A gait ground reaction force prediction method based on OpenSim and black-box optimization technology

Abstract

The application discloses a gait ground reaction force prediction method based on OpenSim and black box optimization technology and belongs to the technical field of human musculoskeletal dynamics simulation. The application comprises the following contents: 1. acquiring kinematics data in a gait process by using a motion capture system; 2. scaling a general skeletal muscle model; 3. performing inverse kinematics analysis on the scaled model; 4. calculating a plantar pressure center based on foot morphology; 5. calculating resultant external force based on kinematics data; 6. distinguishing a double-foot support period and a single-foot support period based on foot kinematics data; 7. calculating resultant external force based on kinematics data; and 8. calculating ground reaction force distribution based on a black box optimization method. Compared with the prior art, the application can obtain double-foot ground reaction force distribution in a walking process only by using kinematics data.

Description

A Gait-Based Ground Reaction Prediction Method Based on OpenSim and Black-Box Optimization Technology Technical Field

[0001] This invention relates to the field of human musculoskeletal dynamics simulation technology, specifically a gait ground reaction force prediction method based on OpenSim and black-box optimization technology. Background Technology

[0002] Musculoskeletal dynamics simulation is now widely used in competitive sports, medical rehabilitation, and many other fields. Ground reaction force is crucial for calculating muscle exertion and activation, making accurate acquisition of ground reaction force particularly important. Currently, there are generally two methods for obtaining ground reaction force. The first is through experimental equipment, such as force plates or foot pressure insoles.

[0003] However, due to equipment limitations, force plates are generally not usable outdoors, and plantar pressure insoles can only obtain ground reaction forces perpendicular to the ground. The second method involves calculating ground reaction forces through kinematics. While ground reaction forces during a single-foot support phase can be obtained using the Newton-Euler equations, ground reaction forces during a two-foot support phase are difficult to obtain. Romain Van Hulle, Cedric Schwartz, and Vincent Denoel proposed a method for calculating the coordinates of the plantar pressure center in their paper, “Afoot / ground contact model for biomechanical inverse dynamics analysis”—Journal of Biomechanics 2019. This non-patented journal is cited in its entirety as prior art in this invention, specifically including:

[0004] Based on experimental data, the center of plantar pressure (COP) x D Position with foot tilt angle value The changes were indeed unrelated to the subject's activity or individual characteristics, therefore the location y of contact point D... D With y A y B y C and The following functional relationship exists between them:

[0005]

[0006]

[0007] Simply relying on this method is still insufficient to accurately calculate ground reaction force. For the reasons mentioned above, we propose a gait-based ground reaction force prediction method based on OpenSim and black-box optimization technology. Summary of the Invention

[0008] 1. The technical problem to be solved by the present invention

[0009] The purpose of this invention is to provide a gait ground reaction force prediction method based on OpenSim and black box optimization technology, which can obtain the distribution of bipedal ground reaction forces during walking using only kinematic data.

[0010] 2. Technical Solution

[0011] To achieve the above objectives, the present invention provides the following technical solution:

[0012] A gait-based ground reaction force prediction method based on OpenSim and black-box optimization technology includes the following steps:

[0013] S1. Use a motion capture system to acquire kinematic data during the subject's gait process;

[0014] S2. Select a general full-body musculoskeletal model containing all joints in the OpenSim software; and scale the general model constructed in S1 based on the subject's body size and weight to make it have a height and weight distribution similar to the subject's.

[0015] S3. Perform inverse kinematics analysis on the scaled model obtained in S2 to obtain rotation and translation data of various parts of the subject's body.

[0016] S4. Based on the scaling model and inverse kinematics results in S3, run inverse kinematics without adding the ground reaction force file, read the residual force values ​​and superimpose them into the ground reaction force file;

[0017] S5. Repeat the operation described in S4 until the residual force value is less than the given threshold, thus obtaining the resultant force of the ground reaction force of both feet during the gait process;

[0018] S6. Calculate the plantar pressure center based on empirical formulas and foot kinematic data;

[0019] S7. Based on foot kinematics data, distinguish between bipedal support phase and unipedal support phase;

[0020] S8. Based on the distinction results obtained in S7, if the current moment belongs to the single-leg support phase, the resultant force calculated in S5 is directly applied to the ground-contacting foot; if the current moment belongs to the double-leg support phase, the black-box optimization method is run to calculate the ground reaction force of the double feet.

[0021] S9. Return to S8 and continue calculating the ground reaction force value at the next moment until the calculation of the ground reaction force at all moments is completed.

[0022] Preferably, the S3 inverse kinematics analysis is performed using the least squares method.

[0023] Preferably, the calculation of the plantar pressure center of gravity in S6 specifically includes the following:

[0024] S6.1 Determine the vertical coordinates of the plantar pressure center based on the positions of the heel, the first metatarsal head, and the big toe;

[0025] S6.2. Based on the empirical formula proposed by Van Hulle et al. (2020), determine the coordinates of the plantar pressure center in the anterior-posterior direction, and establish the functional relationship between the anterior-posterior coordinates of the heel, the first metatarsal bone, and the big toe.

[0026] Preferably, S7 specifically includes the following:

[0027] S7.1 Calculate the velocity at the center of the foot based on kinematic data;

[0028] S7.2. Use peak detection method to identify the maximum value of vertical velocity at the center of the foot, where the moment of the maximum value is the moment when the toes leave the ground.

[0029] S7.3 Combine the peak detection method with the displacement constraint of the heel marker point from the ground to identify the moment when the heel touches the ground;

[0030] S7.4. Based on the timing of the toe-off-ground movement determined in S7.2 and the timing of the heel-on-ground movement determined in S7.3, distinguish between the two-foot support phase and the one-foot support phase.

[0031] Preferably, the calculation of the bipedal ground reaction force using the black-box optimization method described in S8 specifically refers to using the resultant force of the bipedal ground reaction force obtained in S5 as a constraint, and using the black-box optimization method to run the inverse dynamics minimization objective function to obtain the bipedal ground reaction force. The objective function is:

[0032]

[0033] In the formula, F Vt F APt M1, M2, and M3 represent the vertical and forward / backward ground forces acting on the foot at the current moment; M1, M2, and M3 represent the three-axis torques acting on the human body at the current moment.

[0034] 3. Beneficial effects

[0035] This invention proposes a gait ground reaction force prediction method based on OpenSim and black-box optimization technology, providing a more efficient and easier-to-implement approach for GRF estimation during gait. It is expected to have wide applications in medicine, rehabilitation, and biomechanics, and will also contribute to a better understanding and analysis of human movement characteristics. Furthermore, the robustness of this method makes it suitable for various research environments, allowing for rapid expansion into activities such as vertical jumps, single-leg support, and running. Attached Figure Description

[0036] Figure 1 is a schematic diagram of the method flow of the gait ground reaction force prediction method based on OpenSim and black box optimization technology proposed in this invention.

[0037] Figure 2 is a graph showing the functional relationship between the foot segment tilt angle and the front-to-back direction coordinates mentioned in Embodiment 1 of the present invention;

[0038] Figure 3 is a comparison chart of the optimization results mentioned in Embodiment 3 of the present invention. Detailed Implementation

[0039] The following describes in detail, with reference to the accompanying drawings and specific embodiments, a gait ground reaction force prediction method based on OpenSim and black-box optimization technology provided by the present invention.

[0040] It should also be noted that, in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments. For some well-known technologies, those skilled in the art can also use other alternative methods to implement them. Moreover, the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0041] This invention encompasses any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of this invention. To provide the public with a thorough understanding of this invention, specific details are described in detail in the following preferred embodiments; however, those skilled in the art will fully understand the invention even without these details. Furthermore, to avoid unnecessary misunderstanding of the essence of this invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0042] Example 1:

[0043] This invention proposes a gait-based ground reaction force prediction method based on OpenSim and black-box optimization technology, characterized by the following steps:

[0044] S1. Use a motion capture system to acquire kinematic data during the subject's gait process;

[0045] S2. Select a general full-body musculoskeletal model containing all joints in the OpenSim software; and scale the general model constructed in S1 based on the subject's body size and weight to make it have a height and weight distribution similar to the subject's.

[0046] S3. Perform inverse kinematics analysis on the scaled model obtained in S2 using the least squares method to obtain rotation and translation data of various parts of the subject's body.

[0047] S4. Based on the scaling model and inverse kinematics results in S3, run inverse kinematics without adding the ground reaction force file, read the residual force values ​​and superimpose them into the ground reaction force file;

[0048] S5. Repeat the operation described in S4 until the residual force value is less than the given threshold, thus obtaining the resultant force of the ground reaction force of both feet during the gait process;

[0049] S6. Calculate the plantar pressure center based on empirical formulas and foot kinematic data, specifically including the following:

[0050] S6.1 Determine the vertical coordinates of the plantar pressure center based on the positions of the heel, the first metatarsal head, and the big toe;

[0051] S6.2. Determine the coordinates of the plantar pressure center in the anterior-posterior direction based on empirical formulas, and establish the functional relationship between the foot inclination angle and the coordinates in the anterior-posterior direction. The specific content is as follows:

[0052] Foot kinematics mainly includes temporal data from three markers (heel, big toe, and first metatarsal). The coordinates of the plantar pressure center are mainly divided into three directions. The vertical coordinate is consistent with the ground; the anterior-posterior coordinate is determined based on the empirical formula proposed by Van Hulle et al. (2020), which establishes the relationship between the foot tilt angle and the anterior-posterior coordinate; and since the left-right coordinate of the plantar pressure center remains basically unchanged during walking, the change in the left-right coordinate value is ignored, and it is assumed to be located at the center.

[0053] Please refer to Figure 2, where θ4 represents the foot tilt angle value, and δ HS δ HO δ MO δ TO represents the tilt angle at the corresponding moments of heel touchdown, heel liftoff, metatarsal liftoff, and toe liftoff. Where x A x B x C These represent the front-back coordinates of the heel, the front-back coordinates of the first metatarsal bone, and the front-back coordinates of the big toe, respectively.

[0054] S7. Based on foot kinematic data, distinguish between the bipedal support phase and the unipedal support phase, specifically including the following:

[0055] S7.1 Calculate the velocity at the center of the foot based on kinematic data;

[0056] S7.2. Use peak detection method to identify the maximum value of vertical velocity at the center of the foot, where the moment of the maximum value is the moment when the toes leave the ground.

[0057] S7.3 Combine the peak detection method with the displacement constraint of the heel marker point from the ground to identify the moment when the heel touches the ground;

[0058] S7.4. Based on the timing of the toe-off-ground movement determined in S7.2 and the timing of the heel-on-ground movement determined in S7.3, distinguish between the two-foot support phase and the one-foot support phase.

[0059] S8. Based on the distinction obtained in S7, if the current moment belongs to the single-leg support phase, the resultant force calculated in S5 is directly applied to the contacting foot; if the current moment belongs to the double-leg support phase, a black-box optimization method is used to calculate the double-leg ground reaction force. Specifically, using the resultant force of the double-leg ground reaction force obtained in S5 as a constraint, the black-box optimization method is used to run the inverse dynamics minimization objective function to obtain the double-leg ground reaction force. The objective function is:

[0060]

[0061] In the formula, F Vt F APt M1, M2, and M3 represent the vertical and forward / backward ground forces acting on the foot at the current moment; M1, M2, and M3 represent the three-axis torques acting on the human body at the current moment.

[0062] S9. Return to S8 and continue calculating the ground reaction force value at the next moment until the calculation of the ground reaction force at all moments is completed.

[0063] Example 2:

[0064] Please refer to Figure 1, which is based on Embodiment 1 but differs in that...

[0065] This invention proposes a gait-based ground reaction force prediction method based on OpenSim and black-box optimization technology. The specific usage process is as follows:

[0066] (1) First, a motion capture experiment was conducted to collect human kinematic and dynamic data. After the experiment, the data was processed.

[0067] (2) The general model is scaled according to the kinematic data to obtain a musculoskeletal dynamic model suitable for the subject;

[0068] (3) Perform inverse kinematics to make the model have a motion sequence similar to the actual motion of the test subject;

[0069] (4) Perform inverse dynamics, first add the initial value of the external force for calculation, then obtain the residual force at the pelvis, and then update the resultant external force value according to its value until the residual force is less than the preset threshold.

[0070] (5) Estimate the COP value based on foot kinematic data at different times and empirical formulas;

[0071] (6) Determine whether the current moment belongs to a single support segment or a double support segment based on foot kinematic data. If it is a single support segment, the external force obtained in (4) can be directly applied to the supporting leg; if it is a double support segment, the following optimization is performed: first, a set of vertical and forward-backward force values ​​of a single foot are randomly given for inverse dynamics, and the residual torque at the pelvis is minimized through black box optimization technology until the maximum number of iterations is reached. Record the calculation results at this time.

[0072] (7) Repeat the above calculations until the last moment of the gait phase is reached.

[0073] Example 3:

[0074] Please refer to Figure 3. The optimization effect of the gait ground reaction prediction method based on OpenSim and black-box optimization technology proposed in this invention will be explained below with specific examples, including the following:

[0075] Figure 3 shows the comparison between the optimized and actual values ​​of fore-and-aft ground reaction force and vertical ground reaction force for different subjects during gait. Slow gait speed is defined as 0.8 m / s, normal gait speed as 1.0 m / s, and fast gait speed as 1.2 m / s. At slow gait speed, the mean rRMSE (rRMSE) of the actual and optimized values ​​of vertical ground reaction force was 7.0%, and the mean rRMSE of the actual and optimized values ​​of fore-and-aft ground reaction force was 14.1%. At normal gait speed, these values ​​were 6.1% and 12.1%, respectively; and at fast gait speed, they were 5.7% and 9.4%, respectively.

[0076] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and its improved concept, should be covered within the scope of protection of the present invention.

Claims

1. A gait-based ground reaction force prediction method based on OpenSim and black-box optimization technology, characterized in that, Includes the following steps: S1. Use a motion capture system to acquire kinematic data during the subject's gait process; S2. In OpenSim software, select a general full-body musculoskeletal model containing all joints; and scale the general model constructed in S1 based on the subject's body size and weight to make it have a height and mass distribution similar to the subject's; S3. Perform inverse kinematics analysis on the scaled model obtained in S2 to obtain rotation and translation data of various parts of the subject's body; S4. Based on the scaled model and inverse kinematics results in S3, run inverse kinematics without adding ground reaction force files, read the residual force values ​​and superimpose them into the ground reaction force files; S5. Repeat step S4 until the residual force value is less than the given threshold, thus obtaining the resultant force of the ground reaction forces of both feet during gait; S6. Calculate the plantar pressure center based on empirical formulas and foot kinematic data, specifically including the following: S6.

1. Determine the vertical coordinates of the plantar pressure center based on the positions of the heel, first metatarsal head, and big toe; S6.

2. Determine the anterior-posterior coordinates of the plantar pressure center based on the following empirical formulas: In the formula, y A y B y C y D These represent the positions of contact points A, B, C, and D, respectively, and x D Indicates the location of the center of pressure on the sole of the foot. Indicate the foot tilt angle value; establish the functional relationship between the anteroposterior coordinates of the heel, the anteroposterior coordinates of the first metatarsal bone, and the anteroposterior coordinates of the big toe; S7, based on foot kinematic data, distinguish between the bipedal support phase and the unipedal support phase; S8. Based on the distinction results obtained in S7, if the current moment belongs to the single-leg support phase, the resultant force calculated in S5 is directly applied to the grounding foot. Specifically, the resultant force of the ground reaction forces of both feet obtained in S5 is used as a constraint, and the black-box optimization method is used to run the inverse dynamics minimization objective function to obtain the ground reaction forces of both feet. If the current moment belongs to the double-leg support phase, the black-box optimization method is run to calculate the ground reaction forces of both feet. S9. Return to S8 and continue calculating the ground reaction force value at the next moment until the calculation of the ground reaction force at all moments is completed.

2. The gait ground reaction prediction method based on OpenSim and black-box optimization technology according to claim 1, characterized in that, In step S3, the inverse kinematics analysis is performed using the least squares method.

3. The gait ground reaction prediction method based on OpenSim and black-box optimization technology according to claim 1, characterized in that, S7 specifically includes the following: S7.1 Calculating the velocity at the center of the foot based on kinematic data; S7.2 Identifying the maximum vertical velocity at the center of the foot using a peak detection method, wherein the moment of the maximum velocity is the moment when the toes lift off the ground; S7.3 Combining the peak detection method with the displacement constraint of the heel marker point from the ground to identify the moment when the heel touches the ground; S7.4 Distinguishing between the two-foot support phase and the one-foot support phase based on the moment when the toes lift off the ground is determined in S7.2 and the moment when the heel touches the ground is determined in S7.

3.

4. The gait ground reaction prediction method based on OpenSim and black-box optimization technology according to claim 1, characterized in that, The objective function described in S8 is: In the formula, F Vt F APt M1, M2, and M3 represent the vertical and forward / backward ground forces acting on the foot at the current moment; M1, M2, and M3 represent the three-axis torques acting on the human body at the current moment.

Images