative torque, transmission power, interference resistance and dynamic interaction force of the coupled joint.

[0055] S3: inputting the human-exoskeleton coupled dynamics model, the solving data and the impedance evaluation index into an evaluator to obtain a real-time coupling adaptability score;

[0056] It can be understood that: the evaluator is designed based on fuzzy logic / neural network.

[0057] In one or more embodiments, the real-time coupling adaptability score corresponds to three scores: 1 point, 2 points and 3 points.

[0058] In one or more embodiments, S3 specifically includes the following steps:

[0059] S31: constructing a joint impedance model;

[0060] S32: inputting the human-exoskeleton coupled dynamics model, the solving data and the impedance evaluation index into an evaluator to obtain a real-time coupling adaptability score;

[0061] Wherein, the impedance evaluation index includes equivalent stiffness parameters, damping parameters, hysteresis parameters and coordination parameters in the joint impedance model;

[0062] The impedance evaluation index also includes the degree of electromyographic activation collected by experiments, the phase difference and torque deviation between human-machine motion; wherein the torque deviation is the deviation between the expected joint torque and the actual joint torque.

[0063] In one or more embodiments, the impedance evaluation index further comprises joint angular velocity, mean power of electromyogram, step frequency variation, and posture stability.

[0064] S4: switching a corresponding human-machine interaction control strategy based on the real-time coupling adaptability score;

[0065] In one or more embodiments, S4 specifically comprises the following steps:

[0066] obtaining a corresponding coordination state level based on the real-time coupling adaptability score;

[0067] In one or more embodiments, the coordination state level corresponds to three levels: good, general, and disorder.

[0068] In one or more embodiments, when the real-time coupling adaptability score is 1, the coordination state level is disorder; when the real-time coupling adaptability score is 2, the coordination state level is general; and when the real-time coupling adaptability score is 3, the coordination state level is good.

[0069] switching to a corresponding human-machine interaction control strategy based on the coordination state level.

[0070] In one or more embodiments, the human-machine interaction control strategy comprises an impedance control strategy, a position control strategy, and a force control strategy.

[0071] In one or more embodiments, the three control strategies can be adjusted and implemented by using two modulation parameters (α and β):

[0072] α = β = 0, position control;

[0073] α > 0, 0 < β < 1, impedance control;

[0074] α > 0, β = 1, force control.

[0075] when the coordination state level is disorder, the corresponding human-machine interaction control strategy is the position control strategy;

[0076] when the coordination state level is general, the corresponding human-machine interaction control strategy is the impedance control strategy;

[0077] when the coordination state level is good, the corresponding human-machine interaction control strategy is the force control strategy.

[0078] S5: minimizing a reward function based on a reinforcement learning algorithm to regulate the control parameters of the human-machine interaction control strategy after switching.

[0079] It can be understood that the control parameters refer to controller parameters.

[0080] In one or more embodiments, the calculation formula of the reward function is:

[0081] R = aR1 + bR2 + gR3;

[0082] In the formula, R represents the reward function; a, b, g represent reward weight indexes; R1 represents an electromyogram power index; R2 represents a stability index; and R3 represents a synergy index.

[0083] In one or more embodiments, a calculation formula of the electromyogram power index is:

[0084]

[0085] wherein, represents a root mean square of an electromyogram signal at an i th moment; i = 1, 2,..., N, and N is a positive integer.

[0086] In one or more embodiments, a calculation formula of the stability index is:

[0087]

[0088] wherein, T i represents a time at which a heel contacts a ground at an i th moment; T i-1 represents a time at which a heel contacts a ground at an (i-1) th moment; ||T i -T i-1 ||represents a Euclidean norm of T i and T i-1 ; i = 1, 2,..., N, and N is a positive integer.

[0089] In one or more embodiments, a calculation formula of the synergy index is:

[0090]

[0091] wherein, τ i represents a human-machine interaction torque at an i th moment; τ i-1 represents a human-machine interaction torque at an (i-1) th moment; ||τ i -τ i-1 ||represents a Euclidean norm of τ i and τ i-1 ; i = 1, 2,..., N, and N is a positive integer.

[0092] In one or more embodiments, S5 specifically comprises the following steps:

[0093] continuously changing a value of the control parameter to obtain a reward function value of the control parameter at different values;

[0094] taking a value of the control parameter corresponding to a minimum reward function value as a final value of the control parameter.

[0095] In one or more embodiments, the reinforcement learning algorithm employs a Proximal Policy Optimization Algorithms (PPO) algorithm of policy gradient.

[0096] The various embodiments in the specification are described in progressive manner, each embodiment focuses on the difference from other embodiments, and the same or similar parts between embodiments can be referred to each other. For the device disclosed by the embodiments, since it corresponds to the method disclosed by the embodiments, the description is relatively simple, and the relevant parts can be referred to the method part.

[0097] The above description of disclosed embodiments enables a person skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for human-robot interaction control of embodied intelligence lower extremity exoskeleton, characterized in that, The method comprises the following steps: S1: constructing a human-exoskeleton coupling dynamics model; S2: jointly solving the human-exoskeleton coupling dynamics model through kinematic equations to obtain solving data; S3: inputting the human-exoskeleton coupling dynamics model, the solving data and impedance evaluation indexes into an evaluator to obtain a real-time coupling adaptability score; S4: switching a corresponding human-computer interaction control strategy based on the real-time coupling adaptability score; S5: minimizing a reward function based on a reinforcement learning algorithm to regulate control parameters of the human-computer interaction control strategy after switching.

2. The embodied intelligence lower extremity exoskeleton human-robot interaction control method according to claim 1, characterized in that, S1 specifically comprises the following steps: S11: constructing a human lower limb multi-joint dynamics model by using a multi-rigid-body system theory and human lower limb biomechanical parameters; constructing a lower limb exoskeleton dynamics model based on exoskeleton structure parameters, driver characteristics and connecting rod constraints; wherein the lower limb exoskeleton dynamics model integrates a dynamic response model and a joint friction model; the dynamic response model is used to simulate the response lag characteristics of an actuator; the joint friction model is used to simulate the friction of a joint; constructing a nonlinear contact model between a foot bottom and the ground based on a Hertz contact model; wherein the nonlinear contact model is used to input contact force data between the foot bottom and the ground collected by a foot bottom pressure sensor in real time; S12: coupling the human lower limb multi-joint dynamics model, the lower limb exoskeleton dynamics model and the nonlinear contact model to obtain the human-exoskeleton coupling dynamics model.

3. The embodied intelligence lower extremity exoskeleton human-robot interaction control method according to claim 1, characterized in that, The solving data comprises relative torque, transmission power, interference resistance and dynamic interaction force of a coupling joint.

4. The embodied intelligence lower extremity exoskeleton human-robot interaction control method of claim 1, wherein, S3 specifically comprises the following steps: S31: constructing a joint impedance model; S32: inputting the human-exoskeleton coupling dynamics model, the solving data and impedance evaluation indexes into an evaluator to obtain a real-time coupling adaptability score; wherein the impedance evaluation indexes comprise equivalent stiffness parameters, damping parameters, hysteresis parameters and coordination degree parameters in the joint impedance model; the impedance evaluation indexes further comprise experimentally collected electromyographic activation levels, phase differences and torque deviations between human-computer motions; wherein the torque deviation is a deviation between an expected joint torque and an actual joint torque.

5. The embodied intelligence lower extremity exoskeleton human-robot interaction control method according to claim 1, characterized in that, S4 specifically comprises the following steps: obtaining a corresponding cooperation state level based on the real-time coupling adaptability score; switching to a corresponding human-computer interaction control strategy based on the cooperation state level.

6. The embodied intelligence lower extremity exoskeleton human-robot interaction control method according to claim 1, characterized in that, The calculation formula of the reward function is: R=αR1+βR2+γR3; wherein R represents the reward function; α, β and γ represent reward weight indexes; R1 represents an electromyographic power index; R2 represents a stability index; and R3 represents a cooperation index.

7. The embodied intelligence lower extremity exoskeleton human-robot interaction control method according to claim 6, characterized in that, The calculation formula of the electromyographic power index is: wherein, RMSi represents the root mean square of the electromyography signal at the ith time instant; i = 1, 2,..., N, N being a positive integer.

8. The embodied intelligence lower extremity exoskeleton human-robot interaction control method according to claim 6, characterized in that, The calculation formula of the stability index is: where T i represents the time of the foot heel contacting the ground at the i th moment; T i-1 represents the time of the foot heel contacting the ground at the i-1 th moment; ||T i -T i-1 ||represents the Euclidean norm of T i and T i-1 ; i=1, 2,..., N, N is a positive integer.

9. The embodied intelligence lower extremity exoskeleton human-robot interaction control method according to claim 6, characterized in that, The calculation formula of the cooperation index is: wherein τ i represents the human-machine interaction torque at the i-th moment; τ i-1 represents the human-machine interaction torque at the (i-1)-th moment; ||τ i -τ i-1 represents the Euclidean norm of τ i and τ i-1 ; i = 1, 2,..., N, N being a positive integer.

10. The embodied intelligence lower extremity exoskeleton human-robot interaction control method of claim 1, wherein, S5 specifically comprises the following steps: constantly changing the value of the control parameter to obtain the reward function value of the control parameter at different values; taking the control parameter value corresponding to the minimum reward function value as the final control parameter value.

Images