A motion following control method, system and storage medium of a wearable robot

Abstract

The application discloses a motion following control method and system of a wearable robot and a storable medium, and relates to the technical field of robot control, wherein the method comprises the following steps: realizing the following control of the wearable robot by constructing a tracking differentiator, improving an inverse dynamics control model and a joint torque representation model, and enabling the wearable robot to follow the motion of a wearer in real time; and the application can generate a control instruction which is more suitable for the actual working condition of a lower-limb wearable robot.

Description

Technical Field

[0001] This invention relates to the field of robot control technology, and more specifically to a method, system, and storage medium for motion following control of a wearable robot. Background Technology

[0002] Currently, exoskeleton robots are wearable devices that can move with the human body's limbs to complete tasks. They mainly consist of a frame worn on the outside of the human body and a power system. Their characteristic is that they rely on the power system to help the human limb joints and muscles move, in order to enhance, extend, compensate for or replace the functions and abilities of the human limbs.

[0003] However, a major challenge in controlling assistive exoskeletons is how to enable users to perform natural movements during interaction with the exoskeleton. Existing control methods largely rely on high-cost sensors, such as those for collecting EEG, EEG, and surface electromyography (EMG) signals. However, these sensors are typically complex to calibrate, susceptible to interference, have limited reliability, often require high learning costs, and are limited in wearability. Therefore, it is necessary to research control strategies that rely less on sensors. Currently, the commonly used control strategy to address these issues is the Sensitivity Amplification Control (SAC) algorithm. However, the SAC algorithm faces the following problems in practical applications: a) Directly calculating the angular acceleration of joints using differentiation poses a challenge in obtaining high-quality angular acceleration signals; low-quality acceleration signals can introduce large disturbances to the system; b) The control system is more sensitive to changes in acceleration than to changes in velocity; using a single amplification factor for the inertial force, centrifugal force, and Coriolis force terms in the control system is not the optimal solution; c) Traditional SAC commands are based on inverse dynamics models for the single-leg swing phase and the standing phase, which leads to sudden and discontinuous changes in control commands during gait phase transitions. This instability in the control system can cause sudden changes in the forces acting on the joints, thus reducing the wearing experience while walking.

[0004] Therefore, how to provide a wearable robot motion following control method that can solve the above problems is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] In view of this, the present invention provides a motion following control method, system and storage medium for wearable robots, which can generate control commands that are more adapted to the actual working conditions of lower limb wearable robots.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A motion-following control method for a wearable robot, wherein the wearable robot is mounted on the lower limb of a wearer during use, and multiple joint motors are provided on the lower limb of the wearer, comprising the following steps:

[0008] The pressure on the wearer's feet is acquired, along with the angle and angular velocity information of the joint motor.

[0009] A tracking differentiator is constructed, and corresponding angular acceleration information is generated based on the tracking differentiator, the angle information, and the angular velocity information. At the same time, an inverse dynamics control model is constructed based on the plantar pressure, the angle information, the angular velocity information, and the angular acceleration information.

[0010] Gait factors are obtained, and the inverse dynamics control model is processed based on the gait factors, the angle information, the angular acceleration information, and the angular velocity information to obtain the corresponding joint torque representation model;

[0011] The wearable robot is controlled by using the joint torque representation model, so that the wearable robot can follow the wearer's movements in real time.

[0012] Preferably, the specific process for constructing the inverse dynamics control model includes:

[0013] Based on the structure and active degrees of freedom of the wearable robot, a five-bar linkage model is constructed in the sagittal plane of the wearer, and the five-bar linkage model is divided into a swing leg and a support leg according to the plantar pressure.

[0014] Based on the Lagrange method and in conjunction with the angle information, angular velocity information, and angular acceleration information, inverse dynamics models of the swing leg and the supporting leg are constructed respectively.

[0015] Preferably, the specific process for constructing the inverse dynamics control model also includes:

[0016] The swing phase inertial sensitivity amplification factor is introduced to improve the swing leg inverse dynamics model, and the support phase inertial sensitivity amplification factor is introduced to improve the support leg inverse dynamics model.

[0017] Preferably, the specific processing steps for obtaining the corresponding joint torque representation model include:

[0018] A gait factor is introduced, wherein the gait factor includes a gait switching factor and a gait support factor;

[0019] A joint torque representation model is constructed based on the improved swing leg inverse dynamics model, the improved support leg inverse dynamics model, the gait switching factor, and the gait support factor.

[0020] Preferably, the specific processing steps for following and controlling the wearable robot further include:

[0021] The wearable robot is controlled by using the current loop of the joint motor and the joint torque representation model.

[0022] The present invention also provides a control system for a motion-following control method for a wearable robot, comprising:

[0023] The acquisition module is used to acquire the plantar pressure of the wearer, and at the same time acquire the angle information and angular velocity information of the joint motor;

[0024] The first model building module is used to build a tracking differentiator and generate corresponding angular acceleration information based on the tracking differentiator, the angle information, and the angular velocity information. At the same time, it builds an inverse dynamics control model based on the plantar pressure, the angle information, the angular velocity information, and the angular acceleration information.

[0025] The second model construction module is used to obtain gait factors and process the inverse dynamics control model based on the gait factors, the angle information, the angular acceleration information and the angular velocity information to obtain the corresponding joint torque representation model.

[0026] The control module is used to perform follow control on the wearable robot using the joint torque representation model, so that the wearable robot follows the wearer's movements in real time.

[0027] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the motion-following control method for a wearable robot as described in any of the preceding claims.

[0028] As can be seen from the above technical solutions, compared with the prior art, the present invention discloses a method, system, and storage medium for motion following control of a wearable robot, which has the following beneficial effects:

[0029] 1. It integrates a tracking differentiator, which reduces the jitter and number of spikes in the joint angular acceleration signal, and improves the input accuracy for calculating the inverse dynamics model;

[0030] 2. To address the issue of varying sensitivities of the control system to speed and acceleration signals, a finely subdivided sensitivity amplification factor was set to generate control commands that are more adapted to the actual working conditions of the lower limb exoskeleton robot.

[0031] 3. A gait switching factor was designed. When the lower limb exoskeleton switches between the swing phase and the support phase, the system model can be considered to change continuously without abrupt changes, thus achieving smoother and more continuous torque command generation.

[0032] 4. A support factor is introduced. When the lower limb exoskeleton is in the double support phase, the control torque command is distributed proportionally between the two legs according to the pressure on the soles of the feet.

[0033] 5. A smooth motion algorithm for weight-bearing assistive exoskeletons is proposed. Using only two sensors—joint angle sensors and plantar pressure sensors—it improves the walking experience of wearing the exoskeleton. Attached Figure Description

[0034] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0035] Figure 1 A flowchart of an overall motion-following control method for a wearable robot provided by the present invention;

[0036] Figure 2 This is a block diagram of the SAC algorithm provided in an embodiment of the present invention;

[0037] Figure 3 This is a block diagram of the ESAC algorithm provided in an embodiment of the present invention;

[0038] Figure 4 This invention provides a schematic diagram of the motion following control system structure for a wearable robot.

[0039] Figures 5(a)-5(b) A comparison chart of different algorithms used to calculate hip and knee joint angular acceleration in embodiments of the present invention;

[0040] Figures 6(a)-6(b) A comparison diagram of the actual torque generated by SAC and ESAC provided in the embodiments of the present invention. Detailed Implementation

[0041] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0042] See appendix Figure 1 As shown in the figure, an embodiment of the present invention discloses a motion following control method for a wearable robot. The wearable robot is mounted on the lower limb of the wearer during use, and multiple joint motors are provided on the lower limb. The method includes the following steps:

[0043] It acquires the wearer's plantar pressure, as well as the angle and angular velocity information of the joint motor. The angle and angular velocity information comes from the encoder feedback of the joint motor itself, which is directly acquired through the joint motor's CAN (Controller Area Network) communication interface.

[0044] Construct a tracking differentiator, and generate corresponding angular acceleration information based on the tracking differentiator, angle information, and angular velocity information. At the same time, construct an inverse dynamics control model based on plantar pressure, angle information, angular velocity information, and angular acceleration information.

[0045] Gait factors are obtained, and the inverse dynamics control model is processed based on gait factors, angle information, angular acceleration information, and angular velocity information to obtain the corresponding joint torque representation model.

[0046] The wearable robot is controlled by using a joint torque representation model, enabling the wearable robot to follow the wearer's movements in real time.

[0047] Specifically, the implementation process of the tracking differentiator is as follows:

[0048] Assuming the current and speed loops are accurate, the position setting value and the position feedback value are added or subtracted, then passed through a PI module, an optional filter, and a limiting module to output the speed value. This speed value then drives the motor via the speed loop and the current loop. The motor, through an encoder, feeds back the position parameters to the system, forming a closed loop. However, the tracking differentiator is discrete to match the discrete signals of the articulated motor.

[0049] The Tracking Differentiator (TD) uses the fast optimal control synthesis function fhan(.) for discrete systems, which reduces noise in both the original signal and the differentiated signal, and avoids high-frequency chattering that occurs in digital computation. The algorithm for the discrete form of the tracking differentiator is shown below:

[0050]

[0051] Where e0(k) is the input signal sequence of the tracking differentiator algorithm, e1(k) is the tracking signal (or filtered signal) output by the tracking differentiator algorithm relative to e0(k), e2(k) is the differential signal (or derivative signal) output by the tracking differentiator algorithm relative to e0(k), r is the speed factor determining the tracking speed, h0 is the filtering factor, and h is the integration step size calculated by the digital controller. h0 is chosen as a parameter appropriately larger than the integration step size h (e.g., h0 = 20h) to suppress noise amplification in the differential signal. Note:

[0052]

[0053] Then the definition of fhan can be expressed as follows: for fh = fhan(x1, x2, r, h):

[0054]

[0055] In the formula, x = (x1, x2) represents the independent variables position and velocity, respectively; fh is the calculated optimal acceleration; l is the length of the linear segment interval of the function fsg; r is the input parameter of the function fhan; e1(k) tracks e0(k), where e0(k) is equivalent to the input discrete position signal sequence, i.e., the ideal position (angle) signal; e1(k) is equivalent to the actual acquired output discrete position signal sequence, i.e., the actual position (angle) signal; e2(k) is equivalent to the actual velocity (angular velocity) signal; and fh is the calculated optimal acceleration (angular acceleration) signal. Experiments were conducted to evaluate how the tracking differentiator (TD) algorithm and the baseline difference algorithm affect the smoothness of the generated angular acceleration signal. Specific results can be found in [link to relevant documentation]. Figures 5(a)-5(b) As shown.

[0056] In a specific embodiment, the specific process of constructing the inverse dynamics control model includes:

[0057] Based on the structure and active degrees of freedom of the wearable robot, a five-bar model is constructed in the sagittal plane of the wearer, and the five-bar model is divided into swing legs and supporting legs according to the plantar pressure.

[0058] Based on the Lagrange method and combined with angle, angular velocity, and angular acceleration information, inverse dynamic models for the swing leg and supporting leg are constructed respectively. These two models together form the inverse dynamic model, the specific expression of which is as follows:

[0059]

[0060]

[0061] In the formula, τ swFor the inverse dynamics model of the lower limb exoskeleton when one leg is in the swing phase, M sw C represents the inertia term of the swing leg model. sw For the centrifugal force and Coriolis force terms of the swing leg model, G sw Let θ be the gravity term for the swing leg model. sw The angle of the swinging leg joint. The joint angular velocity of the swinging leg, τ is the joint acceleration of the swinging leg angle. st For the inverse dynamics model of the lower limb exoskeleton with one leg in the support phase, M st For the inertia term of the supporting leg model, C st To support the centrifugal force and Coriolis force terms of the leg model, G st For the gravity term of the supporting leg model, θ st To support the joint angle of the leg, To support the joint angular velocity of the leg, The joint angular acceleration of the supporting leg.

[0062] Specifically, the inverse dynamics control model is constructed using the SAC algorithm. The SAC algorithm defines the force / torque exerted by the human body on the exoskeleton as a sensitivity function, and then amplifies this function, thereby achieving the effect of controlling the exoskeleton movement with a relatively small force. Its principle is to establish a mapping relationship between the human body's torque provided by the wearer and the kinematic variables of the lower limb exoskeleton, forming a new upper-level closed-loop feedback. The block diagram of the SAC algorithm is shown below. Figure 2 As shown.

[0063] Figure 2 G represents the wearer's inverse dynamics transfer function. exo Let G' be the dynamic transfer function of the lower limb exoskeleton. exo The lower limb exoskeleton inverse dynamics model is embedded in the controller program, where α is the system sensitivity amplification factor with a value range of [1, +), and q represents the motion state of each joint of the lower limb exoskeleton, including the angle, angular velocity, and angular acceleration of each joint. h T represents the movement status of various joints in the wearer's lower limbs. h T represents the interaction torque between the lower limb exoskeleton and the wearer's body, referred to the joints. ɑ The torques provided by the motors to the joints of the lower limb exoskeleton are represented by , and T represents the resultant torque on the joints of the lower limb exoskeleton. The three torques satisfy the following relationship:

[0064] T = T h +T ɑ (6)

[0065] And T ɑThe value is determined by the control output of the controller. Since sensitivity amplification control is a control method based on inverse kinematics models, it is highly dependent on the accuracy of the model. When the inverse kinematics model in the controller is completely consistent with the actual inverse kinematics model of the lower limb exoskeleton, we have:

[0066] G' exo =G -1 exo (7)

[0067] This leads to T and T' ɑ The relationship is as follows:

[0068] T a =(1-α) -1 )G' exo G exo T (8)

[0069] Combining equations (3), (4), and (5), we obtain:

[0070] T = αT h (9)

[0071] Therefore, by changing the value of the sensitivity amplification factor α, different torque amplification factors can be achieved. That is, when α = n, the final resultant torque of the lower limb exoskeleton is n times the torque experienced by the wearer, thus achieving a torque amplification effect. Theoretically, when the inverse dynamics of the lower limb exoskeleton robot are modeled with complete accuracy, increasing the sensitivity amplification factor can make the human-robot interaction torque approach zero. However, since the inverse dynamics model obtained from the modeling contains simplifications and assumptions about the actual physical system, there will always be errors. Considering the errors and system stability, in actual control systems, a larger sensitivity amplification factor is not always better. In addition to increasing the accuracy of the model through appropriate methods, it is also necessary to debug and determine the appropriate sensitivity amplification factor based on the test results.

[0072] In one specific embodiment, see Appendix Figure 3 As shown, the specific process of constructing the inverse dynamics control model also includes:

[0073] The inverse dynamics model of the swing leg is improved by introducing sensitivity amplification factors for the swing phase inertia term and the centripetal Coriolis term. Similarly, the inverse dynamics model of the support leg is improved by introducing sensitivity amplification factors for the support phase inertia term and the centripetal Coriolis term. This process is known as the ESAC algorithm, and its specific expression is as follows:

[0074]

[0075]

[0076] In the formula, This is the sensitivity amplification factor for the swing leg inertia term. This is the sensitivity amplification factor for the centripetal Coriolis term of the swing leg. The sensitivity amplification factor for the inertial term of the supporting leg. The amplification factor for the centripetal Coriolis term sensitivity of the supporting leg.

[0077] In a specific embodiment, the specific processing steps for obtaining the corresponding joint torque representation model include:

[0078] A gait factor is introduced, which includes a gait switching factor and a gait support factor;

[0079] A joint torque representation model is constructed based on the improved swing leg inverse dynamics model, the improved support leg inverse dynamics model, the gait switching factor, and the gait support factor.

[0080] Specifically, the gait switching factor β is used to generate smooth and continuous torque commands when the lower limb exoskeleton transitions between the swing leg and the supporting leg, and the gait support factor γ is used to generate smooth and continuous torque commands when the lower limb exoskeleton transitions between the weight-bearing legs during the standing posture. The joint torque representation model τ of the lower limb exoskeleton is described as follows:

[0081] τ=βγτ st +(1-β)τ sw (12)

[0082] Define the pressure acquired by the plantar pressure sensor as F. p Includes left foot plantar pressure F pL and right foot plantar pressure F pR The threshold for plantar pressure on the swing leg is ε. sw The support leg plantar pressure threshold is ε st .

[0083] Then, design the gait switching factor β:

[0084]

[0085] The gait switching factor β represents the inverse dynamics model in which the control torque smoothly and proportionally switches between the gait stages of the lower limb exoskeleton during walking, i.e., when the same leg switches between support and swing states.

[0086] Finally, the gait support factor γ is designed:

[0087]

[0088] The gait support factor γ represents the smooth and proportional distribution of control torque between the two legs when the lower limb exoskeleton is in a standing position, based on the portion of the plantar pressure of both legs that exceeds the plantar pressure threshold of the swing leg.

[0089] In one specific embodiment, the specific processing steps for following and controlling a wearable robot further include:

[0090] The wearable robot is controlled by using the current loop of the joint motor and the joint torque representation model.

[0091] See appendix Figure 4 As shown, this embodiment of the invention also provides a control system for a wearable robot motion following control method utilizing any of the above embodiments, comprising:

[0092] The acquisition module is used to acquire the plantar pressure of the wearer, as well as the angle and angular velocity information of the joint motor;

[0093] The first model building module is used to build a tracking differentiator and generate corresponding angular acceleration information based on the tracking differentiator, angle information, and angular velocity information. At the same time, it builds an inverse dynamics control model based on plantar pressure, angle information, angular velocity information, and angular acceleration information.

[0094] The second model construction module is used to obtain gait factors and process the inverse dynamics control model based on gait factors, angle information, angular acceleration information and angular velocity information to obtain the corresponding joint torque representation model.

[0095] The control module is used to perform follow control of the wearable robot using a joint torque representation model, enabling the wearable robot to follow the wearer's movements in real time.

[0096] This invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the motion-following control method for a wearable robot as described in any of the above embodiments.

[0097] The method provided in this embodiment was used for testing. In this test, the test subject performed a straight forward walking motion from 0 seconds to 10 seconds, and then transitioned to a squatting motion after 10 seconds. The actual torque generated by the ESAC algorithm was compared with that generated by the SAC algorithm on the same joint (the left hip joint was selected in this embodiment). The results are as follows. Figures 6(a)-6(b) As shown.

[0098] The results show that the SAC without any extension produces relatively large jitter when generating actual torque, especially when switching the gait phase and motion mode of the assisted exoskeleton, while the ESAC effectively suppresses this jitter, demonstrating the advantage of the ESAC algorithm in suppressing control torque jitter.

[0099] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0100] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A motion-following control method for a wearable robot, wherein the wearable robot is mounted on the lower limb of a wearer during use, and multiple joint motors are provided on the lower limb of the wearer, characterized in that, Includes the following steps: The pressure on the wearer's feet is acquired, along with the angle and angular velocity information of the joint motor. A tracking differentiator is constructed, and corresponding angular acceleration information is generated based on the tracking differentiator, the angle information, and the angular velocity information. At the same time, an inverse dynamics control model is constructed based on the plantar pressure, the angle information, the angular velocity information, and the angular acceleration information. Gait factors are obtained, and the inverse dynamics control model is processed based on the gait factors, the angle information, the angular acceleration information, and the angular velocity information to obtain the corresponding joint torque representation model; The wearable robot is controlled by using the joint torque representation model, so that the wearable robot can follow the wearer's movements in real time.

2. The motion following control method for a wearable robot according to claim 1, characterized in that, The specific process of constructing an inverse dynamics control model includes: Based on the structure and active degrees of freedom of the wearable robot, a five-bar linkage model is constructed in the sagittal plane of the wearer, and the five-bar linkage model is divided into a swing leg and a support leg according to the plantar pressure. Based on the Lagrange method and in conjunction with the angle information, angular velocity information, and angular acceleration information, inverse dynamics models of the swing leg and the supporting leg are constructed respectively.

3. The motion following control method for a wearable robot according to claim 2, characterized in that, The specific process of constructing the inverse dynamics control model also includes: The swing phase inertial sensitivity amplification factor is introduced to improve the swing leg inverse dynamics model, and the support phase inertial sensitivity amplification factor is introduced to improve the support leg inverse dynamics model.

4. The motion following control method for a wearable robot according to claim 3, characterized in that, The specific processing steps to obtain the corresponding joint torque representation model include: A gait factor is introduced, wherein the gait factor includes a gait switching factor and a gait support factor; A joint torque representation model is constructed based on the improved swing leg inverse dynamics model, the improved support leg inverse dynamics model, the gait switching factor, and the gait support factor.

5. The motion following control method for a wearable robot according to claim 1, characterized in that, The specific processing steps for following and controlling the wearable robot also include: The wearable robot is controlled by using the current loop of the joint motor and the joint torque representation model.

6. A control system utilizing the motion-following control method for a wearable robot according to any one of claims 1-5, characterized in that, include: The acquisition module is used to acquire the plantar pressure of the wearer, and at the same time acquire the angle information and angular velocity information of the joint motor; The first model building module is used to build a tracking differentiator and generate corresponding angular acceleration information based on the tracking differentiator, the angle information, and the angular velocity information. At the same time, it builds an inverse dynamics control model based on the plantar pressure, the angle information, the angular velocity information, and the angular acceleration information. The second model construction module is used to obtain gait factors and process the inverse dynamics control model based on the gait factors, the angle information, the angular acceleration information and the angular velocity information to obtain the corresponding joint torque representation model. The control module is used to perform follow control on the wearable robot using the joint torque representation model, so that the wearable robot follows the wearer's movements in real time.

7. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium, which, when executed by a processor, implements the motion-following control method for a wearable robot as described in any one of claims 1 to 5.

Images