Upper limb rehabilitation robot on-demand assistance interaction control method based on potential energy field constraint

Abstract

The present application relates to a kind of upper limb rehabilitation robot on-demand auxiliary interaction control method based on potential energy field constraint, based on the upper limb rehabilitation robot of three degrees of freedom operation, preset target trajectory and build task space, construct virtual potential energy field based on rehabilitation demand in task space, based on the compliance control strategy of potential energy field, after patient deviates from target trajectory in training process, potential energy field will provide normal auxiliary force, achieve the effect of constraint training trajectory on undesired normal flutter, and using the RBF network based on greedy algorithm realizes the effect of training of different motion ability of patient on-demand auxiliary, realize the safe, stable and controllable motion of upper limb rehabilitation robot in three-dimensional space;Effectively improve the safety of patient in rehabilitation training process, ensure that patient can independently and effectively complete training task.

Description

Technical Field

[0001] This invention relates to the field of robot control technology, and more specifically, to an on-demand assisted interactive control method for an upper limb rehabilitation robot based on potential energy field constraints. Background Technology

[0002] In recent years, with the acceleration of population aging in my country, the number of patients with brain injury and neurodegenerative diseases has been increasing year by year. In particular, 80% of stroke patients experience motor dysfunction, and 45% of these patients experience loss of upper limb function, including musculoskeletal paralysis. Stroke is the third leading cause of death in China. In the field of rehabilitation, therapists repeatedly bend the affected limbs to reduce muscle spasms and stimulate neuroplasticity, enabling the brain to form new neural connections. This traditional method is time-consuming, laborious, and costly. Moreover, my country currently faces a severe shortage of rehabilitation therapists and low rehabilitation efficiency, failing to meet the increasing rehabilitation needs. Therefore, rehabilitation assistive robots have emerged as a new type of rehabilitation assistive medical device. However, for physical therapy, whether assisted by a therapist or a robot, there is a limitation: because the patient's movement guidance is entirely provided externally, the patient has almost no participation in the entire treatment process. This involuntary rehabilitation task has been identified as worthless. In cases of acute stroke, the first 3-6 months of rehabilitation almost entirely determine the patient's final level of paralysis. Furthermore, psychological factors are a crucial element during rehabilitation; therefore, active patient participation in training is essential for its effectiveness. In recent years, mirror therapy and image training have been shown to stimulate brain and neural plasticity, triggering motor, sensory, and pain awareness in different brain regions and inducing active patient participation. These therapies have been applied and practiced worldwide.

[0003] Current upper limb rehabilitation robots simulate the techniques of a therapist by repeatedly bending and flexing the arm. However, these mirror-image resistance devices can only control the input force of the other upper limb by mirroring the input force from one side. They merely replicate the mirrored force as the training target, without controlling the input force itself. When the input force on one side is too large, the resistance device on the other side will output a large acceleration or move a long distance, easily causing injury to the user's upper limb. The main reasons for this are as follows:

[0004] (1) Most rehabilitation robots are designed with a master / slave configuration of two degrees of freedom (DOF) to generate elbow flexion / extension and forearm extension / flexion. However, the end effector of this type of robot can only move on a two-dimensional plane and cannot provide three-dimensional movement for the patient's upper limbs in space. The activity space is limited and the rehabilitation task cannot be changed according to the patient's needs.

[0005] (2) Passive training in upper limb rehabilitation can stimulate the patient's motor plasticity, thereby improving the patient's motor ability. However, this passive training strictly follows the target trajectory set by the therapist, completely disregarding the patient's motor intentions, and even moving against the patient's intentions. Obviously, the patient will gradually lose active participation in passive rehabilitation training. Passive training is very ineffective in stimulating the patient's neuroplasticity. Summary of the Invention

[0006] The technical problem to be solved by this invention is how to achieve safe, stable and controllable movement of an upper limb rehabilitation robot in three-dimensional space. To overcome the shortcomings of the prior art, this invention provides an interactive control method for an upper limb rehabilitation robot based on potential energy field constraints.

[0007] This invention provides an on-demand assisted interactive control method for an upper limb rehabilitation robot based on potential energy field constraints. The method is applied to a three-degree-of-freedom upper limb rehabilitation robot. The upper limb rehabilitation robot includes a robotic arm with three rotational joints and an end effector. The three rotational joints of the robotic arm are driven to rotate by three drive motors. Each of the three rotational joints has an angle encoder for collecting the rotation angle of the joints. The end effector is equipped with a force sensor for collecting data on the force applied by the patient to the end effector during human-machine interaction. The on-demand assisted interactive control method for the upper limb rehabilitation robot includes:

[0008] Step 1: Preset the target motion trajectory of the upper limb rehabilitation robot, and uniformly sample from the target motion trajectory. One sampling point: In the formula, , is represented as the three-dimensional column vector of the end effector handle of the upper limb rehabilitation robot at the i-th point in Cartesian space; A discrete dataset representing the target motion trajectory of the end effector handle;

[0009] Step 2: Establish the forward kinematics model of the upper limb rehabilitation robot by measuring the rotation angles of the three rotary joints of the robotic arm, and calculate the position of the end effector handle in the reference coordinate system based on the forward kinematics model. The actual position on;

[0010] Step 3: Define the task space based on the target motion trajectory and calculate the sampling points in the task space. For other points in the task space The total potential energy; and the design of the negative gradient of the potential energy. Preset mapping factor and combined with mapping factor With negative gradient The normal auxiliary force in the direction of movement of the end effector handle is calculated. ;

[0011] Step 4: Based on the end effector handle in the reference coordinate system Calculate gravity compensation and friction compensation based on the position above;

[0012] Step 5: Based on the end effector handle in the reference coordinate system The position and target motion trajectory are calculated, and the tangential auxiliary force in the direction of the end effector handle's movement is applied. ;

[0013] Step 6: Based on the actual position of the end effector handle calculated in Step 2, establish a projection module to facilitate the operation of the RBF network in three-dimensional space. The projection module is used to calculate the projection of the end effector handle in the constraint direction. , project The force data collected during the human-computer interaction process is input into a Gaussian RBF network for iterative calculation of the maximum force that the patient can apply to the end effector handle. Preset auxiliary coefficient for the force applied by the patient And according to the maximum force Tangential auxiliary force and Calculation of weighted tangential auxiliary force ;

[0014] Step 7: Based on normal auxiliary force and weighted tangential auxiliary force Calculate the compensating force of the end effector handle ;

[0015] Step 8: Based on the gravity compensation and friction compensation in Step 4, and the compensation force calculated in Step 7. And Jacobi matrix Calculate the torques mapped to each rotational joint of the upper limb rehabilitation robot.

[0016] Compared with existing technologies, this application has the following advantages: Based on the construction of a three-degree-of-freedom upper limb rehabilitation robot, a target motion trajectory is preset and a task space is built. A virtual potential energy field based on rehabilitation needs is constructed in the task space. Based on the compliant control strategy of the potential energy field, the potential energy field provides a normal assist force when the patient deviates from the target motion trajectory during training. Combined with the tangential assist force, it achieves the effect of restraining undesirable normal tremors on the training trajectory. Furthermore, an RBF network based on a greedy algorithm is used to achieve the effect of assisting patients with different motor abilities on demand, realizing the safe, stable and controllable movement of the upper limb rehabilitation robot in three-dimensional space. This effectively improves the safety of patients during rehabilitation training and ensures that patients can independently and effectively complete training tasks.

[0017] In one possible implementation, step 2 involves calculating the end effector handle in the reference coordinate system based on a positive kinematics model. The actual locations on include:

[0018] ;

[0019] ;

[0020] ;

[0021] In the formula, The sensors measure the angle values ​​of the three rotary joints of the robotic arm. Indicates the first The rotation axis of the first rotary joint is related to the adjacent first joint. The length of the common normal between the rotation axes of the rotary joints; Indicates the first The first rotary joint and the adjacent first The common normal of each rotational joint and Indicates the first The first rotational joint and the second The common normal of the nth rotational joint is along the nth The distance between the rotation axes of each rotary joint. for The simplified form, for The simplified form, for The simplified form, for The simplified form.

[0022] In one possible implementation, step 3 involves defining the task space based on the target motion trajectory and calculating the sampling points within the task space. For other points in the task space The total potential energy specifically includes:

[0023] Step 301A: Calculate the sampling points in the task space according to Hooke's Law. For each point Potential energy:

[0024] ;

[0025] In the formula, Represents each sampling point The initial potential energy, and for each sampling point initial potential energy The optimization is performed using a convexity solution optimization problem, denoted as: , ,and The constraints are ;point With sampling points There is a virtual spring between the sampling points. Point The attraction is , Representing stiffness, after integration, the point... The elastic potential energy is ;

[0026] Step 302A: Calculate points using the Gaussian kernel function. arrive Potential energy weights at each sampling point :

[0027]

[0028] In the formula, The weights for the influence range of each sampling point;

[0029] Step 303A, for Normalization yields :

[0030] ;

[0031] Step 304A, and Multiplying and summing yields the points in the task space. Total potential energy :

[0032] ;

[0033] In step 3, the negative gradient of the potential energy is designed. Represented as:

[0034] ;

[0035] In step 3, the preset mapping factor and combined with mapping factor With negative gradient The normal auxiliary force in the direction of movement of the end effector handle is calculated. , represented as: .

[0036] In one possible implementation, step 4 is based on the end effector handle in the reference coordinate system. The calculation of gravity compensation and friction compensation for the position on the surface specifically includes:

[0037] The calculation of gravity compensation specifically includes:

[0038] Calculate the torque of the rotary joint:

[0039] ;

[0040] In the formula, For inertia, For the Coriolis force and centripetal force terms, This is the term related to gravity.

[0041] Substituting the robotic arm parameters into the torque calculation formula for the rotary joint yields:

[0042] ;

[0043] ;

[0044] ;

[0045] The gravity compensation for the three rotary joints are as follows: ;

[0046] The calculation method for friction compensation is as follows: The current of the three rotating joints of the upper limb rehabilitation robot during uniform operation is collected, along with the data relationship between speed and friction obtained by changing the operating speed. This speed and friction data are then introduced into the Stribek model to obtain a fitting model, i.e., friction compensation, with speed and friction as independent variables. The Stribek model is as follows: ;in, The parameters to be fitted are... These represent the velocity and friction force in the fitted data, respectively. It represents the natural logarithm.

[0047] In one possible implementation, step 5 involves calculating the tangential assist force in the direction of operation of the end effector handle. Specifically, it includes:

[0048] Calculate the actual position of the upper limb rehabilitation robot Position difference from the target position ;

[0049] position difference The input impedance controller calculates the tangential auxiliary force in the direction of travel of the end effector handle. :

[0050] ;

[0051] In the formula, The derivative of the positional difference; This indicates the stiffness of the impedance controller. This represents the damping of the impedance controller; where, In the formula, The distance traveled along the target's trajectory is represented by the distance traveled; one round trip along the target's trajectory is recorded as one cycle. For a one-way cycle; It is converted into a displacement relative to the starting point in a predefined training direction, generating the target position in three-dimensional space.

[0052] In one possible implementation, step 6 involves calculating the projection of the end effector handle in the constraint direction using a projection module. Specifically, this includes: the projection module obtaining the start and end positions of the end effector handle based on the start and end angles of the end effector handle, and obtaining a vector based on the start and end positions. The vector is obtained based on the difference between the current actual position and the starting position of the end effector handle. ,according to and Performing a vector dot product yields the projection of the end effector handle onto the constraint direction. ;

[0053] Project The force data collected during the human-computer interaction process is input into a Gaussian RBF network for iterative calculation of the maximum force that the patient can apply to the end effector handle. Specifically, it includes:

[0054] ;

[0055] ;

[0056] In the formula, Let n×1 be an n×1 vector containing radial basis functions. It includes The weight vector of the parameters to be estimated; It is the number of RBFs distributed along a straight line within the task space; Indicates the first Gaussian radial basis functions The expression is: In the formula, Evenly distributed in the task space, representing the first The center of a Gaussian RBF; Indicates the current actual location; Scalar constant representing the width of the Gaussian radial basis functions;

[0057] In the absence of a potential energy field constraint, an upper limb end-effector robot is trained to collect kinematic data and force sensor measurement data. The initial weight vector of the least squares-based RBF network is then determined using the kinematic data and force sensor measurement data.

[0058] ;

[0059] ;

[0060] ;

[0061] In the formula, It is a time series containing a Gaussian RBF network. matrix, Indicates the first time, This represents the number of nodes in a Gaussian RBF network. Indicates the end effector handle at the 1st The force sensor measurement data at any given time. It is a matrix H The false reversal; It is the initial weight vector of the Gaussian RBF network;

[0062] In the iterative training of a Gaussian RBF network, based on the initial weight vector Update the weight vector of the Gaussian RBF network using gradient descent algorithm. Specifically, it includes:

[0063] ;

[0064] ;

[0065] ;

[0066] In the formula, Indicates the learning rate. Indicates the sampling points in the task cycle. Indicates the amount of work. The average gradient over one task cycle. and These are the weight vectors for the current task cycle and the next task cycle, respectively.

[0067] In step 6, based on the predictive power Tangential auxiliary force and auxiliary coefficient Calculation of weighted tangential auxiliary force The calculation formula is:

[0068] .

[0069] In one possible implementation, the compensating force of the end effector handle in step 7 .

[0070] In one possible implementation, the Jacobian matrix in step 8 The calculation formula is:

[0071] ;

[0072] In step 8, the torques mapped to each rotational joint of the upper limb rehabilitation robot are calculated. The calculation formula is:

[0073] ;

[0074] In the formula, It is a 3×3 Jacobian matrix. Given a 3×1 matrix, we obtain It is a 3×1 matrix. Attached Figure Description

[0075] Figure 1 This is a structural diagram and coordinate system distribution diagram of the three-degree-of-freedom rehabilitation robot of the present invention;

[0076] Figure 2 This is a system block diagram of the on-demand assisted interactive control method for an upper limb rehabilitation robot based on potential energy field constraints according to the present invention.

[0077] Figure 3 To investigate the trend of the average force exerted by the subjects relative to the number of tasks, the average force of all subjects under different tasks was calculated, and a first-order linear function was used to fit the relationship between the average force and the number of tasks. Detailed Implementation

[0078] First, those skilled in the art should understand that these embodiments are merely used to explain the technical principles of the embodiments of this application and are not intended to limit the scope of protection of the embodiments of this application. Those skilled in the art can make adjustments as needed to adapt to specific application scenarios.

[0079] In the description of the embodiments of this application, it should be noted that, unless otherwise explicitly specified and limited, the terms "connected" and "linked" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms in the embodiments of this application based on the specific circumstances.

[0080] In the embodiments of this application, unless otherwise expressly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "on top of," and "over" the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.

[0081] The present application will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0082] See Figures 1-3 As shown, this application discloses an on-demand assisted interactive control method for an upper limb rehabilitation robot based on potential energy field constraints. This method is applied to a three-degree-of-freedom upper limb rehabilitation robot. The upper limb rehabilitation robot has three rotary joints, three drive motors for on-demand assisted rotation of the rotary joints, and an end effector handle. The three rotary joints are equipped with angle encoders for collecting the rotation angles of the joints. The end effector handle is equipped with a force sensor for collecting data on the force applied by the patient to the end effector handle during human-machine interaction. The on-demand assisted interactive control method for the upper limb rehabilitation robot includes:

[0083] Step 1: For the patient's rehabilitation training, preset the target motion trajectory of the upper limb rehabilitation robot, and uniformly sample from the target motion trajectory. One sampling point: In the formula, , is represented as the three-dimensional column vector of the end effector handle of the upper limb rehabilitation robot at the i-th point in Cartesian space; A discrete dataset representing the target motion trajectory of the end effector handle;

[0084] Step 2: Establish the forward kinematics model of the upper limb rehabilitation robot by measuring the rotation angles of the three rotary joints of the robotic arm, and calculate the position of the end effector handle in the reference coordinate system based on the forward kinematics model. The actual position on, such as Figure 1 As shown, the reference coordinate system in this specific embodiment Coordinate system with the first rotational joint of the upper limb rehabilitation robot The coordinate system is set at the second rotary joint. A coordinate system is set at the third rotary joint. Specifically, it includes:

[0085] ;

[0086] ;

[0087] ;

[0088] In the formula, The sensors measure the angle values ​​of the three rotary joints of the robotic arm. Indicates the first The rotation axis of the first rotary joint is related to the adjacent first joint. The length of the common normal between the rotation axes of the rotary joints, i.e., along the coordinate system. Direction, from arrive The distance of translation; Indicates the first The first rotary joint and the adjacent first The common normal of each rotational joint and Indicates the first The first rotational joint and the second The common normal of the nth rotational joint is along the nth The distance between the rotation axes of each rotary joint, that is, along the coordinate system. Direction, from arrive The distance of translation; for The simplified form, for The simplified form, for The simplified form, for The simplified form;

[0089] Step 3: Define the task space based on the target motion trajectory and calculate the sampling points in the task space. For other points in the task space The total potential energy; and the design of the negative gradient of the potential energy. Preset mapping factor and combined with mapping factor With negative gradient The normal auxiliary force in the direction of movement of the end effector handle is calculated. Among them, regarding the sampling points in the computation task space. For other points in the task space The total potential energy specifically includes:

[0090] Step 301A: Calculate the sampling points in the task space according to Hooke's Law. For each point Potential energy:

[0091] ;

[0092] In the formula, Represents each sampling point The initial potential energy, and for each sampling point initial potential energy The optimization is performed using a convexity solution optimization problem, denoted as: , ,and The constraints are ;point With sampling points There are virtual springs between them. Represented as sampling points Point The attraction This represents stiffness; the higher the stiffness, the stronger the constraint effect of the potential energy field. After integration, the point... The elastic potential energy is ;

[0093] Step 302A: Calculate points using the Gaussian kernel function. arrive Potential energy weights at each sampling point :

[0094] ;

[0095] In the formula, The weights for the influence range of each sampling point;

[0096] Step 303A, for Normalization yields :

[0097] ;

[0098] Step 304A, and Multiplying and summing yields the points in the task space. Total potential energy :

[0099] ;

[0100] Negative gradient of design potential energy Represented as:

[0101] ;

[0102] Regarding the calculation of the normal auxiliary force in the direction of movement of the end effector handle. Specifically, this includes: preset mapping factors and combined with mapping factor With negative gradient The normal auxiliary force in the direction of movement of the end effector handle is calculated. The calculation formula is expressed as: ;

[0103] Step 4: Based on the end effector handle in the reference coordinate system The calculation of gravity compensation and friction compensation is performed based on the position of the object; specifically, the calculation of gravity compensation includes:

[0104] Calculate the torque of the rotary joint:

[0105] ;

[0106] In the formula, For inertia, For the Coriolis force and centripetal force terms, This is the term related to gravity.

[0107] Substituting the robotic arm parameters into the torque calculation formula for the rotary joint yields:

[0108] ;

[0109] ;

[0110] ;

[0111] The gravity compensation for the three rotary joints are as follows: ;

[0112] The calculation method for friction compensation is as follows: The current of the drive motors of the three rotary joints of the upper limb rehabilitation robot running at a constant speed is collected, along with the data relationship between speed and friction obtained by changing the running speed. This speed and friction data are then introduced into the Stribek model to obtain a fitting model, i.e., friction compensation, with speed and friction as independent variables. The Stribek model is as follows: ;in, The parameters to be fitted are... These represent the velocity and friction force in the fitted data, respectively. Represents the natural logarithm;

[0113] Step 5: Based on the end effector handle in the reference coordinate system The position and target motion trajectory are calculated, and the tangential auxiliary force in the direction of the end effector handle's movement is applied. Specifically, it includes:

[0114] Calculate the actual position of the upper limb rehabilitation robot Position difference from the target position ;

[0115] position difference The input impedance controller calculates the tangential auxiliary force in the direction of travel of the end effector handle. :

[0116] ;

[0117] In the formula, The derivative of the positional difference; This indicates the stiffness of the impedance controller. This represents the damping of the impedance controller; where, In the formula, The distance traveled along the target's trajectory is represented by the distance traveled; one round trip along the target's trajectory is recorded as one cycle. For a one-way cycle; It is converted into displacement relative to the starting point in a predefined training direction to generate the target position in three-dimensional space;

[0118] Step 6: Based on the actual position of the end effector handle calculated in Step 2, establish a projection module to facilitate the operation of the RBF network in three-dimensional space. The projection module is used to calculate the projection of the end effector handle in the constraint direction. , project The force data collected during the human-computer interaction process is input into a Gaussian RBF network for iterative calculation of the maximum force that the patient can apply to the end effector handle. Preset auxiliary coefficient for the force applied by the patient And according to the maximum force Tangential auxiliary force and Calculation of weighted tangential auxiliary force ;

[0119] Specifically, the projection of the end effector handle in the constraint direction is calculated through the projection module. Specifically, this includes: the projection module obtaining the start and end positions of the end effector handle based on the start and end angles of the end effector handle, and obtaining a vector based on the start and end positions. The vector is obtained based on the difference between the current actual position and the starting position of the end effector handle. ,according to and Performing a vector dot product yields the projection of the end effector handle onto the constraint direction. ;

[0120] Project The force data collected during the human-computer interaction process is input into a Gaussian RBF network for iterative calculation of the maximum force that the patient can apply to the end effector handle. Specifically, it includes:

[0121] ;

[0122] ;

[0123] In the formula, Let n×1 be an n×1 vector containing radial basis functions. It includes The weight vector of the parameters to be estimated; It is the number of RBFs distributed along a straight line in the task space. In this specific embodiment, in order to balance the approximate accuracy and the computational complexity of the upper limb rehabilitation robot, the number of RBFs is set to 20. Indicates the first Gaussian radial basis functions The expression is: In the formula, Evenly distributed in the task space, representing the first The center of a Gaussian RBF; Indicates the patient's current actual location; The scalar constant representing the width of the Gaussian radial basis function is set in this specific embodiment. ; This represents the distance between the centers of the first and last Gaussian radial odd functions;

[0124] In this specific embodiment, a Gaussian RBF network fits the maximum force applied by the patient over time and is used to model the patient's capabilities. The calculation of the weight vector parameters of the Gaussian RBF network includes an initialization part and an iterative part. Since the update of the Gaussian RBF network is based on the weight vector of the previous task's RBF network, it is necessary to obtain an initial weight vector. Specifically, this includes: training the upper limb end-effector robot to collect kinematic data and force sensor measurement data without potential energy field constraints, and determining the initial weight vector of the least squares-based RBF network using the kinematic data and force sensor measurement data.

[0125] ;

[0126] ;

[0127] ;

[0128] In the formula, It is a time series containing a Gaussian RBF network. matrix, Indicates the first time, This represents the number of nodes in a Gaussian RBF network. Indicates the end effector handle at the 1st The force sensor measurement data at any given time. It is a matrix H The false reversal; It is the initial weight vector of the Gaussian RBF network;

[0129] Obtain the initial weight vector Subsequently, during the iterative training of the Gaussian RBF network, based on the initial weight vector... Update the weight vector of the Gaussian RBF network using gradient descent algorithm. Specifically, it includes:

[0130] ;

[0131] ;

[0132] ;

[0133] In the formula, Indicates the learning rate. Indicates the sampling points in the task cycle. Indicates the amount of work. The average gradient over one task cycle. and These are the weight vectors for the current task cycle and the next task cycle, respectively; where, It is a binary variable whose value is obtained by comparing the measured value from the force sensor with the estimated value from the Gaussian RBF network. Greater than This means that the output of the Gaussian RBF network is less than the maximum force the patient can exert. Therefore, the weight vector needs to be updated to increase the output of the Gaussian RBF network. Conversely, when Less than This means the patient should be able to exert more force; therefore, the control will maintain the weight vector to challenge the patient and stimulate them to exert greater active force. Since the output of the Gaussian RBF network only updates in an increasing trend, the proposed robot-assisted strategy is called a Greedy On-Demand Assist (GAAN) controller. Furthermore, if the patient occasionally reduces the force they exert due to other subjective reasons such as laxity, the controller will not misjudge a decline in their functional ability. Therefore, the Greedy On-Demand Assist control strategy in this specific embodiment builds a model of the patient's motor ability based on the maximum force exerted by the patient during rehabilitation training, rather than the instantaneous force.

[0134] Based on predictive power Tangential auxiliary force and auxiliary coefficient Calculation of weighted tangential auxiliary force The calculation formula is:

[0135] ;

[0136] Step 7: Based on normal auxiliary force and weighted tangential auxiliary force Calculate the compensating force of the end effector handle The specific calculation formula is as follows: ;

[0137] Step 8: Based on the gravity compensation and friction compensation in Step 4, and the compensation force calculated in Step 7. And Jacobi matrix Calculate the torques mapped to each rotational joint of the upper limb rehabilitation robot;

[0138] Among them, the Jacobian matrix The calculation formula is:

[0139] ;

[0140] In step 8, the torques mapped to each rotational joint of the upper limb rehabilitation robot are calculated. The calculation formula is:

[0141] ;

[0142] In the formula, It is a 3×3 Jacobian matrix; Given a 3×1 matrix, we obtain It is a 3×1 matrix.

[0143] Experimental methods:

[0144] The experiment was conducted by six healthy volunteers (4 males and 2 females), aged 24.5 ± 0.6 years and with a height of 173.0 ± 7.2 cm. All experiments were approved by the Ethics Committee of the Ningbo Institute of Materials Technology and Engineering, Chinese Academy of Sciences. Informed consent was provided by all participants. During the experiment, the volunteers lay on a bedridden stroke patient synchronous training rehabilitation robot.

[0145] Each participant was asked to complete three sets of rehabilitation training exercises. Before the training began, participants were shown the virtual environment and the operation was explained, and they were given a few minutes to familiarize themselves with the system. During the experiment, participants completed shooting tasks under the guidance of visual feedback, and the magnitude of the force applied by the participants was determined by their own will. When the participant reached the current target, the virtual person fired a bullet, causing the target ball to break, and the participant began to move towards the starting point. The target ball regenerated 2 seconds after reaching the starting point. The period of each shooting task was set to 10 seconds, and kinematic and dynamic data were recorded at a frequency of 1000 Hz during the experiment.

[0146] The experiment aimed to verify that the developed on-demand assisted controller based on potential field constraints could promote active participation of subjects. Each subject underwent three training exercises with different assistance levels (α=0, 0.5, 1) under the same impedance controller. The initial weight vector of the RBF network was set to the values ​​obtained by each subject in previous experiments. In the training exercise with α=0, the weight vector of the RBF network was updated normally according to the method in 4.2.2, but the network output did not affect the robot controller. Therefore, the control paradigm with α=0 could not provide a challenge for the subjects, while the controllers in the other groups (α=0.5, 1) could provide a challenge using the RBF network. Each training exercise consisted of 10 back-and-forth movements. Before the start of the next new exercise, the weight vector of the RBF network was updated based on the acquired data. The iteration step λ was set to 0.2.

[0147] To further analyze the relationship between participant participation and task cycle at the same challenge level, the number of tasks was used as the independent variable, and the resulting data is shown in the figure:

[0148] Figure 3 The average force of the subjects was plotted as the number of tasks progressed, along with a fitted curve based on a first-order linear function. The slope of the fitted curve represents the trend of change in the subjects' level of participation during the task. Figure 3 (a), (b), and (c) represent the motion results of endpoint AB when α=0, α=0.5, and α=1, respectively; Figure 3 (d), (e), and (f) represent the motion results of endpoint BA when α=0, α=0.5, and α=1, respectively. Figure 3 (a), (b), and (c) show that when α=0, the subjects' exertion in each task cycle remains almost unchanged (slope=-0.005). When α=0.5 or α=1, the subjects' average force increases with the number of tasks (slope=0.323, slope=0.484).

[0149] Figure 3 Similar conclusions can be drawn from the recorded reverse movements in (d), (e), and (f). This indicates that the task challenge provided by the proposed controller compels subjects to exert greater initiative in completing the training task, effectively preventing slackness.

[0150] In the description of the embodiments of this application, it should be noted that the terms "inner" and "outer" and other terms indicating direction or positional relationship are based on the direction or positional relationship shown in the drawings. This is only for the convenience of description and does not indicate or imply that the device or component must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this application.

[0151] In the description of this application, the references to terms such as "an embodiment," "some embodiments," "in this embodiment," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

Claims

1. A method for on-demand assisted interactive control of an upper limb rehabilitation robot based on potential energy field constraints, applied to a three-degree-of-freedom upper limb rehabilitation robot, wherein the upper limb rehabilitation robot has three rotary joints, drive motors that assist the rotation of the rotary joints when needed, and an end effector handle; the three rotary joints are equipped with angle encoders for collecting the rotation angles of the rotary joints; and the end effector handle is equipped with a force sensor for collecting force data applied by the patient to the end effector handle during human-computer interaction, characterized in that... The on-demand assisted interactive control method for the upper limb rehabilitation robot includes: Step 1: Preset the target motion trajectory of the upper limb rehabilitation robot, and uniformly sample from the target motion trajectory. One sampling point: In the formula, Let be the first digit of the end effector of the upper limb rehabilitation robot in Cartesian space. The three-dimensional column vector at the point; A discrete dataset representing the target motion trajectory of the end effector handle; Represents Cartesian space; Step 2: Establish the forward kinematics model of the upper limb rehabilitation robot by measuring the rotation angles of the three rotary joints of the robotic arm, and calculate the position of the end effector handle in the reference coordinate system based on the forward kinematics model. The actual position on; Step 3: Define the task space based on the target motion trajectory and calculate the sampling points in the task space. The total potential energy at all other points in the task space; and the design of the negative gradient of the potential energy. Preset mapping factor and combined with mapping factor With negative gradient The normal auxiliary force in the direction of movement of the end effector handle is calculated. ; Step 4: Based on the end effector handle in the reference coordinate system The position above is used to calculate gravity compensation and friction compensation; Step 5: Based on the end effector handle in the reference coordinate system The position and target motion trajectory are calculated, and the tangential auxiliary force in the direction of the end effector handle's movement is determined. ; Step 6: Based on the actual position of the end effector handle calculated in Step 2, establish a projection module to facilitate the operation of the RBF network in three-dimensional space. The projection module is used to calculate the projection of the end effector handle in the constraint direction. , project The force data collected during the human-computer interaction process is input into a Gaussian RBF network for iterative calculation of the maximum force that the patient can apply to the end effector handle. Preset auxiliary coefficient for the force applied to the patient And according to the maximum force Tangential auxiliary force and Calculation of weighted tangential auxiliary force ; Step 7: Based on normal auxiliary force and weighted tangential auxiliary force Calculate the compensating force of the end effector handle ; Step 8: Based on the gravity compensation and friction compensation in Step 4, and the compensation force calculated in Step 7. And Jacobi matrix Calculate the torques mapped to each rotational joint of the upper limb rehabilitation robot; Jacobian matrix. The calculation formula is: 。 2. The on-demand assisted interactive control method for an upper limb rehabilitation robot based on potential energy field constraints according to claim 1, characterized in that, In step 2, the end effector handle in the reference coordinate system is calculated based on the positive kinematics model. The actual locations on include: ； ； ； In the formula, The sensors measure the angle values ​​of the three rotary joints of the robotic arm. Indicates the first The rotation axis of the first rotary joint is related to the adjacent first joint. The length of the common normal between the rotation axes of the rotary joints; Indicates the first The first rotary joint and the adjacent first The common normal of the first and second rotational joints The first rotational joint and the second The common normal of the nth rotational joint is along the nth The distance between the rotation axes of each rotary joint. for The simplified form, for The simplified form, for The simplified form, for The simplified form.

3. The on-demand assisted interactive control method for an upper limb rehabilitation robot based on potential energy field constraints according to claim 2, characterized in that, In step 3, the task space is defined based on the target motion trajectory, and the sampling points in the task space are calculated. For other points in the task space The total potential energy specifically includes: Step 301A: Calculate the sampling points in the task space according to Hooke's Law. For each point Potential energy: ； In the formula, Represents each sampling point The initial potential energy, and for each sampling point initial potential energy The optimization is performed using a convexity solution optimization problem, denoted as: , ,and The constraints are ;point With sampling points There is a virtual spring between the sampling points. Point The attraction is , Representing stiffness, after integration, the point... The elastic potential energy is ; Step 302A: Calculate points using the Gaussian kernel function. arrive Potential energy weights at each sampling point : ； In the formula, The weights for the influence range of each sampling point; Step 303A, for Normalization yields : ； Step 304A, and Multiplying and summing yields the points in the task space. Total potential energy : ； In step 3, the negative gradient of the potential energy is designed. Represented as: ； In step 3, the preset mapping factor and combined with mapping factor With negative gradient The normal auxiliary force in the direction of movement of the end effector handle is calculated. , represented as: .

4. The on-demand assisted interactive control method for an upper limb rehabilitation robot based on potential energy field constraints according to claim 3, characterized in that, In step 4, the end effector handle is based on the reference coordinate system. The calculation of gravity compensation and friction compensation for the position on the surface specifically includes: The calculation of gravity compensation specifically includes: Calculate the torque of the rotary joint: ； Substituting the robotic arm parameters into the torque calculation formula for the rotary joint yields: ； ； ； The gravity compensation for the three rotary joints are as follows: ; The calculation method for friction compensation is as follows: The current of the three rotating joints of the upper limb rehabilitation robot during uniform operation is collected, along with the data relationship between speed and friction obtained by changing the operating speed. This speed and friction data are then introduced into the Stribek model to obtain a fitting model, i.e., friction compensation, with speed and friction as independent variables. The Stribek model is as follows: ;in, The parameters to be fitted are... These represent the velocity and friction force in the fitted data, respectively. It represents the natural logarithm.

5. The on-demand assisted interactive control method for an upper limb rehabilitation robot based on potential energy field constraints according to claim 4, characterized in that, In step 5, the tangential assist force in the direction of operation of the end effector handle is calculated. Specifically, it includes: Calculate the actual position of the upper limb rehabilitation robot Position difference from the target position ; position difference The input impedance controller calculates the tangential auxiliary force in the direction of travel of the end effector handle. : ； In the formula, The derivative of the positional difference; This indicates the stiffness of the impedance controller. This represents the damping of the impedance controller; where, In the formula, The distance traveled along the target's trajectory is represented by the distance traveled; one round trip along the target's trajectory is recorded as one cycle. For a one-way cycle; It is converted into a displacement relative to the starting point in a predefined training direction, generating the target position in three-dimensional space.

6. The on-demand assisted interactive control method for an upper limb rehabilitation robot based on potential energy field constraints according to claim 5, characterized in that, In step 6, the projection of the end effector handle in the constraint direction is calculated by the projection module. Specifically, this includes: the projection module obtaining the start and end positions of the end effector handle based on the start and end angles of the end effector handle, and obtaining a vector based on the start and end positions. The vector is obtained based on the difference between the current actual position and the starting position of the end effector handle. ,according to and Performing a vector dot product yields the projection of the end effector handle onto the constraint direction. ; Project The force data collected during the human-computer interaction process is input into a Gaussian RBF network for iterative calculation of the maximum force that the patient can apply to the end effector handle. Specifically, it includes: ； ； In the formula, Let n×1 be an n×1 vector containing radial basis functions. It includes The weight vector of the parameters to be estimated; It is the number of RBFs distributed along a straight line within the task space; Indicates the first Gaussian radial basis functions The expression is: In the formula, Evenly distributed in the task space, representing the first The center of a Gaussian RBF; Indicates the current actual location; Scalar constant representing the width of the Gaussian radial basis functions; In the absence of a potential energy field constraint, an upper limb end-effector robot is trained to collect kinematic data and force sensor measurement data. The initial weight vector of the least squares-based RBF network is then determined using the kinematic data and force sensor measurement data. ； ； ； In the formula, It is a time series containing a Gaussian RBF network. matrix, Indicates the first time, This represents the number of nodes in a Gaussian RBF network. Indicates the end effector handle at the 1st The force sensor measurement data at any given time. It is a matrix H The false reversal; It is the initial weight vector of the Gaussian RBF network; In the iterative training of a Gaussian RBF network, based on the initial weight vector Update the weight vector of the Gaussian RBF network using gradient descent algorithm. Specifically, it includes: ； ； ； In the formula, Indicates the learning rate. Indicates the sampling points in the task cycle. Indicates the amount of work. The average gradient over one task cycle. and These are the weight vectors for the current task cycle and the next task cycle, respectively. In step 6, based on the predictive power Tangential auxiliary force and auxiliary coefficient Calculation of weighted tangential auxiliary force The calculation formula is: 。 7. The on-demand assisted interactive control method for an upper limb rehabilitation robot based on potential energy field constraints according to claim 6, characterized in that, The compensating force of the end effector handle in step 7 .

8. The on-demand assisted interactive control method for an upper limb rehabilitation robot based on potential energy field constraints according to claim 7, characterized in that, In step 8, the torques mapped to each rotational joint of the upper limb rehabilitation robot are calculated. The calculation formula is: ； In the formula, It is a 3×3 Jacobian matrix. Given a 3×1 matrix, we obtain It is a 3×1 matrix.

Images