Fmg-based exoskeleton optimization control method, system, terminal and medium

Abstract

This application relates to the field of exoskeleton assistive technology, and discloses a human-in-the-loop exoskeleton optimization control method, system, terminal, and medium based on force-sensitive electromyography (FMG). The method includes: acquiring the user's force-sensitive electromyography (EMG) data and inertial measurement data; obtaining an FMG load index based on the inertial measurement data and the force-sensitive EMG data, and constructing a total cost function based on the FMG load index; defining a current optimization parameter vector, generating an assist torque curve based on the current optimization parameter vector, and outputting a target torque according to the assist torque curve; acquiring the user's physiological and kinematic data during walking after applying the target torque, and obtaining the target assist parameter configuration based on the total cost function and the physiological and kinematic data. This application can achieve targeted optimization of specific muscle exertion by exoskeleton assistance, improving optimization efficiency.

Description

Technical Field

[0001] This application relates to the field of exoskeleton assistive technology, and in particular to human-in-the-loop exoskeleton optimization control methods, systems, terminals and media based on FMG. Background Technology

[0002] Lower limb assistive exoskeleton robots reduce the burden on the human musculoskeletal system and improve walking endurance and work efficiency by outputting assistive torques at joints such as the hip and knee. This is achieved during tasks like weight-bearing walking, prolonged standing, and climbing. Existing lower limb exoskeletons widely employ open-loop assistive trajectory control based on gait phase. Designers pre-determine a joint assistive torque curve that varies with gait phase based on biomechanical analysis. In actual use, assistance is triggered only based on detected gait events or gait phases, without closed-loop adjustments based on the wearer's muscle state. This type of open-loop control is ill-suited to adapting to differences in body size, muscle strength, and fatigue levels among users, and it struggles to respond promptly to changes in muscle load for the same user under different tasks (such as changes in speed, slope, and load). This often results in deviations between the actual assistive effect and the design goals.

[0003] To overcome the lack of feedback between open-loop control and the human body, the "Human-in-the-loop optimization (HIL)" control framework has emerged in recent years. This framework uses exoskeleton assistance parameters as optimization variables and automatically searches for the assistance scheme that minimizes the wearer's metabolic power through iterative experiments. Existing research has shown that using a respiratory metabolic analyzer to measure total body energy expenditure and using metabolic cost as the objective function of HIL can significantly reduce the wearer's metabolic power during walking. However, this type of method relies on bulky and expensive metabolic measurement equipment, requires lengthy warm-up and calibration before experiments, and typically requires several minutes of stable walking for each assistance condition to obtain reliable metabolic estimates. This results in a time-consuming and complex overall optimization process, making it difficult to promote in practical applications.

[0004] Therefore, existing technologies still need to be improved and developed. Summary of the Invention

[0005] The main purpose of this application is to provide a human-in-the-loop exoskeleton optimization control method, system, terminal and medium based on FMG, which aims to solve the problem that the existing optimization control framework using human in-the-loop is not very practical due to its reliance on large and expensive metabolic measurement equipment, long optimization process and complicated experiments.

[0006] The first aspect of this application provides a human-in-the-loop (FMG)-based human-in-the-loop exoskeleton optimization control method, which includes the following steps: Acquire the user's force-sensitive electromyographic data and inertial measurement data; Based on the inertial measurement data and the force-sensitive electromyography data, the FMG load index is obtained, and the total cost function is constructed based on the FMG load index; Define the current optimization parameter vector, generate the assist torque curve based on the current optimization parameter vector, and output the target torque according to the assist torque curve; Acquire physiological and kinematic data of the user during walking after the target torque is applied, and obtain the target assist parameter configuration based on the total cost function and the physiological and kinematic data.

[0007] Optionally, in one embodiment of this application, the force-sensitive electromyography data is an FMG signal, and the inertial measurement data is raw IMU data; The acquisition of the user's force-sensitive electromyographic data and inertial measurement data specifically includes: Acquire raw FMG pressure signals and static baseline values ​​of the user measured by the FMG muscle expansion detection module, and raw IMU data of the user measured by the gait detection module; The static baseline value is removed from the original FMG pressure signal to obtain the FMG variation components; The FMG variation components are low-pass filtered to obtain the FMG signal.

[0008] Optionally, in one embodiment of this application, obtaining the FMG load index based on the inertial measurement data and the force-sensitive electromyography data specifically includes: The gait cycle of the FMG signal is identified based on the raw IMU data, each gait cycle is normalized to a corresponding gait phase, and the FMG signal is mapped onto the gait phase to obtain a gait alignment signal. A key phase window is determined, and the root mean square value of the gait alignment signal is calculated within each key phase window. The FMG load index is obtained based on the multiple root mean square values.

[0009] Optionally, in one embodiment of this application, constructing the total cost function based on the FMG load index specifically includes: The real-time output torque of the exoskeleton joint is obtained, and the root mean square value of the assist torque and the torque change rate are calculated based on the real-time output torque. Obtain the user's current gait trajectory and natural gait trajectory, and calculate the root mean square deviation based on the current gait trajectory and the natural gait trajectory; The total cost function is obtained by weighting and fusing the FMG load index, the root mean square value of the assist torque, the torque change rate, and the root mean square deviation.

[0010] Optionally, in one embodiment of this application, generating the assist torque curve based on the current optimized parameter vector specifically includes: Define a piecewise cosine function; Based on the buckling stage parameters in the current optimized parameter vector, generate buckling stage assist components; Based on the stretching phase parameters in the current optimized parameter vector, generate the stretching phase assist component; An assist torque curve is generated based on the piecewise cosine function, the buckling stage assist component, and the extension stage assist component.

[0011] Optionally, in one embodiment of this application, the physiological and kinematic data includes the target FMG signal and the target gait trajectory; The process of obtaining the target assistance parameter configuration based on the total cost function and the physiological and kinematic data specifically includes: Based on the total cost function, the total cost score of the target is calculated according to the target FMG signal and the target gait trajectory; Based on the total target cost score, the current optimization parameter vector, and the defined parameter feasible region, the updated optimization parameter vector is obtained; The updated total cost score and the updated number of iterations are obtained based on the updated optimization parameter vector; If the difference between the updated total cost score and the target total cost score meets the set requirements, or if the updated number of iterations reaches the preset number of iterations, then the updated optimization parameter vector will be configured as the target assist parameter.

[0012] Optionally, in one embodiment of this application, the total cost function is expressed as: ; in, Score the total cost. For load indicators, To help determine the root mean square value of the torque, The rate of change of torque, This is the root mean square deviation. , and These are the penalty term weighting coefficients for the root mean square value of the assist torque, the rate of change of torque, and the root mean square deviation, respectively.

[0013] A second aspect of this application also provides a human-in-the-loop (FMG)-based human-in-the-loop exoskeleton optimization control system, wherein the FMG-based human-in-the-loop exoskeleton optimization control system is used to implement the FMG-based human-in-the-loop exoskeleton optimization control method described in any of the above solutions; the FMG-based human-in-the-loop exoskeleton optimization control system includes: The signal preprocessing module is used to acquire the user's force-sensitive electromyographic data and inertial measurement data; The cost function construction module is used to obtain the FMG load index based on the inertial measurement data and the force-sensitive electromyography data, and to construct the total cost function based on the FMG load index. The auxiliary torque generation module is used to define the current optimization parameter vector, generate an auxiliary torque curve based on the current optimization parameter vector, and output the target torque according to the auxiliary torque curve. The assist parameter determination module is used to acquire the user's physiological and kinematic data during the walking process after the target torque is applied, and to obtain the target assist parameter configuration based on the total cost function and the physiological and kinematic data.

[0014] A third aspect of this application also provides a terminal, wherein the terminal includes: a memory, a processor, and an FMG-based human-in-the-loop exoskeleton optimization control program stored in the memory and executable on the processor, wherein when the FMG-based human-in-the-loop exoskeleton optimization control program is executed by the processor, it implements the steps of the FMG-based human-in-the-loop exoskeleton optimization control method as described above.

[0015] A fourth aspect of this application also provides a computer-readable storage medium, wherein the computer-readable storage medium stores an FMG-based human-in-the-loop exoskeleton optimization control program, which, when executed by a processor, implements the steps of the FMG-based human-in-the-loop exoskeleton optimization control method as described above.

[0016] Beneficial effects: This application provides a human-in-the-loop exoskeleton optimization control method, system, terminal and medium based on FMG. This application eliminates the need for expensive and bulky metabolic instruments or complex EMG acquisition systems. By directly monitoring the mechanical activity of target muscles (such as the rectus femoris) and constructing a cost function directly related to its load, it can achieve targeted optimization of exoskeleton-assisted force exertion on specific muscles, improve optimization efficiency and enhance the comfort of the wearer while walking. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1This is a flowchart of a preferred embodiment of the human-in-the-loop exoskeleton optimization control method based on FMG in this application; Figure 2 This is a schematic diagram illustrating the working principle of the human-in-the-loop exoskeleton optimization control method based on FMG in this application; Figure 3 For this application Figure 2 A diagram illustrating the degree of muscle expansion in the middle. Figure 4 This is a schematic diagram illustrating the application of the human-in-the-loop exoskeleton optimization control system based on FMG in this application; Figure 5 This is a structural diagram of a preferred embodiment of the human-in-the-loop exoskeleton optimization control system based on FMG of this application; Figure 6 This is a structural diagram of a preferred embodiment of the terminal of this application.

[0019] Explanation of reference numerals in the attached figures: 100. Signal preprocessing module; 200. Cost function construction module; 300. Auxiliary torque generation module; 400. Auxiliary parameter determination module. Detailed Implementation

[0020] To make the objectives, technical solutions, and effects of this application clearer and more explicit, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. The described embodiments are only possible technical implementations of this application and not all possible implementations. Based on the embodiments in this application, those skilled in the art can obtain other embodiments without creative effort, and these embodiments are also within the protection scope of this application.

[0021] First, let's introduce the terms used in the embodiments of this application: FMG, Force Myography; HIL, Human-in-the-loop, is an optimization of the concept of "human in the loop". EMG, Electromyography, surface electromyography; IMU, Inertial Measurement Unit; RPTF, Resistive Polymer Thick Film, is a technology that uses resistive polymer thick film. LPF, Low-Pass Filter; RMS, Root Mean Square; Jerk, Jerk (derivative of acceleration), jerk (the derivative of acceleration, referring to the rate of change of torque).

[0022] In related technologies, besides metabolic indicators, some studies have also attempted to use surface electromyography (EMG) signals as feedback for exoskeleton control or HIL optimization, indirectly characterizing the reduction in muscle load by decreasing the EMG amplitude of specific muscles (such as the rectus femoris). While EMG methods can reflect muscle electrical activity and are related to muscle activation levels, they suffer from problems in long-term wearable applications and real-world environments, including signal susceptibility to sweat, skin resistance, body hair, and electromagnetic interference; sensitivity to electrode placement; and complex wire arrangement. The signal processing chain is also relatively complex, requiring high sampling frequencies and filtering.

[0023] Currently, muscle activity measurement methods based on force myometry (FMG) have been developed. FMG measures muscle belly expansion and circumferential pressure changes caused by muscle contraction through pressure or force sensors placed on the limb surface, thereby obtaining signals related to muscle mechanical activity. Compared to EMG, FMG signals are low-frequency, unipolar mechanical quantities. The sensors do not need to be attached to bare skin and can be directly integrated into exoskeleton straps, cuffs, and other structures, resulting in lower hardware costs, stronger resistance to electromagnetic interference, and suitability for long-term, daily wear. FMG has been validated in upper limb prosthetic hand control, gesture recognition, and upper limb support exoskeletons. Some studies have combined FMG with joint kinematics to estimate lower limb joint torques, providing muscle-level information input for exoskeleton control. However, existing FMG research mainly focuses on gesture recognition, upper limb assistance, and joint torque estimation, and has not yet been deeply applied to the HIL (High-Intensity Linkage) framework of exoskeletons.

[0024] Existing open-loop assist control cannot adaptively adjust assist based on the wearer's real-time muscle load, making it difficult to achieve targeted load reduction for key muscles such as the rectus femoris. Current HIL optimization based on metabolism or EMG requires expensive or complex measurement equipment, resulting in long experimental cycles and high costs, making it unsuitable for daily use or product deployment. While existing FMG technology can be used for muscle activity detection, a lower limb exoskeleton HIL optimization framework with FMG signals as the core feedback quantity has not yet been established.

[0025] The human-in-the-loop (HIL) optimization control method based on FMG in this application constructs an objective function related to the mechanical load of the lower limb by placing FMG sensors in key muscles (such as the rectus femoris) and introducing a multi-objective and safety-constrained control strategy. This method achieves rapid, safe, and individualized closed-loop optimization of muscle load without relying on metabolic devices, overcoming the limitations of existing open-loop control and traditional HIL methods.

[0026] The architecture of the embodiments of this application is described below.

[0027] The exoskeleton includes: a lower limb assistive exoskeleton body for outputting assistive torque at the knee (or hip) joint; an FMG muscle expansion detection module located in the thigh muscles (such as the rectus femoris muscle belly) for measuring local pressure or force changes caused by muscle contraction; a gait detection module, an inertial measurement unit (IMU) located in the thigh (or lower leg or heel) for identifying gait events and gait phase; and a control and optimization unit, including a real-time controller and a human-in-the-loop optimization module, for calculating muscle load indices based on FMG signals and gait phase, and optimizing exoskeleton assist parameters based on these indices.

[0028] The FMG muscle expansion detection module (taking the rectus femoris muscle as an example) employs a ring-shaped flexible band structure, embedding several force-sensitive resistor sensor units based on RPTF (Resistive Polymer Thick Film) technology. The band is secured to the rectus femoris muscle belly using Velcro or buckles, with appropriate pre-tension applied so that the sensors are in their linear operating range when the muscle relaxes. When the rectus femoris contracts, the muscle belly expands, leading to increased local pressure and thus a change in sensor resistance. The controller samples the sensor output voltage and converts it to resistance to obtain the raw FMG signal.

[0029] The technical solutions of this application will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0030] The preferred embodiment of this application describes a human-in-the-loop exoskeleton optimization control method based on FMG, such as... Figure 1 As shown, the human-in-the-loop exoskeleton optimization control method based on FMG includes the following steps: In step S10, the user's force-sensitive electromyographic data and inertial measurement data are acquired.

[0031] In one possible implementation, the force-sensitive electromyography (FMG) data is an FMG signal, and the inertial measurement data is raw IMU data. Step S10 specifically includes: acquiring the raw FMG pressure signal and static baseline value of the user measured by the FMG muscle expansion detection module, and the raw IMU data of the user measured by the gait detection module; removing the static baseline value from the raw FMG pressure signal to obtain the FMG variation component; and performing low-pass filtering on the FMG variation component to obtain the FMG signal.

[0032] In this embodiment, the control and optimization unit preprocesses the FMG signal, and the steps are as follows: Step S11: Remove static baseline: The subject is placed in an upright, static position and raw FMG pressure signals are continuously acquired. 30s. Perform a moving average (window can be 1s) or equivalent low-frequency filtering on this data segment to obtain the baseline B(t); define the session baseline using the time average of this segment. .

[0033] Step S12, Extraction of Changed Components: During the experiment, the real-time pressure signal was processed according to... ; Decompose it. Defined as a change component related to rectus femoris contraction.

[0034] Step S13, Low-pass filtering: To suppress high-frequency noise, the filter is... Apply a zero-phase Butterworth low-pass filter with a cutoff frequency of 10Hz; the filtered signal is denoted as... . Muscle expansion information within the 0–10 Hz range is retained as input signals for subsequent calculations and optimizations.

[0035] in, The original FMG pressure signal (varying over time). This is the static baseline value. The FMG changes after baseline removal. This is the filtered FMG signal.

[0036] Specifically, see Figure 2 and Figure 3 , Figure 3 Muscle expansion was represented as the flat-ground-gait interaction force (N=286, Avg T=1.04s). The input was a raw, unprocessed pressure signal. It's understandable that even if the thigh is completely relaxed and the muscles are not engaged, the FMG sensor will still output a non-zero pressure value. This is because the strap itself has a pretension (the force applied to secure the strap), and there is also the weight of the leg itself and the static pressure of the soft tissue. This ever-present force is called the static baseline. By stripping away the constant pressure and retaining the dynamically changing pressure, the FMG variation component is obtained by subtracting this background noise from the total pressure measured in real time. .

[0037] The signal at this time (FMG change components) Although the baseline has been removed, it still contains high-frequency noise from the impact vibrations during walking (vibration when the foot lands), friction between the straps and clothing, and the sensor's own electronic noise. It's understandable that muscle contraction is a mechanical movement, with its frequency primarily concentrated in the low-frequency range of 0-10Hz; signals above 10Hz are essentially noise. This application uses a low-pass filter (LPF) with a cutoff frequency of 10Hz to allow only the true signal from 0-10Hz to pass, filtering out noise above 10Hz to obtain a clean FMG signal. This clean FMG signal is then used as input for calculating muscle load indicators and subsequently fused with data from the gait detection module. The purpose of this preprocessing is to transform the raw, mixed physical pressure signal into a clean digital signal that is only related to active muscle contraction.

[0038] In step S20, the FMG load index is obtained based on the inertial measurement data and the force-sensitive electromyography data, and the total cost function is constructed based on the FMG load index.

[0039] In one possible implementation, the step of obtaining the load index in step S20 specifically includes: identifying the gait cycle of the FMG signal based on the raw IMU data, normalizing each gait cycle to a corresponding gait phase, and mapping the FMG signal onto the gait phase to obtain a gait alignment signal; determining a key phase window, calculating the root mean square value of the gait alignment signal within each key phase window, and obtaining the FMG load index based on multiple root mean square values.

[0040] In one possible implementation, the step of constructing the total cost function in step S20 specifically includes: acquiring the real-time output torque of the exoskeleton joint, and calculating the root mean square value of the assist torque and the torque change rate based on the real-time output torque; acquiring the user's current gait trajectory and natural gait trajectory, and calculating the root mean square deviation based on the current gait trajectory and the natural gait trajectory; and performing weighted fusion based on the FMG load index, the root mean square value of the assist torque, the torque change rate, and the root mean square deviation to obtain the total cost function.

[0041] In this embodiment, during the construction of the rectus femoris load index and cost function, the controller first uses the gait detection module to identify gait cycles and normalizes each gait cycle to a 0–100% gait phase. For the gait intervals where the rectus femoris load is most concentrated (e.g., the mid-to-late support phase and the early swing phase), within each gait cycle, based on the pre-processed signal... Calculate the FMG load index of the rectus femoris muscle Based on this, a total cost function is constructed. The details are as follows: Step S21, Statistical calculation of phase window: Calculate the root mean square value or average value of FMG after processing within the phase window, which reflects the expansion intensity of the rectus femoris muscle in that phase; In the target phase window Above, calculation ; If multiple sensitive phase windows exist, they are weighted according to a preset weight. Perform a weighted sum: ; in, Normalized gait phase (0-100%). For the target gait phase window, For signal The root mean square value.

[0042] Step S22, Multi-objective weighted total cost: To constrain the assist intensity and gait morphology, the load index is further weighted... Root mean square value of joint assist torque Torque change rate and kinematic deviations from natural gait Introducing a penalty term forms a multi-objective weighted total cost: The total cost function is expressed as: ; in, Score the total cost. For load indicators (core load reduction targets). To assist the root mean square value of the torque (constraint strength). This represents the rate of change of torque (constraint smoothness). This is the root mean square deviation (i.e., kinematic deviation, which constrains the naturalness of gait). , and These are the penalty term weighting coefficients for the root mean square value of the assist torque, the rate of change of torque, and the root mean square deviation, respectively.

[0043] The aforementioned cost function ensures targeted optimization of the rectus femoris load while avoiding surface load reduction by increasing the assist torque or twisting the gait.

[0044] Specifically, in the gait alignment and segmentation process, preprocessed FMG signals and raw IMU data are input for event recognition. The IMU data is used to calculate the heel strike and toe lift-off times. Then, the gait is segmented, defining one heel strike to the next as a complete gait cycle. Finally, normalization is performed, stretching or compressing the time axis of this cycle to a standard 0–100%, denoted as . In this way, regardless of whether a person walks fast or slow, the system has a unified comparison benchmark. This allows the continuous physiological signals to be segmented into independent steps according to walking rhythm, outputting each gait cycle. The corresponding gait alignment signal.

[0045] Understandably, attitude calculation algorithms (such as complementary filtering and Kalman filtering) fuse accelerometer and gyroscope data to calculate the real-time tilt angle and angular velocity of the thigh or lower leg in space. Gait event detection identifies heel strike and toe lift based on specific patterns of tilt angle or angular velocity (such as zero crossing, peak, and threshold). Period division and normalization define one gait cycle from one heel strike to the next, and normalize the time axis of this cycle to 0-100%. Signal resampling and mapping... Each data point in the data is mapped to a corresponding gait cycle (gait phase) based on the time it occurs. The gait alignment signal is obtained from the above.

[0046] Then, multiple sensitive phase windows are defined. Based on biomechanical knowledge, the system anticipates that the rectus femoris muscle primarily exerts force in two regions: the mid-to-late support phase (controlling knee flexion and stabilizing the body) and the early swing phase (hip flexion and leg lift). Within these two windows, the root mean square (RMS) value of the signal is calculated based on the gait alignment signal. A higher RMS value indicates higher force intensity, meaning more intense muscle contraction and greater fatigue. A weighted summation is then performed; if both windows are important, they are assigned different weights, and the summation yields the final load index. .

[0047] Next, the root mean square value of the assist torque is calculated based on the real-time output torque of the exoskeleton joints to prevent the optimization algorithm from penalizing excessive assist intensity by increasing it without limit. The torque change rate (jerk) is calculated based on the real-time output torque of the exoskeleton joints to prevent drastic changes in the assist waveform and penalize uneven assist, ensuring wearing comfort. The deviation between the current gait and the natural gait is calculated based on the current joint angle trajectory (current gait trajectory) and the pre-stored natural gait trajectory without exoskeleton, preventing gait distortion and penalizing unnatural gait, ensuring natural walking. Finally, the core objective and all penalty terms are combined according to weights to form a single total cost function, and the final output total cost score serves as the basis for the optimization algorithm to solve for the optimal assist parameters.

[0048] In step S30, a current optimization parameter vector is defined, a boost torque curve is generated based on the current optimization parameter vector, and a target torque is output according to the boost torque curve.

[0049] In one possible implementation, step S30 specifically includes: defining a piecewise cosine function; generating a buckling stage assist component based on the buckling stage parameters in the current optimization parameter vector; generating an extension stage assist component based on the extension stage parameters in the current optimization parameter vector; and generating an assist torque curve based on the piecewise cosine function, the buckling stage assist component, and the extension stage assist component.

[0050] In this embodiment of the application, the human-in-the-loop optimization process based on FMG includes: Step S31, Optimization Variable Definition: Establish an optimization parameter vector that includes two stages: buckling (Flex) and extension (Ext). ; The variables include normalized gait phase. The start-up time, peak time, and turn-off time, and the corresponding peak torque. Among them, To optimize the parameter vector, This marks the beginning of the buckling phase. This is the peak moment of the buckling phase. This refers to the moment of closure during the buckling phase. This represents the peak moment during the buckling stage. This marks the start of the stretching phase. This is the peak moment of the extension phase. This is the moment when the extension phase is shut down. This represents the peak torque during the extension phase. This is the transpose symbol.

[0051] Step S32, Parameter Constraint Space: Define the feasible region of parameters. Strict timing constraints must be met: ; Simultaneously satisfy the minimum duration constraint: , ; And the safety constraints on the torque amplitude: ,in, .

[0052] Among them, subscript Indicates the buckling stage, subscript Indicates the extension phase; This is the minimum duration of action during the buckling phase. This represents the peak torque.

[0053] Step S33, Generating the Assisted Curve: Define a piecewise cosine form of the unit shape function. ,in, For unit shape function, To enable phase, For peak phase, For the off phase. Construct the bimodal half-cosine total assist torque: .in, For the total assist torque, To provide a boosting torque during the buckling phase, Provides a boosting torque during the extension phase.

[0054] Among them, the buckling component: ; Stretch component: .

[0055] Specifically, a feasible region for the parameters is defined to ensure that the search is only conducted within a reasonable range. A standardized, smooth convex shape is defined as the basic unit for constructing the assist (i.e., the basic waveform unit, piecewise cosine function). The parameters of the buckling stage are substituted into the basic waveform unit to generate a positive assist convexity; the parameters of the extension stage are substituted into the basic waveform unit to generate a negative assist convexity; the assist components of the buckling and extension stages are superimposed on the time axis to form a complete torque curve that may have two peaks; the synthesized curve is stored as a lookup table or function, which is used by the real-time controller to look up the table and output the torque according to the current gait phase, ultimately outputting a smooth bimodal assist torque curve.

[0056] In step S40, physiological and kinematic data of the user during the walking process after the target torque is applied are obtained, and the target assist parameter configuration is obtained based on the total cost function and the physiological and kinematic data.

[0057] In one possible implementation, the physiological and kinematic data includes the target FMG signal and the target gait trajectory. Step S40 specifically includes: calculating a target total cost score based on the total cost function, according to the target FMG signal and the target gait trajectory; obtaining an updated optimized parameter vector based on the target total cost score, the current optimized parameter vector, and the defined parameter feasible region; obtaining an updated total cost score and an updated iteration count based on the updated optimized parameter vector; if the difference between the updated total cost score and the target total cost score meets a set requirement, or the updated iteration count reaches a preset iteration count, then the updated optimized parameter vector is configured as the target assist parameter.

[0058] In this embodiment of the application, the FMG signal is first preprocessed: the mean FMG value of the subject when standing still is used as the basis for the preprocessing. Using this as a reference, calculate the baseline removal signal. The preprocessed rectus femoris activation signal was obtained through a 10Hz low-pass filter (LPF). .

[0059] Then, the objective function is constructed: the rectus femoris load index is calculated. The weighted root mean square value within two time windows, namely buckling and extension, is used. ; And the weights satisfy and .

[0060] Then perform total cost and optimization solution: Introduce a penalty term including the effective value of torque. Torque smoothness and relative to natural gait reference kinematic deviation Construct the total cost function By solving To obtain the optimal assist parameter configuration.

[0061] Specifically, see Figure 4 This allows the wearer to walk while the exoskeleton provides current assistance, and collects pre-processed signals. Based on the joint angles and the collected data, a comprehensive score is calculated under the current assist parameters, and a weighted total cost score is obtained. Based on the score of this experiment, the optimization algorithm determines the parameters for the next set of trials. It searches within the parameter feasible region to predict which region's parameters might yield a lower total cost score, generating new optimized parameters. It then determines whether the optimal parameters have been reached (the score no longer decreases) or the maximum number of trials has been reached. If convergence is achieved, the optimal parameters are output; otherwise, it returns to continue iterating, ultimately outputting the optimal assist parameters for real-time exoskeleton control.

[0062] The present invention has the following beneficial effects: This invention enables closed-loop optimization control based on specific muscle loads. By acquiring muscle expansion information through FMG sensors deployed in the lower limb muscles, constructing a muscle load index, and using it as the core cost function for human-loop optimization, the invention achieves a directional reduction in muscle force exerted by the exoskeleton, directly corresponding to improved walking endurance and comfort.

[0063] It can reduce costs and shorten optimization time. The FMG sensor is low-cost and simple in structure, and can be directly integrated into the exoskeleton strap, eliminating the need for expensive and bulky metabolic equipment or demanding EMG acquisition systems; the FMG signal has a fast response and does not require a long waiting time for stabilization, which significantly shortens the evaluation time for each assist condition. The overall HIL optimization process time is shortened from tens of minutes in traditional metabolic HIL to an acceptable short time, making it easy to use in laboratory settings and everyday environments.

[0064] It can balance the smoothness of the assistive function with the naturalness of the gait. This invention addresses the objective function... A torque accelerometer (Jerk) penalty term was introduced. Penalty for kinematic deviation Compared to optimization methods that only aim to reduce muscle activity, this invention can effectively suppress the generation of high-frequency jitter or abrupt torque and limit the deviation of the optimized gait trajectory from the natural reference trajectory. Thus, while reducing muscle load, it ensures the smoothness of human-computer interaction and avoids gait stiffness or safety hazards caused by excessive assistance.

[0065] In summary, this invention enables rapid, safe, and individualized closed-loop optimization control of lower limb muscle load (such as key muscles like the rectus femoris) while ensuring a simple system structure and controllable hardware costs. It overcomes the shortcomings of existing open-loop control and traditional HIL methods, and has clear engineering application value and prospects for promotion.

[0066] Next, referring to the accompanying drawings, a human-in-the-loop exoskeleton optimization control system based on FMG according to an embodiment of this application is described, which is used to implement the human-in-the-loop exoskeleton optimization control method based on FMG as described in any of the above schemes.

[0067] Figure 5 This is a structural diagram of the human-in-the-loop exoskeleton optimization control system based on FMG, according to an embodiment of this application.

[0068] like Figure 5 As shown, the FMG-based human-in-the-loop exoskeleton optimization control system includes: a signal preprocessing module 100, a cost function construction module 200, an auxiliary torque generation module 300, and an assist parameter determination module 400.

[0069] Specifically, the signal preprocessing module 100 is used to acquire the user's force-sensitive electromyographic data and inertial measurement data; The cost function construction module 200 is used to obtain the FMG load index based on the inertial measurement data and the force-sensitive electromyography data, and to construct the total cost function based on the FMG load index. The auxiliary torque generation module 300 is used to define the current optimization parameter vector, generate an auxiliary torque curve according to the current optimization parameter vector, and output the target torque according to the auxiliary torque curve. The assist parameter determination module 400 is used to acquire the physiological and kinematic data of the user during the walking process after the target torque is applied, and to obtain the target assist parameter configuration based on the total cost function and the physiological and kinematic data.

[0070] Figure 6 A structural diagram of a terminal provided in an embodiment of this application. The terminal may include: The memory 501, the processor 502, and the computer program stored on the memory 501 and capable of running on the processor 502.

[0071] When the processor 502 executes the program, it implements the human-in-the-loop exoskeleton optimization control method based on FMG provided in the above embodiments.

[0072] Furthermore, the terminal also includes: Communication interface 503 is used for communication between memory 501 and processor 502.

[0073] The memory 501 is used to store computer programs that can run on the processor 502.

[0074] Memory 501 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0075] If the memory 501, processor 502, and communication interface 503 are implemented independently, then the communication interface 503, memory 501, and processor 502 can be interconnected via a bus to complete communication between them. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EIS) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, Figure 6 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0076] Optionally, in a specific implementation, if the memory 501, processor 502, and communication interface 503 are integrated on a single chip, then the memory 501, processor 502, and communication interface 503 can communicate with each other through an internal interface.

[0077] Processor 502 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application.

[0078] This embodiment also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described human-in-the-loop exoskeleton optimization control method based on FMG.

[0079] One embodiment of this application provides a computer program product, including a computer program that, when executed by a processor, implements the features described in this application. Figure 1 The corresponding embodiments provide an FMG-based human-in-the-loop exoskeleton optimization control method.

[0080] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0081] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0082] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0083] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable storage medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable storage medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable storage medium could be paper or other suitable media on which the program can be printed, since the program can be obtained electronically by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0084] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0085] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0086] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0087] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

[0088] It should be understood that the application of this application is not limited to the examples above. Those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

[0089] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A human-in-the-loop exoskeleton optimization control method based on FMG, characterized in that, The FMG-based human-in-the-loop exoskeleton optimization control method includes: Acquire the user's force-sensitive electromyographic data and inertial measurement data; Based on the inertial measurement data and the force-sensitive electromyography data, the FMG load index is obtained, and the total cost function is constructed based on the FMG load index; Define the current optimization parameter vector, generate the assist torque curve based on the current optimization parameter vector, and output the target torque according to the assist torque curve; Acquire physiological and kinematic data of the user during walking after the target torque is applied, and obtain the target assist parameter configuration based on the total cost function and the physiological and kinematic data.

2. The human-in-the-loop exoskeleton optimization control method based on FMG according to claim 1, characterized in that, The force-sensitive electromyography data is an FMG signal, and the inertial measurement data is raw IMU data; The acquisition of the user's force-sensitive electromyographic data and inertial measurement data specifically includes: Acquire raw FMG pressure signals and static baseline values ​​of the user measured by the FMG muscle expansion detection module, and raw IMU data of the user measured by the gait detection module; The static baseline value is removed from the original FMG pressure signal to obtain the FMG variation components; The FMG variation components are low-pass filtered to obtain the FMG signal.

3. The human-in-the-loop exoskeleton optimization control method based on FMG according to claim 2, characterized in that, The step of obtaining the FMG load index based on the inertial measurement data and the force-sensitive electromyography data specifically includes: The gait cycle of the FMG signal is identified based on the raw IMU data, each gait cycle is normalized to a corresponding gait phase, and the FMG signal is mapped onto the gait phase to obtain a gait alignment signal. A key phase window is determined, and the root mean square value of the gait alignment signal is calculated within each key phase window. The FMG load index is obtained based on the multiple root mean square values.

4. The human-in-the-loop exoskeleton optimization control method based on FMG according to claim 3, characterized in that, The construction of the total cost function based on the FMG load index specifically includes: The real-time output torque of the exoskeleton joint is obtained, and the root mean square value of the assist torque and the torque change rate are calculated based on the real-time output torque. Obtain the user's current gait trajectory and natural gait trajectory, and calculate the root mean square deviation based on the current gait trajectory and the natural gait trajectory; The total cost function is obtained by weighting and fusing the FMG load index, the root mean square value of the assist torque, the torque change rate, and the root mean square deviation.

5. The human-in-the-loop exoskeleton optimization control method based on FMG according to claim 4, characterized in that, The step of generating the assist torque curve based on the current optimized parameter vector specifically includes: Define a piecewise cosine function; Based on the buckling stage parameters in the current optimized parameter vector, generate buckling stage assist components; Based on the stretching phase parameters in the current optimized parameter vector, generate the stretching phase assist component; An assist torque curve is generated based on the piecewise cosine function, the buckling stage assist component, and the extension stage assist component.

6. The human-in-the-loop exoskeleton optimization control method based on FMG according to claim 4, characterized in that, The physiological and kinematic data include the target FMG signal and the target gait trajectory; The process of obtaining the target assistance parameter configuration based on the total cost function and the physiological and kinematic data specifically includes: Based on the total cost function, the total cost score of the target is calculated according to the target FMG signal and the target gait trajectory; Based on the total target cost score, the current optimization parameter vector, and the defined parameter feasible region, the updated optimization parameter vector is obtained; The updated total cost score and the updated number of iterations are obtained based on the updated optimization parameter vector; If the difference between the updated total cost score and the target total cost score meets the set requirements, or if the updated number of iterations reaches the preset number of iterations, then the updated optimization parameter vector will be configured as the target assist parameter.

7. The human-in-the-loop exoskeleton optimization control method based on FMG according to claim 6, characterized in that, The total cost function is expressed as: ； in, Score the total cost. For load indicators, To help determine the root mean square value of the torque, The rate of change of torque, This is the root mean square deviation. , and These are the penalty term weighting coefficients for the root mean square value of the assist torque, the rate of change of torque, and the root mean square deviation, respectively.

8. A human-in-the-loop exoskeleton optimization control system based on FMG, characterized in that, The FMG-based human-in-the-loop exoskeleton optimization control system is used to implement the FMG-based human-in-the-loop exoskeleton optimization control method according to any one of claims 1-7, wherein the FMG-based human-in-the-loop exoskeleton optimization control system comprises: The signal preprocessing module is used to acquire the user's force-sensitive electromyographic data and inertial measurement data; The cost function construction module is used to obtain the FMG load index based on the inertial measurement data and the force-sensitive electromyography data, and to construct the total cost function based on the FMG load index. The auxiliary torque generation module is used to define the current optimization parameter vector, generate an auxiliary torque curve based on the current optimization parameter vector, and output the target torque according to the auxiliary torque curve. The assist parameter determination module is used to acquire the user's physiological and kinematic data during the walking process after the target torque is applied, and to obtain the target assist parameter configuration based on the total cost function and the physiological and kinematic data.

9. A terminal, characterized in that, The terminal includes: a memory, a processor, and an FMG-based human-in-the-loop exoskeleton optimization control program stored in the memory and executable on the processor. When the FMG-based human-in-the-loop exoskeleton optimization control program is executed by the processor, it implements the steps of the FMG-based human-in-the-loop exoskeleton optimization control method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores an FMG-based human-in-the-loop exoskeleton optimization control program, which, when executed by a processor, implements the steps of the FMG-based human-in-the-loop exoskeleton optimization control method as described in any one of claims 1-7.

Images