Prosthetic control method, system, and apparatus

Abstract

This application belongs to the field of prosthetic limb control, specifically disclosing a prosthetic limb control method, system, and device. The method includes: acquiring real-time motion data of multiple points on the prosthesis and the user's torso; inputting the real-time motion data of multiple points into a first neural network to predict the motion data of each point in different preset motion modes at future times; using the predicted motion data of each point as a reference sequence and the real-time motion data of each point as a control sequence to determine the similarity between the reference sequence and the control sequence for each point; combining the similarity of each point in each preset motion mode to determine the probability of being in each preset motion mode at future times; inputting the real-time motion data of multiple points into a second neural network to obtain prosthetic limb control parameters corresponding to different preset motion modes; and weighted summing of the prosthetic limb control parameters in different preset motion modes to obtain the final control parameters of the prosthesis. This application achieves smooth control of the prosthesis.

Description

Technical Field

[0001] This application belongs to the field of prosthetic control, and more specifically, relates to a prosthetic control method, system and device. Background Technology

[0002] Traditional lower limb prostheses, taking passive prostheses worn by those with knee amputations as an example, directly connect to the amputee's thigh stump, resulting in the loss of rotational freedom of the knee and ankle joints. Therefore, when walking, climbing stairs, or navigating slopes, asymmetrical muscle movements are required to compensate for the reduced flexibility of the knee and ankle joints. However, this can lead to muscle strain and increase the risk of injury over time. Therefore, powered thigh prostheses, which mimic the movement of the amputee's thigh and allow for autonomous rotation of the knee and ankle joints, offer significant advantages in reducing asymmetrical compensatory movements in amputees.

[0003] For powered thigh prostheses, the primary condition and key to assisting amputees in movement lies in accurately identifying the amputee's movement objectives and then precisely matching the movement of the prosthesis's knee and ankle joints with the movement of the amputee's residual thigh limb. Good human-machine gait cycle matching helps amputees increase their trust in the prosthesis, improves the user experience, and further promotes the widespread adoption of powered thigh prostheses. Common methods for gait cycle progression matching in existing thigh prostheses involve constructing a function from the movement angle and cycle progression of the residual thigh limb, and then mapping the gait cycle progression based on the thigh angle, its integral, or a more complex functional relationship. Other methods include prosthesis-customized movement speeds that allow users to control the prosthesis's landing time and thus the joint angle movement.

[0004] Existing methods for determining the movement pattern of prostheses based on external environmental factors and amputee movement information mainly include the following: A common method is to determine the movement pattern based on the swing speed of the amputee's thigh and whether the prosthesis is on the ground, using a finite state machine to determine the state based on the collected sensor information; another method is to install a depth camera on the thigh socket of the prosthesis to acquire the terrain conditions in the direction of travel and change the prosthesis's movement state. These methods, which determine the movement state by defining specific nodes of terrain and amputee movement information, usually have the following problems: they typically require the amputee to perform specific actions to switch movement states (slopes, stairs, obstacles), and once a movement state is entered, the single control parameter cannot well adapt to the user's own movement habits. In summary, while existing control methods can meet the movement needs of amputees, they cannot fully realize the expression of the amputee's movement intentions, allowing the user to completely control the progress and speed of the prosthesis's movement. Therefore, further exploration of new methods for amputee movement intention recognition is needed. Summary of the Invention

[0005] In view of the shortcomings of the prior art, the purpose of this application is to provide a prosthetic limb control method, system and device, which aims to solve the problem that the existing prosthetic limb control methods only consider a single movement state or terrain state and cannot adapt well to the user's movement habits.

[0006] To achieve the above objectives, in a first aspect, this application provides a prosthesis control method, wherein the prosthesis is positioned at the user's amputated thigh to assist the user in walking, and the control method includes:

[0007] Acquire real-time motion data at multiple points on the prosthesis and the user's torso;

[0008] The real-time motion data of the multiple points are input into the first neural network to predict the motion data of each point under different preset motion modes at future times.

[0009] The motion data of each predicted point is used as a reference sequence, and the real-time motion data of each point is used as a control sequence. The similarity between the reference sequence and the control sequence corresponding to each point under each preset motion mode is determined. By combining the similarity of each point under each preset motion mode, the probability of a user carrying a prosthesis being in each preset motion mode at a future time is determined.

[0010] The real-time motion data of the multiple points are input into the second neural network to obtain the prosthesis control parameters corresponding to different preset motion modes;

[0011] By combining the probability of being in each preset movement mode, the control parameters of the prosthesis under different preset movement modes are weighted and summed to obtain the final control parameters of the prosthesis.

[0012] In one possible implementation, real-time motion data is acquired at multiple points on the prosthesis and the user's torso, including:

[0013] Real-time motion data of multiple first-type points are acquired by using multiple sensors installed on the prosthesis;

[0014] By combining real-time motion data of multiple first-type points on the prosthesis with inverse motion equations, real-time motion data of multiple second-type points on the user's torso are obtained.

[0015] The sensors installed on the prosthesis may include simple sensors such as an inertial measurement unit (IMU) and a six-axis force sensor.

[0016] In one possible implementation, the plurality of first-type points include: amputated thigh, prosthetic lower leg, prosthetic ankle joint, and prosthetic foot;

[0017] The multiple second-category points include: the center point of the chest and abdomen, the center point of the pelvis, the healthy thigh, and the healthy calf.

[0018] In one possible implementation, the first neural network is trained using real-time motion data samples from multiple points and motion data samples under different preset motion modes at corresponding future times.

[0019] The second neural network is trained using real-time motion data samples from multiple points and corresponding prosthetic control parameter samples under different preset motion modes.

[0020] The different preset movement modes correspond to different walking environments for users carrying prostheses; the walking environment includes at least one of the following: flat ground, slopes of different gradients, stairs of different heights and / or lengths, and obstacles of different heights and / or lengths.

[0021] In one possible implementation, the similarity between the reference sequence and the control sequence for each point under each preset motion mode is obtained through the following steps:

[0022] Obtain real-time updated control sequences based on real-time motion data of the locations;

[0023] A heuristic iterative algorithm is used to transform the control sequence updated at a single time point to obtain the optimal transformation parameters that minimize the dynamic time warping (DTW) distance between the transformed control sequence and the reference sequence; the transformation includes scaling and / or translation.

[0024] The single-time similarity between the updated control sequence and the reference sequence at each time point is determined by combining the optimal transformation parameters.

[0025] The similarity scores at each moment within a preset time period are weighted and summed to obtain the similarity scores between the reference sequence and the control sequence within the preset time period. These scores are then used as the similarity scores between the reference sequence and the control sequence for the corresponding points. Among these scores, the weight of the single-moment similarity scores at later moments within the preset time period is greater than that at earlier moments.

[0026] In one possible implementation, the single-time similarity is:

[0027]

[0028]

[0029] in, Represents the similarity at time n; To and Numerical functions, for or ; This represents the preset calculation coefficients, and t represents the sampling period of the real-time motion data. This indicates the sampling rate of real-time motion data; This represents the maximum value in the reference sequence within the time period t. and These represent the horizontal scaling parameter and the vertical scaling parameter, respectively. and These represent the horizontal translation parameter and the vertical translation parameter, respectively.

[0030] The similarity between the reference sequence and the control sequence within the preset time period is:

[0031] in, express Similarity between the reference sequence and the control sequence within a given time period.

[0032] In one possible implementation, the probability of being in each preset motion mode is obtained through the following steps:

[0033] In a single preset motion pattern, the similarity of the reference sequence and the control sequence at each point is weighted and summed to obtain the total similarity of the preset motion pattern; among them, the points farther away from the user's amputated thigh have lower weight values ​​when weighting and summing.

[0034] Normalize the total similarity of each preset motion pattern;

[0035] By combining the total similarity normalization, the probability of each preset motion mode at the current moment is updated to obtain the probability of each preset motion mode at the future moment.

[0036] In one possible implementation, the probability of each preset motion mode at the future moment... for:

[0037]

[0038] in, This indicates the probability of each preset motion mode at the current moment. Indicate the probability update weight, This represents the normalized overall similarity of each preset motion pattern.

[0039] In one possible implementation, the final control parameters of the prosthesis for:

[0040]

[0041] in, This represents real-time motion data from multiple points. n Indicates the total number of preset sports modes. Indicates the future time. i The possibility of preset motion modes Indicates the first i Prosthetic control parameters for a preset movement pattern.

[0042] Secondly, this application provides a prosthetic limb control system, wherein the prosthesis is positioned at the user's amputated thigh to assist the user in walking, and the control system includes:

[0043] The motion data acquisition module is used to acquire real-time motion data from multiple points on the prosthesis and the user's torso;

[0044] The motion data prediction module is used to input the real-time motion data of the multiple points into the first neural network to predict the motion data of each point under different preset motion modes at future times.

[0045] The motion pattern determination module is used to take the predicted motion data of each point as a reference sequence and the real-time motion data of each point as a control sequence to determine the similarity between the reference sequence and the control sequence for each point under each preset motion pattern; and combine the similarity of each point under each preset motion pattern to determine the probability that a user carrying a prosthesis will be in each preset motion pattern at a future time.

[0046] The control parameter acquisition module is used to input the real-time motion data of the multiple points into the second neural network to obtain the prosthesis control parameters corresponding to different preset motion modes; and to weight and sum the prosthesis control parameters under different preset motion modes based on the probability of being in each preset motion mode to obtain the final control parameters of the prosthesis.

[0047] Thirdly, this application provides an electronic device, comprising: at least one memory for storing a program; and at least one processor for executing the program stored in the memory, wherein when the program stored in the memory is executed, the processor is configured to execute the method described in the first aspect or any possible implementation thereof.

[0048] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when run on a processor, causes the processor to perform the method described in the first aspect or any possible implementation thereof.

[0049] Fifthly, this application provides a computer program product that, when run on a processor, causes the processor to perform the method described in the first aspect or any possible implementation thereof.

[0050] Overall, the technical solutions conceived in this application have the following beneficial effects compared with the prior art:

[0051] This application provides a prosthesis control method, system, and device. By comparing the similarity between predicted motion data and real-time motion data, the likelihood of an amputee wearing a prosthesis being in different motion modes at future moments is predicted. Then, the control parameters under different motion modes are weighted and summed to generate hybrid control parameters, instead of using only a single motion state or terrain state. This fully considers the amputee's autonomous movement habits, achieves smooth transition control of the prosthesis, and realizes seamless switching and rapid identification of motion states.

[0052] This application provides a prosthesis control method, system, and device. By directly weighting the potentially suitable motion parameters and corresponding motion states for each amputee, automatic parameter adjustment is achieved, saving engineers' adjustment time and helping amputees quickly adapt to the prosthesis. Furthermore, the solution provided in this application does not require common terrain or motion state judgment sensors, including depth vision cameras, infrared rangefinders, and ultrasonic rangefinders. It only uses simple sensors such as IMUs and six-axis force sensors to comprehensively determine motion data and terrain patterns, thereby deriving the amputee's motion state and achieving prosthesis control that incorporates the amputee's movement habits. Attached Figure Description

[0053] Figure 1 This is a flowchart illustrating the prosthesis control method provided in an embodiment of this application;

[0054] Figure 2 This is a diagram showing the distribution of prosthetic sensors and the kinematic measurement points provided in the embodiments of this application;

[0055] Figure 3 This is a schematic diagram of the second neural network architecture provided in an embodiment of this application;

[0056] Figure 4 This is a schematic diagram of the prosthesis control parameter calculation framework provided in the embodiments of this application;

[0057] Figure 5 This is a label framework diagram of the output data of two neural networks provided in the embodiments of this application;

[0058] Figure 6 This is an architecture diagram of the prosthetic control system provided in an embodiment of this application;

[0059] Figure 7 This is an architectural diagram of the electronic device provided in the embodiments of this application. Detailed Implementation

[0060] To make the objectives, technical solutions, and advantages of this application 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 application.

[0061] In this article, the term "and / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. The symbol " / " in this article indicates that the related objects are in an "or" relationship; for example, A / B means A or B.

[0062] The terms "first" and "second," etc., used in the specification and claims herein are used to distinguish different objects, not to describe a specific order of objects. For example, "first neural network" and "second neural network," etc., are used to distinguish different neural networks, not to describe a specific order of neural networks.

[0063] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0064] In the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more, for example, multiple points means two or more points, etc.

[0065] The embodiments of this application are described below with reference to the accompanying drawings.

[0066] Figure 1 This is a flowchart of the prosthetic limb control method provided in the embodiments of this application; as follows: Figure 1 As shown, it includes the following steps:

[0067] Step S101: Obtain real-time motion data of multiple points on the prosthesis and the user's torso.

[0068] Optionally, real-time motion data at multiple points on the prosthesis and the user's torso can be acquired, including:

[0069] Real-time motion data of multiple first-type points are acquired by using multiple sensors installed on the prosthesis;

[0070] By combining real-time motion data of multiple first-type points on the prosthesis with inverse motion equations, real-time motion data of multiple second-type points on the user's torso are obtained.

[0071] like Figure 2 As shown, by way of example, multiple first-class points include: amputated thigh, prosthetic lower leg, prosthetic ankle joint, and prosthetic foot; multiple second-class points include: center point of chest and abdomen, center point of pelvis, healthy thigh, and healthy lower leg.

[0072] The core of the solution provided in this application lies in using sensor data, such as IMU sensors located on the patient's prosthetic thigh and calf, and other parts, as well as a six-axis force sensor on the prosthetic ankle joint, to solve the inverse motion equation. This allows for the calculation of the motion velocities of the main human torso, such as the center point of the chest and abdomen, the healthy leg and the amputated leg, as well as the angular velocity accelerations of the hip joints, knee joints, and ankle joints, and their relative motion data.

[0073] More specifically, the real-time motion data of the aforementioned multiple points includes: real-time motion data I of the first type of points and real-time motion data O of the second type of points.

[0074]

[0075]

[0076] in, Indicates thigh acceleration. Indicates the angular velocity of the thigh. This indicates the acceleration of the lower leg. Indicates the angular velocity of the lower leg. Indicates foot acceleration, Indicates foot angular velocity, This indicates that the ankle joint is under stress. This indicates the torque acting on the ankle joint. Indicates thigh length. Indicates the length of the lower leg. This represents the distance vector from the center of the ankle joint to the heel, from the center of the ankle joint to the forefoot landing point, and from the forefoot landing point to the heel landing point. This indicates the distance from the center point of the chest and abdomen to the center point of the pelvis. This indicates the distance from the center point of the pelvis to the center of rotation of the healthy hip joint. This indicates the distance from the center of rotation of the healthy hip joint to the center of rotation of the healthy knee joint. This indicates the distance from the center of rotation of the healthy knee joint to the center of rotation of the healthy ankle joint. This represents the distance vector from the center of the healthy ankle joint to the location of the healthy heel, from the center of the healthy ankle joint to the landing point of the forefoot of the healthy foot, and from the landing point of the forefoot of the healthy foot to the landing point of the heel of the healthy foot. This represents the estimated angular velocity at the center point of the thoracic and abdominal cavities. This represents the estimated acceleration at the center of the thoracic and abdominal cavities. This represents the estimated angular velocity at the center of the pelvis. This represents the estimated acceleration at the center of the pelvis. This represents the estimated angular velocity of the healthy thigh. This represents the estimated acceleration of the healthy thigh. This represents the estimated angular velocity of the healthy side leg. This represents the estimated acceleration of the healthy side leg.

[0077] Step S102: Input the real-time motion data of the multiple points into the first neural network to predict the motion data of each point under different preset motion modes at future times.

[0078] Optionally, the first neural network is trained using real-time motion data samples from multiple points and motion data samples under different preset motion modes at corresponding future times; the different motion modes correspond to different walking environments for users carrying prostheses; the walking environment includes at least one of the following types: flat ground, slopes of different gradients, stairs of different heights and / or different lengths, and obstacles of different heights and / or different lengths.

[0079] Step S103: Use the predicted motion data of each point as a reference sequence and the real-time motion data of each point as a control sequence to determine the similarity between the reference sequence and the control sequence for each point under each preset motion mode; combine the similarity of each point under each preset motion mode to determine the probability that a user carrying a prosthesis will be in each preset motion mode at a future time.

[0080] Specifically, the similarity between the reference sequence and the control sequence for each point in each preset motion mode is obtained through the following steps:

[0081] Obtain real-time updated control sequences based on real-time motion data of the locations;

[0082] A heuristic iterative algorithm is used to transform the control sequence updated at a single time step, and the optimal transformation parameters are obtained to minimize the DTW distance between the transformed control sequence and the reference sequence; the transformation includes scaling and / or translation.

[0083] The single-time similarity between the updated control sequence and the reference sequence at each time step is determined by combining the optimal transformation parameters.

[0084] The similarity scores at each moment within a preset time period are weighted and summed to obtain the similarity scores between the reference sequence and the control sequence within the preset time period. These scores are then used as the similarity scores between the reference sequence and the control sequence for the corresponding points. Among these scores, the weight of the single-moment similarity scores at later moments within the preset time period is greater than that at earlier moments.

[0085] Understandably, this process uses a heuristic loop algorithm to scale and translate the updated reference sequence to obtain a transformed reference sequence. Then, the data segments corresponding to the time length positions of the transformed reference sequence and the reference sequence are truncated to calculate the DTW distance. Based on the preceding heuristic algorithm, an optimal transformation parameter (including two for translation and two for scaling, for a total of four transformation parameters) is obtained. When the minimum DTW distance is obtained, the single-time similarity between the reference sequence and the reference sequence is obtained based on the evaluation function of the four transformation parameters. When the reference sequence is updated to the same time label as the reference sequence, the single-time similarity is weighted to obtain the time segment similarity. Thus, the similarity calculation of a data segment on a time segment is completed.

[0086] The greater the magnitude of the transformation, i.e. the larger the transformation parameter, the greater the difference between the reference sequence and the control sequence before the transformation, i.e., the lower the similarity between the reference sequence and the control sequence.

[0087] In one possible implementation, the single-time similarity is:

[0088]

[0089]

[0090] in, Represents the similarity at time n; To and Numerical functions, for or ; This represents the preset calculation coefficients, and t represents the sampling period of the real-time motion data. This indicates the sampling rate of real-time motion data; This represents the maximum value in the reference sequence within the time period t. and These represent the horizontal scaling parameter and the vertical scaling parameter, respectively. and These represent the horizontal translation parameter and the vertical translation parameter, respectively.

[0091] Because the longer the motion data appears, the more obvious the motion characteristics become, the similarity between the reference sequence and the control sequence within a preset time period can be defined as:

[0092]

[0093] in, express The similarity between the reference sequence and the control sequence within a time period; referring to the above formula, it can be seen that the larger n is, the larger its corresponding weight value is.

[0094] In one possible implementation, the probability of being in each preset motion mode is obtained through the following steps:

[0095] In a single preset motion pattern, the similarity of each point is weighted and summed to obtain the total similarity of the preset motion pattern; among them, the points farther away from the user's amputated thigh have lower weight values ​​when weighting and summing.

[0096] Normalize the total similarity of each preset motion pattern;

[0097] By combining the total similarity normalization, the probability of each preset motion mode at the current moment is updated to obtain the probability of each preset motion mode at the future moment.

[0098] Understandably, after completing the motion data prediction, the trunk motion data O is calculated using inverse kinematics from the real-time prosthetic motion data I. The gait progress prediction algorithm, calculated using similarity matching, calculates the gait progress within the most recent predicted gait cycle. Therefore, it can return matching parameters for each type of data (angular velocity and velocity of the chest and abdomen, thighs, and relative motion data between trunk segments). The predicted gait pattern and the motion data of each major trunk segment should perfectly match the actual motion data of the human body and prosthesis working together. All parameters in the matching calculation are ideal values; for example, translation and rotation parameters are 0, and scaling parameters are uniformly 1, i.e., no scaling. In reality, due to the accumulation of errors in inverse kinematics calculation and neural network prediction, the matching calculation parameters and ideal matching values ​​will differ to some extent.

[0099] When an amputee walks on a flat surface, even with the accumulation of inverse kinematics and neural network predictions, the error in the matching parameters will not be too large as long as the predicted movement pattern is also flat-surface walking. The error in the matching calculation only exceeds the acceptable range when the movement pattern prediction is incorrect, leading to a significant difference between the real-time movement trajectory data and the movement data under this pattern, resulting in a large discrepancy between the matching parameters and the ideal values. Therefore, these differences can be used to correct the gait pattern predicted by the neural network, thereby correcting the prosthetic limb movement pattern to better suit the amputee's own movement habits and characteristics.

[0100] In other words, this progress prediction system has self-correcting capabilities and high real-time discriminative ability. It can automatically adjust the prediction strategy and gait progress based on the amputee's real-time movement status. Furthermore, by utilizing the attention mechanism of the neural network, it can increase the proportion of losses in the prosthetic thigh, knee, calf, and ankle joints in the loss function, ensuring that the prosthesis can assist the amputee in completing motor tasks normally, even with a certain amount of matching error.

[0101] Step S104: Input the real-time motion data of multiple points into the second neural network to obtain the prosthesis control parameters corresponding to different preset motion modes.

[0102] Optionally, the second neural network is trained using real-time motion data samples from multiple points and corresponding prosthetic control parameter samples under different preset motion modes.

[0103] like Figure 3 As shown, by inputting the calculated estimated data and the actual data into the second neural network, the probability of the corresponding state is output ( Figure 3 (represented by a thin black arrow) Figure 3 The vertical axis represents the probability of the second neural network predicting the corresponding time from highest to lowest, and the horizontal axis represents time. For the conversion of the user's actual movement intent, see [link to relevant documentation]. Figure 3 The actual state transformation is shown in the figure. The light green box represents the most likely state after adjusting the state probability through similarity calculation.

[0104] It should be noted that in this application, the user's true movement intention is the corresponding true movement mode. The user's true movement intention is reflected by real-time movement data. Considering the uncertainty of the movement environment, the true intention may be adjusted in real time, and the user can smoothly transition between different movement modes when adjusting the movement mode in real time. However, when traditional prosthetic control switches between different movement modes, it may mechanically complete the previous movement mode before starting the next movement mode. Such control method cannot achieve a smooth transition between movement modes, resulting in the prosthetic control not being able to match the user's true movement intention well.

[0105] Therefore, this application matches predicted motion data for future moments with actual motion data to determine the probability of being in each motion mode at future moments, i.e., the probability of being in each motion mode at future moments is determined by referring to the amputee's actual motion intention. Then, referring to the control parameters under each motion mode, a weighted sum is obtained by combining the probability of each motion mode to obtain a comprehensive control parameter, thereby achieving control of the prosthesis. Through the above method, the control of the prosthesis can better match the amputee's actual motion intention, and the smooth transition control of the prosthesis can be achieved by combining the amputee's actual motion intention, allowing the amputee to completely control the progress and speed of the prosthesis's movement, and realizing the maximum manifestation of the amputee's motion intention.

[0106] Step S105: The prosthesis control parameters under different preset motion modes are weighted and summed based on the probability of being in each preset motion mode to obtain the final control parameters of the prosthesis.

[0107] Furthermore, now that the similarity of all states and all corresponding data points has been calculated, due to errors in the motion and inverse kinematics calculations of the prosthetic motor, the accuracy (or the reliability of the similarity) of points farther away from the thigh is actually lower. Therefore, when calculating the total similarity of a state, the weight of the similarity of points farther away from the thigh is lower.

[0108]

[0109] Among them, f, d, e,... This is a credibility weight, which can be adjusted based on the actual situation.

[0110] Furthermore, after the confidence level calculation is completed, the state probabilities need to be adjusted, including:

[0111] A. Normalization:

[0112] B. Possibility Update:

[0113] in It's the updated weight of the probability. Because The computation speed is 5 Hz, while the feedforward computation... The update speed is 10Hz, therefore It will be for the next calculation calculated at the same time. Continue to use.

[0114] After calculating the probability, the probability and the corresponding state control parameters are weighted and calculated using the probability as the weight to obtain the final control parameters, which are then sent to the lower-level machine to complete one system calculation.

[0115] In one possible implementation, the final control parameters of the prosthesis for:

[0116]

[0117] in, This represents real-time motion data from multiple points. n Indicates the total number of preset sports modes. Indicates the future time. i The possibility of preset motion modes Indicates the first i Prosthetic control parameters for a preset movement pattern.

[0118] It is understood that the embodiments of this application sum the current state and the possible future states, which can directly avoid the non-smooth transition of the finite state machine state transition (for possible states, we use parameter weights to select the calculation weights of the control parameters), based on the weight of the current state and the advance parameter setting of the most likely predicted future state; in the method provided by the embodiments of this application, the possible future states and probabilities are predicted first, and then the control parameters of the set state are generated by weighting and summing according to the prosthetic control parameters and probabilities of these states.

[0119] More specifically, such as Figure 4 The diagram shown is a calculation sequence diagram arranged according to the calculation time order provided in an embodiment of this application. Figure 4 The dashed box in the middle represents the Multi-Input Feature Matching (MFGE) algorithm, where M stands for Multi-Input and FGE is the feature matching algorithm.

[0120] MFGE comprises: a data network and similarity calculation; the aforementioned data network is the first neural network; the first neural network is used to predict the amputee's motion data at future moments. Similarity calculation refers to calculating the similarity between the predicted motion data and real-time motion data; event probability adjustment refers to adjusting the probability of each motion pattern at the current moment based on the calculated similarity to determine the probability of each motion pattern at future moments.

[0121] The event network, or second neural network, is used to combine real-time motion data to obtain prosthetic control parameters corresponding to different motion modes. For example... Figure 5 As shown, the event network outputs the possibilities of all states (motion modes) (i.e., prosthetic control parameters under different motion modes). Motion modes include ramps (variable: slope), stairs (variable: stair height, length), and obstacles (variable: obstacle height, length). Different combinations of variables in each state are considered as a sub-state and assigned possibilities; that is, as long as the variables are different, a state is defined. The parameters for each state are manually set in advance.

[0122] The data network outputs motion data for all states (i.e., motion modes), including predicted output data for all input point data. Specifically, if the data in one row of the input matrix is ​​the angular velocity of the thigh point, then the data at the corresponding position in the output matrix is ​​the predicted motion data for the next time interval t.

[0123] Both of the above data networks and the time network can be obtained using existing network training methods, which will not be elaborated here.

[0124] Figure 6 This is an architecture diagram of the prosthetic control system provided in this application, such as... Figure 6 As shown, it includes:

[0125] The motion data acquisition module 610 is used to acquire real-time motion data of multiple points on the prosthesis and the user's torso;

[0126] The motion data prediction module 620 is used to input the real-time motion data of the multiple points into the first neural network to predict the motion data of each point under different preset motion modes at future times.

[0127] The motion pattern determination module 630 is used to take the predicted motion data of each point as a reference sequence and the real-time motion data of each point as a comparison sequence to determine the similarity between the reference sequence and the comparison sequence for each point under each preset motion pattern; and combine the similarity of each point under each preset motion pattern to determine the probability that a user carrying a prosthesis will be in each preset motion pattern at a future time.

[0128] The control parameter acquisition module 640 is used to input the real-time motion data of the multiple points into the second neural network to obtain the prosthesis control parameters corresponding to different preset motion modes; and to weight and sum the prosthesis control parameters in different preset motion modes based on the probability of being in each preset motion mode to obtain the final control parameters of the prosthesis.

[0129] It should be understood that the above system is used to execute the methods in the above embodiments. The corresponding program modules in the system are similar in implementation principle and technical effect to those described in the above methods. The working process of the system can be referred to the corresponding process in the above methods, and will not be repeated here.

[0130] Based on the methods in the above embodiments, this application provides an electronic device, such as... Figure 7 As shown, the electronic device may include a processor 710, a communication interface 720, a memory 730, and a communication bus 740, wherein the processor 710, the communication interface 720, and the memory 730 communicate with each other via the communication bus 740. The processor 710 may call logical instructions in the memory 730 to execute the methods in the above embodiments.

[0131] Furthermore, the logical instructions in the aforementioned memory 730 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application.

[0132] Based on the methods in the above embodiments, this application provides a computer-readable storage medium storing a computer program that, when run on a processor, causes the processor to execute the methods in the above embodiments.

[0133] Based on the methods in the above embodiments, this application provides a computer program product that, when run on a processor, causes the processor to execute the methods in the above embodiments.

[0134] It is understood that the processor in the embodiments of this application can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. A general-purpose processor can be a microprocessor or any conventional processor.

[0135] The method steps in this application embodiment can be implemented in hardware or by a processor executing software instructions. The software instructions can consist of corresponding software modules, which can be stored in random access memory (RAM), flash memory, read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), registers, hard disks, portable hard disks, CD-ROMs, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor, enabling the processor to read information from and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can reside in an ASIC.

[0136] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted through the computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state disk (SSD)).

[0137] It is understood that the various numerical designations used in the embodiments of this application are merely for the convenience of description and are not intended to limit the scope of the embodiments of this application.

[0138] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A prosthesis control method for assisting a user in walking, the prosthesis being provided at an amputated thigh of the user, characterized by, Control methods include: Acquire real-time motion data at multiple points on the prosthesis and the user's torso; The real-time motion data of the multiple points are input into the first neural network to predict the motion data of each point under different preset motion modes at future times. The motion data of each predicted point is used as a reference sequence, and the real-time motion data of each point is used as a control sequence. The similarity between the reference sequence and the control sequence corresponding to each point under each preset motion mode is determined. By combining the similarity of each point under each preset motion mode, the probability of a user carrying a prosthesis being in each preset motion mode at a future time is determined. The real-time motion data of the multiple points are input into the second neural network to obtain the prosthesis control parameters corresponding to different preset motion modes; The prosthesis control parameters under different preset movement modes are weighted and summed based on the probability of being in each preset movement mode to obtain the final control parameters of the prosthesis. Acquire real-time motion data at multiple points on the prosthesis and the user's torso, including: By using multiple sensors installed on the prosthesis, real-time motion data of multiple first-type points are acquired; by combining the real-time motion data of multiple first-type points on the prosthesis, inverse motion equations are solved to obtain real-time motion data of multiple second-type points on the user's torso. Final control parameters of the prosthesis are: in, This represents real-time motion data from multiple points. n Indicates the total number of preset sports modes. Indicates the future time. i The possibility of preset motion modes Indicates the first i Prosthetic control parameters for a preset movement pattern; The similarity between the reference sequence and the control sequence for each point in each preset motion mode is obtained through the following steps: Obtain real-time updated control sequences based on real-time motion data of the locations; A heuristic iterative algorithm is used to transform the control sequence updated at a single time point, and the optimal transformation parameters are obtained to minimize the dynamic time adjustment distance between the transformed control sequence and the reference sequence. The single-time similarity between the updated control sequence and the reference sequence at each time point is determined by combining the optimal transformation parameters. The similarity scores at each moment within a preset time period are weighted and summed to obtain the similarity scores between the reference sequence and the control sequence within the preset time period. These scores are then used as the similarity scores between the reference sequence and the control sequence for the corresponding points. Among these scores, the weight of the single-moment similarity scores at later moments within the preset time period is greater than that at earlier moments. The probability of being in each preset motion mode is obtained through the following steps: In a single preset motion pattern, the similarity of the reference sequence and the control sequence at each point is weighted and summed to obtain the total similarity of the preset motion pattern; among them, the points farther away from the user's amputated thigh have lower weight values ​​when weighting and summing. Normalize the total similarity of each preset motion pattern; By combining the total similarity normalization, the probability of each preset motion mode at the current moment is updated to obtain the probability of each preset motion mode at the future moment.

2. The method of claim 1, wherein, The plurality of first-class points include: amputated thigh, prosthetic calf, prosthetic ankle joint, and prosthetic foot; the plurality of second-class points include: center point of chest and abdomen, center point of pelvis, healthy thigh, and healthy calf.

3. The method of claim 1, wherein, The first neural network is trained using real-time motion data samples from multiple points and motion data samples under different preset motion modes at corresponding future times. The second neural network is trained using real-time motion data samples from multiple points and corresponding prosthetic control parameter samples under different preset motion modes. The different preset movement modes correspond to different walking environments for users with prostheses.

4. The method of claim 1, wherein, The transformations include scaling and / or translation; The single-time similarity is: in, Represents the similarity at time n; To and Numerical functions, for or ; This represents the preset calculation coefficients, and t represents the sampling period of the real-time motion data. This indicates the sampling rate of real-time motion data; This represents the maximum value in the reference sequence within the time period t. and These represent the horizontal scaling parameter and the vertical scaling parameter, respectively. and These represent the horizontal translation parameter and the vertical translation parameter, respectively. The similarity between the reference sequence and the control sequence within the preset time period is: wherein, represents similarity of the reference sequence and the control sequence over the time period.

5. The method of claim 1, wherein, the likelihood of each preset motion pattern at the future time is: wherein, represents the likelihood of each preset motion pattern at the current time, represents the likelihood update weight, represents the total similarity normalization of each preset motion pattern.

6. A prosthesis control system, the prosthesis being provided at an amputated thigh of a user for assisting the user in walking, characterized in that The control system includes: The motion data acquisition module is used to acquire real-time motion data from multiple points on the prosthesis and the user's torso; The motion data prediction module is used to input the real-time motion data of the multiple points into the first neural network to predict the motion data of each point under different preset motion modes at future times. The motion pattern determination module is used to take the predicted motion data of each point as a reference sequence and the real-time motion data of each point as a control sequence to determine the similarity between the reference sequence and the control sequence for each point under each preset motion pattern; and combine the similarity of each point under each preset motion pattern to determine the probability that a user carrying a prosthesis will be in each preset motion pattern at a future time. The control parameter acquisition module is used to input the real-time motion data of the multiple points into the second neural network to obtain the prosthesis control parameters corresponding to different preset motion modes; and to weight and sum the prosthesis control parameters in different preset motion modes based on the probability of being in each preset motion mode to obtain the final control parameters of the prosthesis. Acquire real-time motion data at multiple points on the prosthesis and the user's torso, including: By using multiple sensors installed on the prosthesis, real-time motion data of multiple first-type points are acquired; by combining the real-time motion data of multiple first-type points on the prosthesis, inverse motion equations are solved to obtain real-time motion data of multiple second-type points on the user's torso. Final control parameters of the prosthesis are: in, This represents real-time motion data from multiple points. n Indicates the total number of preset sports modes. Indicates the future time. i The possibility of preset motion modes Indicates the first i Prosthetic control parameters for preset movement patterns; The similarity between the reference sequence and the control sequence for each point in each preset motion mode is obtained through the following steps: Obtain real-time updated control sequences based on real-time motion data of the locations; A heuristic iterative algorithm is used to transform the control sequence updated at a single time point, and the optimal transformation parameters are obtained to minimize the dynamic time adjustment distance between the transformed control sequence and the reference sequence. The single-time similarity between the updated control sequence and the reference sequence at each time point is determined by combining the optimal transformation parameters. The similarity scores at each moment within a preset time period are weighted and summed to obtain the similarity scores between the reference sequence and the control sequence within the preset time period. These scores are then used as the similarity scores between the reference sequence and the control sequence for the corresponding points. Among these scores, the weight of the single-moment similarity scores at later moments within the preset time period is greater than that at earlier moments. The probability of being in each preset motion mode is obtained through the following steps: In a single preset motion pattern, the similarity of the reference sequence and the control sequence at each point is weighted and summed to obtain the total similarity of the preset motion pattern; among them, the points farther away from the user's amputated thigh have lower weight values ​​when weighting and summing. Normalize the total similarity of each preset motion pattern; By combining the total similarity normalization, the probability of each preset motion mode at the current moment is updated to obtain the probability of each preset motion mode at the future moment.

7. An electronic device, comprising: include: At least one memory for storing computer programs; At least one processor is configured to execute a program stored in the memory, wherein when the program stored in the memory is executed, the processor is configured to perform the method as described in any one of claims 1-5.

Images