A real-time detection method and system for the binding state of an exoskeleton robot lower limb

Abstract

The application discloses a real-time detection method and system for the binding state of the lower limbs of an exoskeleton robot. The method comprises collecting the gait of a human body currently in a reversing motion state and extracting features therefrom, determining the current gait phase through the feature extraction; and then detecting the binding state of the lower limbs corresponding to the gait phase in real time. The system comprises a state observer module, a gait phase identification module, a binding detection module, a self-anti-interference controller module and a CMSIS-DSP library module. The application significantly improves the real-time performance and reliability of the binding fall-off identification through the phase-by-phase dynamic detection strategy.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of robot control, and particularly relates to a real-time detection method and system for a lower limb binding state of an exoskeleton robot. BACKGROUND

[0002] An exoskeleton robot is a wearable auxiliary device capable of enhancing or assisting the movement ability of a human limb, and is widely applied to the fields of rehabilitation medicine, industrial carrying, military load bearing, etc. In order to realize effective cooperation with the human body, the exoskeleton robot must be closely bound with the body of a user, and the robot joints are mechanically connected with the human joints through a binding belt, a support or a customized support structure, so as to realize action synchronization and force transmission.

[0003] In actual application, the binding quality directly affects the motion control performance of the exoskeleton and the user comfort. If the binding is too loose, the robot action may be out of synchronization with the human action, and problems such as force-position decoupling, control instability and frictional slip may occur; if the binding is too tight, local compression, blood circulation obstruction or user discomfort may easily occur. In addition, in the dynamic motion process such as walking, running or load operation, the binding state may change over time, such as loosening, mispositioning or shifting, which further reduces the use safety and control accuracy.

[0004] Therefore, real-time monitoring of the binding state becomes one of the key technologies for improving the human-machine cooperation performance of the exoskeleton. Current researches explore the use of pressure sensors, strain gauges, inertial measurement units or visual systems to detect the binding state, and still face challenges such as high cost, complex layout and easy interference. SUMMARY

[0005] The application aims to provide a real-time detection method and system for a lower limb binding state of an exoskeleton robot, and improve the reliability and safety of exoskeleton control.

[0006] In order to achieve the purpose of the application, on the one hand, the application provides a real-time detection method for a lower limb binding state of an exoskeleton robot, comprising the following steps:

[0007] Step 1, the current human gait in the reversing motion state is collected and features are extracted therefrom, and the current gait phase is determined through the feature extraction;

[0008] Step 2, when the current gait phase is determined to be in the standing phase, whether the binding falls off is determined through the mean and variance strategy of the speed error;

[0009] Step 3, when the current gait phase is determined to be in the swing phase, whether the binding falls off is determined according to the kinematic consistency strategy;

[0010] Step 4, the real-time detection of the lower limb binding state is completed through the above judgment.

[0011] In another aspect, the application also provides a real-time detection system for the lower limb binding state of an exoskeleton robot, comprising the following modules:

[0012] A state observer module is used to perform low-pass filtering on the collected raw sensor data, improve the data quality, and generate standardized input required for gait phase recognition;

[0013] A gait phase recognition module is used to extract gait motion features according to joint angle and angular velocity data, determine whether the gait is in a standing phase or a swing phase through joint angle difference and angular velocity difference, and provide prior information for subsequent binding state discrimination;

[0014] A binding detection module is used to execute corresponding falling detection strategies in different gait phases;

[0015] A self-disturbance controller module is used to perform anti-disturbance stability adjustment on the motor joint speed in standing phase binding detection, so as to ensure that gravity compensation can be provided when the binding falls off the leg rod;

[0016] A CMSIS-DSP library module is used to calculate the mean and variance of the joint angular velocity error in standing phase binding detection, provide functions for algorithm acceleration and calculation of mean and variance, and ensure that the high-frequency response requirement of lower limb motion is met.

[0017] Compared with the prior art, the application has the following significant progress: (1) the application adopts different strategies in standing phase and swing phase to realize effective identification of the lower limb binding falling or relaxation state, and improve the reliability and safety of exoskeleton control; (2) the application can adapt to real-time monitoring of the binding state in a dynamic gait environment, and has the advantages of high accuracy, good real-time performance, and independence from additional hardware sensors.

[0018] To more clearly illustrate the functional characteristics and structural parameters of the application, the following further describes the application in conjunction with the accompanying drawings and specific embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0019] The accompanying drawings described herein are used to provide further understanding of the application, and form a part of this application. The illustrative embodiments of the application and their descriptions serve to explain the application, and do not constitute an improper limitation on the application. In the drawings:

[0020] Figure 1 is a step flowchart of the application;

[0021] Figure 2 is a standing phase binding detection schematic diagram of the application. DETAILED DESCRIPTION

[0022] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the 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 of the present application; based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0023] The present application discloses a real-time detection method for the binding state of the lower limbs of an exoskeleton robot. Figure 1 , comprising the following steps:

[0024] Step 1: The current gait in the reversing motion state is collected and the features are extracted, and the current gait phase is determined by the feature extraction;

[0025] Step 2: When the current gait phase is determined to be in the standing phase, whether the binding falls off is determined by the mean and variance strategy of the speed error;

[0026] Step 3: When the current gait phase is determined to be in the swing phase, whether the binding falls off is determined according to the kinematic consistency strategy;

[0027] Step 4: The real-time detection of the binding state of the lower limbs is completed after the above determination.

[0028] The reversing motion state in step 1 includes the angle, angular velocity curve and inertial measurement unit (IMU) linear acceleration change of the leg joint.

[0029] The feature extraction in step 1 is the comparison result of the collected joint angle difference and joint angular velocity difference with the preset threshold value; when the joint angle difference and the joint angular velocity difference are less than or equal to the preset threshold value, it is determined that the current gait phase is in the standing phase; when the joint angle difference is greater than the preset threshold value and the joint angular velocity difference is positive (i.e. the angle difference shows an expanding trend but has not reached the swing peak value), it is determined that the current gait phase is in the swing phase.

[0030] The preset threshold value of the joint angle difference is 0.4 rad, and the preset threshold value of the joint angular velocity difference is 0.5 rad / s.

[0031] In combination with Figure 2 , step 2 specifically comprises the following steps:

[0032] Step 2-1: Set the angular velocity expected value, and detect the actual value of the joint angular velocity of the exoskeleton robot every control period (such as 10 ms);

[0033] Step 2-2, set a sliding time window T (such as 300 ms) to observe the error of the joint angular velocity, and the real-time data collected is first filtered by a state observer, and then the sample value of the speed error is determined by the angular velocity expected value and the actual value of the joint angular velocity:

[0034] ;

[0035] wherein, is the angular velocity expected value, is the actual value of the joint angular velocity, is the sample value of the speed error;

[0036] The sliding time window T defines the time range of observing the speed error, and the real-time data collected in the window generates the sample value of the speed error by calculation, and the window sliding dynamically updates the sample set.

[0037] Step 2-3, when the sample value reaches the upper limit of time set by the sliding time window, the mean value of the speed error is determined by calculating the mean function arm_mean_f32 (e_buf, N, & mean) using the CMSIS-DSP library (a digital signal processing function library based on the Cortex microcontroller software interface standard) The variance of the speed error is determined by calculating the variance function arm_variance_f32 (e_buf, N, & var) , wherein e_buf represents the sample array, N represents the sample data amount, mean represents the calculation of the mean, and var represents the calculation of the variance;

[0038] Step 2-4, the mean value and the variance of the speed error are obtained, and it is judged whether the current mean value and variance meet certain conditions, and whether the binding falls off occurs is judged.

[0039] The conditions met in the step 2-4 are as follows:

[0040] ;

[0041] wherein, is the lower limit of the mean value of the speed error, is the upper limit of the mean value of the speed error, is the threshold value of the variance of the speed error, is the mean value of the speed error of the data in the sliding time window, is the variance of the speed error of the data in the sliding time window;

[0042] When the mean of the speed error is within the interval of ±0.05 rad / s and the variance of the speed error is lower than 0.03 rad2 / s2, it is determined that the current binding is in a risk state of falling or slipping, and an alarm and a disablement protection of the exoskeleton robot (forced power-off of the device) are performed.

[0043] When the mean of the speed error is out of the preset interval or the variance of the speed error is out of the preset threshold, it is determined that the current binding is in a normal state, and the binding state is continuously detected.

[0044] The step 3 specifically comprises the following steps:

[0045] Step 3-1, collecting motion features, including angles and angular velocities when the support leg and the swing leg are switched, recording gait reversing angles of continuous N (such as 5) gait cycles and reversing angular velocities ;

[0046] Step 3-2, based on the gait reversing angle, simultaneously using linear fitting to estimate a theoretical reversing angle of the current gait cycle;

[0047] Step 3-3, according to the reversing angular velocity, estimating a range of joint angular velocity when the theoretical reversing angle is reached through deviation , wherein is a minimum value of the reversing angular velocity of continuous multiple cycles, is a maximum value of the reversing angular velocity of continuous multiple cycles;

[0048] Step 3-4, when the real-time measured current joint angular velocity reaches the theoretical reversing angle, judging whether the joint angular velocity satisfies certain conditions, and judging whether the binding falls off based on the conditions:

[0049] The theoretical reversing angle of the step 3-2 is specifically as follows:

[0050] ;

[0051] wherein, represents a parameter of linear fitting, , represents a gait reversing angle of the most recent N cycles, represents an initial gait reversing angle (usually 0.2 rad).

[0052] The conditions satisfied in the step 3-4 are specifically as follows:

[0053] ;

[0054] wherein, For speed redundancy tolerance, in actual application process, due to the response time difference of different motors, control filter parameter setting is different, or signal processing delay and other factors, even in the normal state of binding, the actual joint angular velocity may exceed the reversing speed range fitted by the former N periods, the parameter is mainly to avoid the misjudgment caused thereby, the value of the tolerance delta is set according to the speed fluctuation data in the system debugging process, filter bandwidth setting and historical maximum deviation, and is usually 0.1-0.3 rad / s; is the actual value of each joint angular velocity;

[0055] If the above formula is satisfied, it is judged that the binding falls off or slips, and alarm and exoskeleton robot failure protection are carried out; otherwise, return to the step 3-1 and continue to detect the binding.

[0056] The real-time detection system of the lower limb binding state of the exoskeleton robot of the application comprises the following modules:

[0057] The state observer module is used for low-pass filtering the collected original sensor data (such as joint angle, angular velocity) and improving the data quality, and generating the standardized input required for gait phase recognition;

[0058] The gait phase recognition module is used for extracting gait motion characteristics according to joint angle and angular velocity data, judging whether the gait is in the standing phase or the swing phase through the joint angle difference and the angular velocity difference, and providing prior information for subsequent binding state discrimination;

[0059] The binding detection module is used for executing corresponding falling detection strategies in different gait phases: such as judging whether there is binding loosening and falling in the standing phase based on the mean value and variance of the joint angular velocity error, and judging whether there is binding loosening and falling in the swing phase according to whether the current reversing angular velocity deviates from the maximum value interval of the angular velocity of the previous several periods;

[0060] The active disturbance rejection controller (ADRC) module is used for anti-disturbance stability regulation of the motor joint speed in the standing phase binding detection, so as to provide gravity compensation when the binding falling leg rod rises, reduce the speed error, and enhance the robustness of the binding detection function; when the exoskeleton robot is in the standing phase, the active disturbance rejection controller (ADRC) module is started, the target joint angular velocity is fed forward compensated and disturbance suppressed, and the ADRC module can estimate the equivalent disturbance (such as gravity term, friction, etc.) according to the real-time system state, and compensate the control input online, so as to improve the accuracy and response performance of the joint speed control;

[0061] The CMSIS-DSP library module is used to calculate the mean and variance of joint angular velocity error in standing bound detection, provides algorithm acceleration and functions to calculate the mean and variance, and ensures that the high-frequency response requirement of lower limb movement is met.

[0062] It should be noted that the relational terms herein such as first and second and the like are used solely to distinguish one entity or action from another, without necessarily requiring or implying any actual relationship or order between such entities or actions. Moreover, the terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0063] Although embodiments of the present application have been shown and described, it is to be understood that various modifications, substitutions, replacements and changes can be made to these embodiments without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A method for real-time detection of a state of a lower extremity of an exoskeleton robot, characterized by, The method comprises the following steps: Step 1, by collecting the current gait in the reversing motion state of the human body and extracting the features thereof, judging the current gait phase through the feature extraction; Step 2, when judging that the current gait phase is in the standing phase, judging whether the binding falls off through the mean and variance strategy of the speed error; Step 2-1, setting the angular velocity expected value, and detecting the actual value of the angular velocity of each joint of the exoskeleton robot every control cycle; Step 2-2, setting a sliding time window T to observe the error of the joint angular velocity, and determining the sample value of the speed error by first filtering the real-time data collected through the state observer and then through the angular velocity expected value and the actual value of the angular velocity of each joint; Step 2-3, when the sample value reaches the upper limit of time set by the sliding time window, determining the mean of the speed error by calculating the mean function and determining the variance of the speed error by calculating the variance function through the CMSIS-DSP library; Step 2-4, after obtaining the mean and variance of the speed error, judging whether the current mean and variance meet certain conditions, and judging whether the binding falls off according to the conditions; The conditions met in step 2-4 are as follows: ； wherein, is a lower limit of the velocity error mean, is an upper limit of the velocity error mean, is a threshold of the velocity error variance, is the velocity error mean of the data within the sliding time window, is the velocity error variance of the data within the sliding time window; When the mean of the speed error is within the interval of ±0.05 rad / s, and the variance of the speed error is lower than 0.03 rad2 / s2, it is determined that the current binding is in a falling or slipping risk state, and an alarm and exoskeleton robot disable protection are performed; When the mean of the speed error exceeds the preset interval or the variance of the speed error exceeds the preset threshold, it is determined that the current binding is in a normal state, and the binding state is continuously detected; Step 3, when judging that the current gait phase is in the swing phase, judging whether the binding falls off according to the kinematic consistency strategy; Step 4, completing the real-time detection of the lower limb binding state through the above judgments. 2.The real-time detection method of the exoskeleton robot lower limb binding state according to claim 1, characterized in that, The reversing motion state in step 1 includes the angle, angular velocity curve, and inertial measurement unit linear acceleration change of the leg joint.

3. The method of claim 2, wherein the method further comprises: determining a state of the exoskeleton robot lower limbs based on the determined state of the exoskeleton robot lower limbs. The feature extraction in step 1 is through the comparison result of the joint angle difference, joint angular velocity difference, and its preset threshold; when the joint angle difference and the joint angular velocity difference are less than or equal to the preset threshold, it is judged that the current gait phase is in the standing phase; when the joint angle difference is greater than the preset threshold, and the joint angular velocity difference is positive, it is judged that the current gait phase is in the swing phase.

4. The method of claim 3, wherein the method further comprises: determining a state of the exoskeleton robot lower limbs based on the determined state of the exoskeleton robot lower limbs. Step 3 specifically comprises the following steps: Step 3-1, collecting motion features, including the angle and angular velocity when the supporting leg and the swinging leg switch, and recording the gait reversing angle and reversing angular velocity of consecutive N gait cycles; Step 3-2, based on the gait reversing angle, simultaneously estimating the theoretical reversing angle of the current gait cycle by linear fitting; Step 3-3, based on the slew rate, estimating a range of joint angular velocity to reach the theoretical commutation angle by a deviation wherein is a minimum value of the slew rate over a plurality of consecutive periods, is a maximum value of the slew rate over a plurality of consecutive periods; Step 3-4, when the real-time measured current joint angular velocity reaches the theoretical reversing angle, judging whether the joint angular velocity meets certain conditions, and judging whether the binding falls off according to the conditions; The conditions met in step 3-4 are as follows: ； wherein is the velocity redundancy tolerance; is the actual value of the joint angular velocity; If the above formula is satisfied, it is judged that the binding falls off or slips, and an alarm and exoskeleton robot disability protection are performed; otherwise, it returns to the step 3-1 to continue the binding detection.

5. The real-time detection method of the exoskeleton robot lower limb binding state according to claim 4, characterized in that, theoretical tangential angle of step 3-2 , as shown in the following formula: ； wherein, denotes a parameter of the linear fit, , denotes the gait reversal angle of the most recent Nth cycle, denotes the initial gait reversal angle.

6. The method of claim 3, wherein the method further comprises: determining a state of the exoskeleton robot lower limbs based on the determined state of the exoskeleton robot lower limbs. The preset threshold of the joint angle difference is 0.4 rad, and the preset threshold of the joint angular velocity difference is 0.5 rad / s.

7. A real-time detection system for exoskeleton robot lower limb binding state, for implementing the method of any one of claims 1-6, characterized in that, The method comprises the following modules: A state observer module is configured to perform low-pass filtering on the collected original sensor data, improve the data quality, and generate standardized input required for gait phase recognition; A gait phase recognition module is configured to extract gait motion features according to joint angle and angular velocity data, judge whether the gait is in a standing phase or a swing phase through joint angle difference and angular velocity difference, and provide prior information for subsequent binding state discrimination; A binding detection module is configured to execute corresponding falling detection strategies in different gait phases; A disturbance-rejection controller module is configured to perform disturbance-rejection stability regulation on the motor joint velocity in the standing phase binding detection, so as to ensure that the gravity compensation can be provided when the binding falls off the leg rod. A CMSIS-DSP library module is configured to calculate the mean and variance of the joint angular velocity error in the standing phase binding detection, provide an algorithm acceleration and a function for calculating the mean and variance, and ensure that the high-frequency response requirement of the lower limb motion is met.

Images