Exoskeleton control methods, storage media, and exoskeletons

Abstract

This disclosure relates to the field of exoskeleton technology, and discloses an exoskeleton control method, storage medium, and exoskeleton. The exoskeleton control method includes: acquiring gait feature values; inputting the gait feature values ​​into a fuzzy neural network model to obtain a target gait output value; determining candidate gait phases and previous gait phases based on the target gait output value, wherein the gait output value of the previous gait phase is temporally adjacent to the target gait output value; determining the target gait phase based on the candidate gait phases and the previous gait phase; and controlling the exoskeleton based on the target gait phase. This embodiment identifies gait phases by comprehensively considering previous gait phases and candidate gait phases, which helps improve the accuracy and reliability of gait phase identification, thereby enabling the exoskeleton to more effectively drive the user's limbs, thus providing human-machine compliance.

Description

Technical Field

[0001] This disclosure relates to the field of exoskeleton technology, specifically to an exoskeleton control method, storage medium, and exoskeleton. Background Technology

[0002] Exoskeletons, as powerful wearable mechanical devices, are increasingly attracting the attention of scholars and researchers both domestically and internationally, becoming a new research hotspot. Exoskeletons require the rapid and accurate prediction of human movement intentions in order to effectively control their movement.

[0003] The relevant technology is based on neural network algorithms. It uses a gait training dataset of human walking to train a neural network model, which then uses the model to identify the user's movement intention from real-time gait data and controls the exoskeleton accordingly. However, gait training datasets are typically limited and do not easily cover the recognition of various gait data from different users. Furthermore, gait data from different users at the same gait phase is usually different, and even for the same user at the same gait phase, gait data can vary. Additionally, the neural network model provided by the technology still suffers from low detection accuracy. This leads to unreliable gait phase recognition by the neural network model, making it prone to errors. Consequently, the exoskeleton may hinder human movement, resulting in a lack of human-machine coordination.

[0004] Public content

[0005] One objective of this disclosure is to provide a control method, storage medium, and exoskeleton for an exoskeleton, aiming to improve the poor reliability of gait phase recognition in related technologies.

[0006] In a first aspect, embodiments of this disclosure provide a method for controlling an exoskeleton, comprising:

[0007] Obtain gait feature values;

[0008] The gait feature values ​​are input into a fuzzy neural network model to obtain the target gait output value;

[0009] Based on the target gait output value, candidate gait phases and previous gait phases are determined, wherein the gait output value of the previous gait phase is temporally adjacent to the target gait output value;

[0010] The target gait phase is determined based on the candidate gait phase and the previous gait phase;

[0011] The exoskeleton is controlled based on the target gait phase.

[0012] Optionally, determining the target gait phase based on the candidate gait phase and the previous gait phase includes:

[0013] Based on a preset standard gait timing sequence, the gait phase arranged after the previous gait phase is determined as the control gait phase;

[0014] Determine whether the candidate gait phase matches the control gait phase;

[0015] If a match is found, the candidate gait phase is determined to be the target gait phase.

[0016] Optionally, if the candidate gait phase does not match the control gait phase, the method further includes:

[0017] Obtain plantar pressure data;

[0018] Based on the plantar pressure data, a reference gait phase is determined;

[0019] The target gait phase is determined based on the reference gait phase and the candidate gait phase.

[0020] Optionally, determining the target gait phase based on the reference gait phase and the candidate gait phase includes:

[0021] Calculate the first weighted label value based on the gait label value corresponding to the reference gait phase and the first weight coefficient;

[0022] The second weighted label value is calculated based on the target gait output value of the candidate gait phase and the second weighting coefficient;

[0023] Add the first weighted label value and the second weighted label value to obtain the weighted result;

[0024] The target gait phase is determined based on the weighted result.

[0025] Optionally, determining the target gait phase based on the weighted result includes:

[0026] The weighted result is integerized to obtain the total label value;

[0027] The gait phase whose gait label value is the total label value is selected as the target gait phase.

[0028] Optionally, each type of gait phase is configured with a gait label value, and determining the candidate gait phase based on the target gait output value includes:

[0029] Based on the target gait output value, generate a target label value;

[0030] Select the gait label value corresponding to the target label value as the candidate label value;

[0031] The gait phase corresponding to the candidate label value is selected as the candidate gait phase.

[0032] Optionally, when the gait label value and the target label value are integers, generating the target label value based on the target gait output value includes:

[0033] Determine whether the numerical type of the target gait output value is an integer;

[0034] If so, the target gait output value is used as the target label value;

[0035] If not, the target gait output value is integerized to obtain the target label value.

[0036] Optionally, determining the previous gait phase based on the target gait output value includes:

[0037] The gait output value that is temporally adjacent to the target gait output value is determined to be the previous gait output value;

[0038] The previous gait phase is determined based on the previous gait output value.

[0039] In a second aspect, embodiments of this disclosure provide a storage medium storing computer-executable instructions for causing an exoskeleton to perform the aforementioned exoskeleton control method.

[0040] In a third aspect, embodiments of this disclosure provide an exoskeleton, comprising:

[0041] At least one processor; and,

[0042] A memory communicatively connected to the at least one processor; wherein,

[0043] The memory stores instructions that can be executed by the at least one processor, which enables the at least one processor to perform the exoskeleton control method described above.

[0044] In the exoskeleton control method provided in this embodiment, gait feature values ​​are acquired, input into a fuzzy neural network model to obtain a target gait output value, candidate gait phases and previous gait phases are determined based on the target gait output value, the gait output value of the previous gait phase is temporally adjacent to the target gait output value, the target gait phase is determined based on the candidate gait phases and previous gait phases, and the exoskeleton is controlled based on the target gait phase. This embodiment identifies gait phase by combining previous gait phases and candidate gait phases, which helps improve the accuracy and reliability of gait phase identification, thereby enabling the exoskeleton to more effectively drive the user's limbs to move, thus providing human-machine compliance. Attached Figure Description

[0045] One or more embodiments are illustrated by way of example with reference numerals in the accompanying drawings. These illustrations do not constitute a limitation on the embodiments. Elements with the same reference numerals in the drawings are denoted as similar elements. Unless otherwise stated, the figures in the drawings are not to be limited by scale.

[0046] Figure 1 This is a schematic diagram of the structure of an exoskeleton provided in an embodiment of the present disclosure;

[0047] Figure 2 for Figure 1 The diagram shows the structure of the signal acquisition shoe.

[0048] Figure 3 This is a schematic diagram showing the position of the IMU sensor on the exoskeleton according to an embodiment of this disclosure;

[0049] Figure 4 A schematic diagram of the circuit structure of an exoskeleton provided in an embodiment of this disclosure;

[0050] Figure 5 A flowchart illustrating an exoskeleton control method provided in an embodiment of this disclosure;

[0051] Figure 6 A schematic diagram of a fuzzy neural network model provided in an embodiment of this disclosure;

[0052] Figure 7 This is a schematic diagram illustrating the use of the fuzzy neural network model provided in this embodiment to identify gait phase.

[0053] Figure 8 Phase timing diagram of gait phase provided in embodiments of this disclosure;

[0054] Figure 9 This is a schematic diagram of the circuit structure of an electronic device provided in an embodiment of the present disclosure. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of this disclosure clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this disclosure. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without inventive effort are within the scope of protection of this disclosure.

[0056] It should be noted that, unless there is a conflict, the various features in the embodiments of this disclosure can be combined with each other, all of which are within the protection scope of this disclosure. Furthermore, although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than the module division in the device or the order in the flowchart. Moreover, the terms "first," "second," and "third" used in this disclosure do not limit the data or execution order, but only distinguish identical or similar items with substantially the same function and effect.

[0057] This disclosure provides an exoskeleton, which can be referred to in conjunction with other embodiments. Figure 1 , Figure 2 and Figure 3 The exoskeleton 100 includes a back frame assembly 200, a waist frame assembly 300, a lower limb assembly 400, a signal acquisition shoe 500, an IMU sensing assembly 600, and a joint drive unit 700.

[0058] The back brace assembly 200 is used to secure the shoulders and back of the human body. The lumbar support assembly 300 is connected to the back brace assembly 200 and is used to secure the lower back of the human body. The lower limb assembly 400 is connected to the lumbar support assembly 300 and is used to secure the thighs and calves of the human body. The signal acquisition shoe 500 is connected to the lower limb assembly 400 and is used for wearing on the feet of the human body.

[0059] Please combine Figure 4 The signal acquisition shoe 500 includes a shoe body 501, multiple pressure sensors 502, and a controller 503. The controller 503 is electrically connected to each pressure sensor 502. Each pressure sensor 502 is positioned at a designated foot location on the shoe body 501, wherein the designated foot location includes the right heel, right toe, right midfoot, left heel, left toe, and left midfoot.

[0060] Multiple pressure sensors 502 include a first pressure sensor 504, a second pressure sensor 505 and a third pressure sensor 506, wherein each pressure sensor is disposed at a designated foot position on each side of the foot.

[0061] A first pressure sensor 504 is located at the toe, a second pressure sensor 505 is located in the middle of the foot, and a third pressure sensor 506 is located in the heel. The first pressure sensor 504 collects the pressure applied by the user's toe to obtain a pressure signal from the toe. The second pressure sensor 505 collects the pressure applied by the user's middle of the foot to obtain a pressure signal from the middle of the foot. The third pressure sensor 506 collects the pressure applied by the user's heel to obtain a pressure signal from the heel.

[0062] The controller 503 is electrically connected to each pressure sensor 502 and is used to process pressure signals.

[0063] The IMU sensing component 600 is used to detect the acceleration and angular velocity of each joint in the exoskeleton 100, and the controller 503 calculates the joint angle based on the acceleration and angular velocity.

[0064] In some embodiments, the IMU sensing component 600 includes five IMU sensors, namely a first IMU sensor 601, a second IMU sensor 602, a third IMU sensor 603, a fourth IMU sensor 604, and a fifth IMU sensor 605, and the controller 503 is electrically connected to the first IMU sensor 601, the second IMU sensor 602, the third IMU sensor 603, the fourth IMU sensor 604, and the fifth IMU sensor 605.

[0065] The first IMU sensor 601 is strapped to the lower back using a strap; the second IMU sensor 602 is strapped to the left thigh using a strap; the third IMU sensor 603 is strapped to the left calf using a strap; the fourth IMU sensor 604 is strapped to the right thigh using a strap; and the fifth IMU sensor 605 is strapped to the right calf using a strap. The IMU sensors provide Euler angles and quaternions. Before strapping the IMUs to the user's thighs and calves, an IMU calibration procedure is required. This calibration procedure requires the user to stand still for 30 seconds before engaging in activity to determine the orientation of the spatial coordinate system. Once the orientation axis is determined, joint angles can be calculated in the sagittal plane.

[0066] The joint drive unit 700 is used to switch motion modes and perform servo control under the control of the controller 503. The motion modes may include position control, speed control, and torque control. The joint drive unit 700 may be a joint module integrating a motor servo system, encoder, driver, and motor, or it may be other power actuators, such as pneumatic or hydraulic transmission systems.

[0067] As another aspect of this disclosure, this disclosure provides a method for controlling an exoskeleton. Please refer to... Figure 5 The control method for exoskeletons includes the following steps:

[0068] S51: Obtain gait feature values.

[0069] In this step, gait feature values ​​are numerical values ​​used to represent the characteristics of gait phase. These gait feature values ​​can be represented by any suitable numerical value, such as Euclidean distance, variance, standard deviation, or absolute mean. Gait phase refers to the specific gait movements within each movement pattern. Different designers can further subdivide the various gait movements within a gait cycle. Movement patterns are derived by classifying basic movements frequently encountered in daily life. This paper discloses eight movement patterns: sitting, standing, walking, running, climbing stairs, descending stairs, ascending incline, and descending incline. Sitting and standing are non-periodic movement patterns, while walking, running, climbing stairs, descending stairs, ascending incline, and descending incline are periodic movement patterns.

[0070] The gait phases in this article include gait phase ① with the right foot forward and both feet supporting the body, gait phase ② with the left foot swinging, gait phase ③ with the left foot forward and both feet supporting the body, and gait phase ④ with the right foot swinging.

[0071] Typically, people sequentially perform gait phases within a gait cycle. Under normal circumstances, the various gait phases within the same gait cycle constitute the standard gait timing sequence. As mentioned earlier, under normal circumstances, people perform actions according to the following standard gait timing sequence, as shown below: First standard gait timing sequence: Gait phase ① → Gait phase ② → Gait phase ③ → Gait phase ④.

[0072] Second standard gait timing sequence: Gait phase ② → Gait phase ③ → Gait phase ④ → Gait phase ①.

[0073] The third standard gait timing sequence is: gait phase ③ → gait phase ④ → gait phase ① → gait phase ②.

[0074] Fourth standard gait timing sequence: Gait phase ④ → Gait phase ① → Gait phase ② → Gait phase ③.

[0075] S52: Input the gait feature values ​​into the fuzzy neural network model to obtain the target gait output value.

[0076] In this step, the fuzzy neural network model is trained in advance by the designer based on the gait training dataset. The training of the fuzzy neural network model includes the following steps: obtaining the gait training dataset, filtering out gait training datasets that do not meet the training conditions, performing equal-length processing on each gait training dataset to obtain equal-length gait training datasets, segmenting the equal-length gait training datasets according to the characteristics of the gait phase to obtain multiple training data subsets, assigning gait label values ​​corresponding to the gait phase to each training data subset, normalizing each training data subset to obtain normalized training data subsets, performing feature processing on the normalized training data subsets to obtain training data features, and training the fuzzy neural network model based on the gait label values ​​of the normalized training data subsets and the training data features, combined with the fuzzy neural network algorithm.

[0077] The gait training dataset consists of gait motion data collected from users within the same gait cycle under specified movement modes. To obtain the gait training dataset, the designers equipped the subjects with signal acquisition shoes and various IMU sensors. Ensuring the proper functioning of all pressure sensors and IMU sensors, the designers instructed the subjects to perform the experiment according to the eight movement modes mentioned above, recording the gait motion data for each mode to obtain the gait training dataset. In the periodic movement mode, the designers recorded multiple sets of gait motion data, each set collected within a complete gait cycle. In the non-periodic movement mode, the designers specified the duration of the gait motion data, for example, 60 seconds. When the subjects performed the movements according to the specified movement modes, they always started and ended with both legs standing. In the experiment, the sampling frequency of the pressure sensors was 100Hz, and the sampling frequency of the IMU sensors was 200Hz. After the experiment, all gait motion data were exported to a CSV file. The format of the CSV file is shown in Table 1.

[0078] Table 1

[0079] t FSR1 FSR2 IMU1 IMU2 IMU3 IMU4 IMU5 t1 t2 t3 t4 t5 … tn

[0080] Filtering gait training datasets that do not meet the training conditions includes: filtering out abnormal gait training datasets and retaining gait training datasets that meet the training conditions.

[0081] In some embodiments, the gait training dataset includes multiple training data points, which are pressure signals collected by a pressure sensor or angular velocities collected by an IMU sensor. Filtering abnormal gait training datasets includes: determining the number of training data points in the gait training dataset, judging whether the number of data points is greater than a preset number threshold, if not, filtering the gait training dataset, if so, retaining the gait training dataset.

[0082] In some embodiments, filtering abnormal gait training datasets includes: calculating the standard deviation of each gait training dataset, determining whether the standard deviation is greater than a preset difference value, and if so, retaining the gait training dataset; otherwise, filtering the gait training dataset.

[0083] In some embodiments, filtering abnormal gait training datasets includes: calculating the standard deviation and mean of each gait training dataset, calculating the ratio of the standard deviation to the mean, determining whether the absolute value of the ratio is greater than a preset value, and if so, retaining the gait training dataset; otherwise, filtering the gait training dataset.

[0084] In some embodiments, retaining the gait training dataset that meets the training conditions includes: removing all training data points within the initial duration of the gait training dataset to obtain the remaining training data points. Since the data from the first 10 seconds of each training session is typically invalid and needs to be removed, this improves the training effect.

[0085] In some embodiments, retaining the gait training dataset that meets the training conditions further includes: filtering the remaining training data points using a low-pass filter to obtain filtered training data points, and using the filtered training data points as the gait training dataset that meets the training conditions. Considering that when subjects perform movements in motion mode, other movements such as turning or stopping may occur midway, and that the human body's movement frequency is usually low, this embodiment can use a low-pass filter to filter high-frequency noise, thereby obtaining a more reliable gait training dataset.

[0086] In some embodiments, equal-length processing of each gait training dataset includes: performing equal-length processing on each gait training dataset according to an interpolation algorithm, so that the number of gait motion data contained in each gait training dataset of the same duration is equal. In this way, this embodiment can set the number of data from sensors with different sampling frequencies to be the same within the same duration, or set the number of data from sensors with the same sampling frequency but with some missing data to be the same, which can facilitate reliable subsequent data processing.

[0087] In some embodiments, before performing equal-length processing on each gait training dataset, in order to reduce the time required to train the model and prevent overfitting, this embodiment performs dimensionality reduction processing on each gait training dataset according to the principal component analysis algorithm. This removes attributes with low correlation to class labels from the input samples, eliminates redundant attributes, and thus reduces computational complexity. For example, the contribution rate is set to 95%.

[0088] This embodiment divides the equal-length gait training dataset into multiple training data subsets based on the characteristics of different gait phases. Each training data subset represents the characteristics of a corresponding gait phase. Each training data subset is labeled with a gait label value corresponding to the gait phase. As shown in Table 2:

[0089] Table 2

[0090] Gait phase Gait label value With the right foot forward, both feet provide support (gait phase ①). 1 Left foot swing (gait phase ②) 2 With the left foot forward, both feet provide support (gait phase 3) 3 Right foot swing (gait phase ④) 4

[0091] As shown in Table 2, the gait label value for "right foot forward with both feet supporting" is 1, the gait label value for "left foot swinging" is 2, the gait label value for "left foot forward with both feet supporting" is 3, and the gait label value for "right foot swinging" is 4.

[0092] This embodiment can use a normalization algorithm to normalize each training data subset in the interval [-1,1] to obtain a normalized training data subset, thus eliminating the data differences from different types of sensors.

[0093] This embodiment performs feature processing on a normalized subset of training data to obtain training data features. In some embodiments, feature processing on the normalized subset of training data includes: windowing the normalized subset of training data to obtain a windowed subset of training data, and calculating the standard deviation, mean, or absolute mean of the windowed subset of training data as training data features.

[0094] This embodiment substitutes gait label values ​​and training data features into a fuzzy neural network model, and uses a fuzzy neural network algorithm to train the fuzzy neural network model. Please refer to [link to relevant documentation]. Figure 6 The fuzzy neural network model includes an input layer 61, a fuzzification layer 62, a fuzzy rule inference layer 63, a defuzzification layer 64, and an output layer 65. The input layer 61 introduces training data features into the fuzzification layer 62. The fuzzification layer 62 uses a membership function to fuzzify each training data feature. The membership function can be a Gaussian membership function or a triangular membership function; this paper chooses the Gaussian membership function, whose expression is: Where c is the center value, σ is the width, and x i Let be the feature of the i-th training data.

[0095] The fuzzy rule inference layer 63 performs logical inference on the fuzzy values ​​to obtain fuzzy output values. The defuzzification layer 64 is used to clarify the fuzzy output values ​​to obtain accurate defuzzified values, which can be used as the target gait output values. The output layer 65 outputs the defuzzified values, which can be integers or decimals. For example, the defuzzified values ​​may be 0.5, 1, 1.2, 1.5, 2, 2.2, 3, 3.6, 4, or 4.3, etc.

[0096] This embodiment optimizes the parameters of each membership function and the fuzzy rule base based on the fuzzy neural network algorithm, thereby improving the performance of the fuzzy inference system and enabling reliable and accurate determination of the target gait phase.

[0097] In this embodiment, gait feature values ​​are substituted from input layer 61 into the fuzzy neural network model, and the fuzzy neural network model outputs the target gait output value. Please refer to [link to relevant documentation]. Figure 7 , Figure 7 The waveforms shown above are the pressure signals collected by the pressure sensors at the right heel, right toe, left heel, and left toe. The waveforms below show the target gait output values ​​(i.e., the output gait label values) from the fuzzy neural network. Figure 7 It can be seen that, with the limited test samples available, the phase recognition accuracy is relatively high.

[0098] Depend on Figure 7 We can also know that: the predicted time T d =t p -t r , among which, T d To predict time, t p To predict the time point of the output for the first time using a fuzzy neural network model, t r The actual measured values ​​of the characteristic points of state transition are given. The average deviation between the predicted and actual values ​​of the phase switching event is 61.5 ms, which meets the prediction time requirements of human-machine coordinated control.

[0099] S53: Based on the target gait output value, determine the candidate gait phase and the previous gait phase. The gait output value of the previous gait phase is adjacent to the target gait output value in time.

[0100] In this step, the candidate gait phase is the gait phase indicated by the target gait output value. Specifically, when the target gait output value is between 0.5 and 1.4, the gait phase indicated by the target gait output value is a double-leg support with the right foot forward. When the target gait output value is between 1.5 and 2.4, the gait phase indicated by the target gait output value is a left-foot swing. When the target gait output value is between 2.5 and 3.4, the gait phase indicated by the target gait output value is a double-leg support with the left foot forward. When the target gait output value is between 3.5 and 4.4, the gait phase indicated by the target gait output value is a right-foot swing.

[0101] The previous gait phase is the gait phase that precedes the candidate gait phase. In some embodiments, determining the previous gait phase based on the target gait output value includes the following steps: determining the gait output value that is temporally adjacent to the target gait output value as the previous gait output value, and determining the previous gait phase based on the previous gait output value.

[0102] The exoskeleton sequentially saves gait output values ​​during each motion intent determination, allowing it to recall previous gait output values ​​to determine the previous gait phase. For example, during the k-th motion intent determination, the exoskeleton saves the k-th gait output value 1.2, and the k-th gait phase is gait phase ①. The (k+1)-th gait output value is 2.3, and the (k+1)-th gait phase is gait phase ②. Specifically, for the (k+1)-th gait phase, the (k+1)-th gait output value 2.3 is the target gait output value, the k-th gait output value 1.2 is the previous gait output value, the (k+1)-th gait phase is the target gait phase, and the k-th gait phase is the previous gait phase.

[0103] As mentioned above, each type of gait phase is configured with a gait label value. In some embodiments, determining candidate gait phases based on the target gait output value includes the following steps:

[0104] S531: Generate target label values ​​based on the target gait output values.

[0105] S533: Select the gait label value corresponding to the target label value as the candidate label value.

[0106] S535: Select the gait phase corresponding to the candidate label value as the candidate gait phase.

[0107] In S531, in some embodiments, when the numerical types of the gait label value and the target label value are decimal types, generating the target label value based on the target gait output value includes: using the target gait output value as the target label value.

[0108] In some embodiments, when the numerical types of the gait label value and the target label value are integers, generating the target label value based on the target gait output value includes: determining whether the numerical type of the target gait output value is an integer; if so, using the target gait output value as the target label value; otherwise, performing an integerization operation on the target gait output value to obtain the target label value.

[0109] In some embodiments, performing an integer operation on the target gait output value to obtain the target label value includes: performing an integer operation on the target gait output value according to the rounding rules to obtain the target label value.

[0110] For example, when the target gait output value is 1.1, the exoskeleton can integerize the target gait output value to obtain the target label value 1. Similarly, when the target gait output value is 0.8, the exoskeleton can integerize the target gait output value to obtain the target label value 1. When the target gait output value is 2, the exoskeleton uses the target gait output value as the target label value 2. When the target gait output value is 2.4, the exoskeleton can integerize the target gait output value to obtain the target label value 2. This embodiment can integerize the target gait output value, which is beneficial for accurately and reliably selecting candidate gait phases.

[0111] In S533 and S535, the exoskeleton searches the tag library for gait tag values ​​corresponding to the target tag value as candidate tag values. For example, if the target tag value is 2, and the gait tag value of gait phase ② is 2 in the tag library as shown in Table 2, then the gait tag value of gait phase ② is a candidate tag value, and gait phase ② is a candidate gait phase.

[0112] S54: Determine the target gait phase based on the candidate gait phase and the previous gait phase.

[0113] In some embodiments, determining the target gait phase based on the candidate gait phase and the previous gait phase in this step includes: determining whether the gait order formed by the previous gait phase and the candidate gait phase matches the local order in the standard gait order; if they match, the candidate gait phase is determined as the target gait phase; if they do not match, the target gait phase is determined.

[0114] For example, when the candidate gait phase is gait phase ② and the previous gait phase is gait phase ①, the gait sequence formed by "gait phase ① → gait phase ②" matches the local sequence formed by "gait phase ① → gait phase ②" in the first standard gait sequence. Therefore, the exoskeleton determines gait phase ② as the target gait phase.

[0115] When the candidate gait phase is gait phase ②, the previous gait phase is gait phase ④. The gait sequence formed by "gait phase ④ → gait phase ②" does not match any local sequence from the first standard gait sequence to the fourth standard gait sequence. The exoskeleton needs to continue to determine the target gait phase.

[0116] In some embodiments, continuing to determine the target gait phase includes using the previous gait phase as the target gait phase.

[0117] In some embodiments, further determining the target gait phase includes: acquiring plantar pressure data, wherein the plantar pressure data is a pressure signal collected from the user's foot applied to the exoskeleton; determining a reference gait phase based on the plantar pressure data; and determining a target gait phase based on the reference gait phase and candidate gait phases.

[0118] In some embodiments, determining the target gait phase based on the candidate gait phase and the previous gait phase includes: determining the gait phase arranged after the previous gait phase as the reference gait phase according to a preset standard gait timing sequence; determining whether the candidate gait phase matches the reference gait phase; and if they match, determining the candidate gait phase as the target gait phase.

[0119] For example, if the candidate gait phase is gait phase ② and the previous gait phase is gait phase ①, then according to the standard gait timing, the control gait phase is gait phase ②. Since the candidate gait phase matches the control gait phase, gait phase ② is determined as the target gait phase.

[0120] When the candidate gait phase is gait phase ②, the previous gait phase is gait phase ④, and according to the standard gait timing, the control gait phase is gait phase ①. Since the candidate gait phase does not match the control gait phase, the exoskeleton needs to continue determining the target gait phase.

[0121] The exoskeleton can correct misjudgments caused by the bias of the fuzzy neural network model itself by judging whether the gait order composed of the previous gait phase and the candidate gait phase matches the local order in the standard gait order, or by judging whether the candidate gait phase matches the control gait phase. This can also avoid misjudgments caused by external interference factors interfering with the original gait data and causing the fuzzy neural network model to make mistakes, which is beneficial to the reliability of gait phase identification.

[0122] S55: Controls the exoskeleton based on the target gait phase.

[0123] In this step, controlling the exoskeleton based on the target gait phase includes: determining the gait trajectory corresponding to the target gait phase, and controlling each joint to perform actions according to the gait trajectory. For example, when the target gait phase is gait phase ①, the exoskeleton determines the gait trajectory corresponding to gait phase ① and controls each joint to move according to the gait trajectory corresponding to gait phase ①. When the target gait phase is gait phase ②, the exoskeleton determines the gait trajectory corresponding to gait phase ② and controls each joint to move according to the gait trajectory corresponding to gait phase ②, and so on.

[0124] This embodiment identifies gait phase by integrating previous gait phases and candidate gait phases, which helps improve the accuracy and reliability of gait phase identification. This, in turn, allows the exoskeleton to more effectively drive the user's limbs to move, thereby providing human-machine compliance.

[0125] In some embodiments, if the candidate gait phase does not match the control gait phase, the exoskeleton control method further includes the following steps:

[0126] S56: Obtain plantar pressure data.

[0127] S57: Determine the reference gait phase based on plantar pressure data.

[0128] S58: Determine the target gait phase based on the reference gait phase and candidate gait phases.

[0129] In S56, plantar pressure data includes pressure signals from the right heel and right toe, as well as pressure signals from the left heel and left toe. When a user wears the exoskeleton for walking, their feet are fitted with signal-collecting shoes. The user's feet apply pressure to the soles of these shoes, and pressure sensors at designated foot locations collect the pressure exerted by the user's feet on the soles, thus obtaining pressure signals. However, using plantar pressure data alone to identify gait phase has limitations; it is not easy to accurately identify gait phase, but it can assist in gait phase correction. Please refer to [link to relevant documentation]. Figure 8 :

[0130] When the target gait phase is a two-footed support with the right foot forward (gait phase ①), the pressure signal collected by the pressure sensor of the right heel and right toe changes from low level to high level, while the pressure signal collected by the pressure sensor of the left heel and left toe remains at high level.

[0131] When the target gait phase is the left foot swing (gait phase ②), the pressure signal collected by the pressure sensor of the right heel and right toe remains at a high level, while the pressure signal collected by the pressure sensor of the left heel and left toe changes from a high level to a low level.

[0132] When the target gait phase is a double-footed support with the left foot forward (gait phase ③), the pressure signals collected by the pressure sensors of the right heel and right toe remain at a high level, while the pressure signals collected by the pressure sensors of the left heel and left toe change from a low level to a high level.

[0133] When the target gait phase is the right foot swing (gait phase ④), the pressure signal collected by the pressure sensor of the right heel and right toe changes from high level to low level, while the pressure signal collected by the pressure sensor of the left heel and left toe remains at high level.

[0134] In S57, determining the reference gait phase based on plantar pressure data includes the following steps: determining the first level change of the pressure signal of the right heel and right toe and the second level change of the pressure signal of the left heel and left toe, and determining the reference gait phase based on the first level change and the second level change.

[0135] When the first level change is from low to high, and the second level change remains high, the exoskeleton determines the right foot forward as the reference gait phase for the two-foot support.

[0136] When the first level change remains high and the second level change changes from high to low, the exoskeleton determines the left foot swing as the reference gait phase.

[0137] When the first level change remains high and the second level change changes from low to high, the exoskeleton determines the left foot forward as the reference gait phase for the two-foot support.

[0138] When the first level change changes from high to low, and the second level change remains high, the exoskeleton determines the right foot swing as the reference gait phase.

[0139] In S58, when the candidate gait phase does not match the control gait phase, this embodiment can comprehensively refer to the gait phase and the candidate gait phase to determine the target gait phase, so as to reliably re-identify the movement intention so as to smoothly control the interaction between the exoskeleton and the human body.

[0140] In some embodiments, determining the target gait phase based on the reference gait phase and the candidate gait phase includes the following steps:

[0141] S581: Calculate the first weighted label value based on the gait label value corresponding to the reference gait phase and the first weight coefficient.

[0142] S582: Calculate the second weighted label value based on the target gait output value of the candidate gait phase and the second weighting coefficient.

[0143] S583: Add the first weighted label value and the second weighted label value to obtain the weighted result.

[0144] S584: Determine the target gait phase based on the weighted results.

[0145] In S581, the exoskeleton calculates the first weighted label value according to the following formula: tag1=χ*tac, where tag1 is the first weighted label value, χ is the first weight coefficient, and tac is the gait label value corresponding to the reference gait phase.

[0146] In S582, the exoskeleton calculates the second weighted label value according to the following formula: tag2 = δ * tad, where tag2 is the second weighted label value, δ is the second weight coefficient, and tad is the target gait output value of the candidate gait phase. In some embodiments, the sum of the first weight coefficient and the second weight coefficient is the natural number 1, where the first weight coefficient is greater than the second weight coefficient, for example, the first weight coefficient is 0.8 and the second weight coefficient is 0.2.

[0147] In S583, the exoskeleton calculates the weighted result according to the following formula, as shown below: tag ad = tag1 + tag2, where tag ad This is the weighted result.

[0148] In S584, in some embodiments, when the numerical types of the gait label value and the target label value are decimal types, determining the target gait phase based on the weighted result includes: selecting the gait phase whose gait label value is the weighted result as the target gait phase.

[0149] In some embodiments, when the gait label value and the target label value are integers, determining the target gait phase based on the weighted result includes: integerizing the weighted result to obtain a total label value, and selecting the gait phase whose gait label value is the total label value as the target gait phase. This embodiment uses a reference gait phase obtained from plantar pressure data to correct the candidate gait phase, which helps to reliably determine the target gait phase, thereby ensuring that the exoskeleton can interact with the human body reliably, effectively, and with high compliance.

[0150] In some embodiments, obtaining gait feature values ​​includes: acquiring joint angle data, which is collected by various IMU sensors located at different joints of the exoskeleton; filtering out abnormal joint angle data to obtain filtered joint angle data; performing equal-length processing on each filtered joint angle data to obtain equal-length joint angle data; performing normalization processing on each equal-length joint angle data to obtain normalized joint angle data; and performing feature processing on the normalized joint angle data to obtain gait feature values.

[0151] In some embodiments, the exoskeleton can also be trained to use a fuzzy neural network model for recognizing motion patterns. For example, following the method described above, motion label values ​​can be assigned to eight motion patterns, as shown in Table 3.

[0152] Table 3

[0153] Sports Mode Sports tag value Meditation mode 1 Standing mode 2 Walking mode 3 Running mode 4 Staircase mode 5 Downstairs mode 6 uphill 7 downhill 8

[0154] The exoskeleton uses gait data and motion label values, combined with a fuzzy neural network algorithm, to train and generate a fuzzy neural network model.

[0155] It should be noted that in the above embodiments, there is no necessarily a certain order between the steps. Those skilled in the art can understand from the description of the embodiments of this disclosure that the steps may have different execution orders in different embodiments, that is, they may be executed in parallel or in interchange, etc.

[0156] Please see Figure 9 , Figure 9 This is a schematic diagram of the circuit structure of an electronic device provided in an embodiment of this disclosure, wherein the electronic device can be an exoskeleton or an electronic product. Figure 9 As shown, the electronic device 900 includes one or more processors 91 and a memory 92. Wherein, Figure 9 Taking a processor 91 as an example, the processor 91 and the memory 92 can be connected via a bus or other means. Figure 9 Taking the example of a connection between China and Israel via a bus.

[0157] The memory 92, as a storage medium, can be used to store non-volatile software programs, non-volatile computer-executable programs, and modules, such as the program instructions / modules corresponding to the exoskeleton control method in the embodiments of this disclosure. The processor 91 implements the functions of the exoskeleton control method provided in the above method embodiments by running the non-volatile software programs, instructions, and modules stored in the memory 92.

[0158] Memory 92 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, memory 92 may optionally include memory remotely located relative to processor 91, which can be connected to processor 91 via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0159] The program instructions / modules are stored in the memory 92 and, when executed by one or more processors 91, execute the exoskeleton control method in any of the above method embodiments.

[0160] This disclosure also provides a storage medium storing computer-executable instructions that are executed by one or more processors, for example... Figure 9 One of the processors 91 can enable the above one or more processors to execute the exoskeleton control method in any of the above method embodiments.

[0161] This disclosure also provides a computer program product, the computer program product including a computer program stored on a non-volatile computer-readable storage medium, the computer program including program instructions, which, when executed by an electronic device, cause the electronic device to perform any of the exoskeleton control methods described herein.

[0162] The device or equipment embodiments described above are merely illustrative. The unit modules described as separate components may or may not be physically separate. The components shown as module units may or may not be physical units; that is, they may be located in one place or distributed across multiple network module units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0163] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented using software plus a general-purpose hardware platform, or of course, using hardware. Based on this understanding, the above technical solutions, in essence or the parts that contribute to the related technology, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0164] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this disclosure, and not to limit them; under the concept of this disclosure, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of different aspects of this disclosure as described above, which are not provided in detail for the sake of brevity; although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features; and these 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 disclosure.

Claims

1. A method for controlling an exoskeleton, characterized in that, include: Gait feature values ​​are obtained, wherein the gait feature values ​​are numerical values ​​used to represent the characteristics of gait phase; The gait feature values ​​are input into a fuzzy neural network model to obtain the target gait output value; Based on the target gait output value, candidate gait phases and previous gait phases are determined, wherein the gait output value of the previous gait phase is temporally adjacent to the target gait output value; Each type of gait phase is configured with a gait label value. The step of determining candidate gait phases based on the target gait output value includes: generating a target label value based on the target gait output value; selecting a gait label value corresponding to the target label value as a candidate label value; and selecting a gait phase corresponding to the candidate label value as a candidate gait phase. Determining a target gait phase based on the candidate gait phase and the previous gait phase includes: determining a gait phase following the previous gait phase as a control gait phase according to a preset standard gait timing sequence; determining whether the candidate gait phase matches the control gait phase; if they match, determining the candidate gait phase as the target gait phase. The exoskeleton is controlled based on the target gait phase.

2. The method according to claim 1, characterized in that, If the candidate gait phase does not match the control gait phase, the method further includes: Obtain plantar pressure data; Based on the plantar pressure data, a reference gait phase is determined; The target gait phase is determined based on the reference gait phase and the candidate gait phase.

3. The method according to claim 2, characterized in that, The step of determining the target gait phase based on the reference gait phase and the candidate gait phase includes: Calculate the first weighted label value based on the gait label value corresponding to the reference gait phase and the first weight coefficient; The second weighted label value is calculated based on the target gait output value of the candidate gait phase and the second weighting coefficient; Add the first weighted label value and the second weighted label value to obtain the weighted result; The target gait phase is determined based on the weighted result.

4. The method according to claim 3, characterized in that Determining the target gait phase based on the weighted result includes: The weighted result is integerized to obtain the total label value; The gait phase whose gait label value is the total label value is selected as the target gait phase.

5. The method according to claim 1, characterized in that, When the gait label value and the target label value are integers, generating the target label value based on the target gait output value includes: Determine whether the numerical type of the target gait output value is an integer; If so, the target gait output value is used as the target label value; If not, the target gait output value is integerized to obtain the target label value.

6. The method according to any one of claims 1 to 4, characterized in that, Determining the previous gait phase based on the target gait output value includes: The gait output value that is temporally adjacent to the target gait output value is determined as the previous gait output value, and the previous gait phase is determined based on the previous gait output value.

7. A storage medium, characterized in that, The exoskeleton is stored with computer-executable instructions for causing it to perform the control method of the exoskeleton as described in any one of claims 1 to 6.

8. An exoskeleton, characterized in that, include: At least one processor; as well as, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the exoskeleton control method as described in any one of claims 1 to 6.

Images