Double upper limb robot interaction method and system based on motion performance self-adaptive assistance

Abstract

The invention belongs to the technical field of training robots, and particularly discloses a double upper limb robot interaction method and system based on motion performance adaptive assistance, and the method comprises the steps: S1, obtaining interaction data of each user during training; s2, generating a motion intention parameter, dynamically adjusting an impedance compensation parameter based on the motion intention parameter, and generating a variable impedance compensation force; s3, quantifying the exercise performance of the user according to the exercise state data of the user in the task cycle, and generating an exercise performance score; s4, obtaining the difference degree of the exercise performance between the users; s5, the ideal auxiliary force provided for each user in the next training period is dynamically adjusted; s6, generating a self-adaptive auxiliary force pointing to the training target; and S7, the real-time interaction force, the variable impedance compensation force and the self-adaptive auxiliary force of the corresponding user are fused to serve as input of a training robot admittance control scheme corresponding to the user, the expected speed and motion instruction of the mechanical arm are generated, and the mechanical arm is controlled to execute corresponding operation.

Description

Technical Field

[0001] The present invention belongs to the technical field of training robots, and in particular relates to a two-person upper limb robot interaction method and system based on adaptive assistance of motion performance. Background Art

[0002] Stroke, commonly known as stroke, is divided into two types: ischemic stroke and hemorrhagic stroke. It is a disease caused by damage to cerebral blood vessels due to various reasons, resulting in focal or global brain tissue damage. Among them, more than 80% of stroke patients have their upper limb motor function restricted to some extent, which will have a serious negative impact on the patient's quality of life; and training is the key to recovering limb function. Compared with traditional single-person training, two-person interactive training can utilize social interaction and competition mechanisms to stimulate patients' motivation and participation, which may bring better results.

[0003] However, in two-person interactive training, patients often have different levels of athletic ability in terms of muscle strength, coordination, etc. If the assistance or training difficulty provided by the system is the same for all users, users with stronger athletic abilities will find it too simple and lacking in challenge, while users with weaker athletic abilities will find it difficult and easily frustrated, resulting in an imbalance in training intensity and limiting training effectiveness and user participation.

[0004] Existing technologies often use preset difficulty or manual adjustment to solve the problem of user ability differences in two-person training. However, this type of adjustment method is difficult to adapt to the user's status in real time and cannot be adjusted in real time according to user needs.

[0005] At the level of individual human-computer interaction, especially for patients who require robot assistance, the smoothness of interaction and the accuracy of task execution are also very important. Although traditional admittance control can provide a certain degree of smoothness, it is unstable under low damping and lacks smoothness under high damping. Fixed interaction impedance is difficult to meet the user's dynamic needs for smoothness and stability under different motion tasks. How to dynamically adjust the "feel" of the interaction according to the user's intention while ensuring system stability, so that low-speed fine operations are more stable and accurate, and high-speed approach movements are smoother and more labor-saving, is also the key to improving training experience and results. Summary of the Invention

[0006] In order to solve the above technical problems, the present invention provides a two-person upper limb robot interaction method and system based on motion performance adaptive assistance, which can intelligently balance the training intensity between users and provide real-time dynamic optimization interaction for individual users.

[0007] Based on the above objectives, the present invention is achieved through the following technical solutions:

[0008] A first aspect of the present invention provides a two-person upper limb robot interaction method based on adaptive assistance of motion performance, comprising the following steps:

[0009] S1. Obtain the interaction data of the robot arm used by each user during training; the interaction data includes the real-time interaction force F obtained from the force sensor at the end of the robot arm. ext and the motion state data of the end effector of the robotic arm; the motion state data includes the current position X and speed V.

[0010] S2. For each user, based on the real-time interaction force F ext and instantaneous rate of change of interaction force Generate a motion intention parameter MA, dynamically adjust the user's impedance compensation parameter based on the motion intention parameter MA, and generate the user's variable impedance compensation force F c The impedance compensation parameters include compensation stiffness diagonal element K ct and the compensating damping diagonal element B ct .

[0011] S3. After completing one or more training task cycles, for each user, quantify their sports performance based on their sports status data during the task cycle and generate a sports performance score TCF. n ; Movement state data includes movement time t to complete the task n and average speed

[0012] S4, based on the two users' sports performance score TCF1 n ,TCF2 n , obtain the difference E in sports performance between users n , E n =|TCF1 n -TCF2 n |.

[0013] S5. According to the difference E n , based on the adaptive adjustment rules, dynamically adjust the ideal auxiliary force F provided to each user i in the next training cycle ideal,i,n+1 .

[0014] S6, based on the ideal auxiliary force F after dynamic adjustment ideal,i,n+1 and the unit direction vector u pointing to the training target target , generating an adaptive auxiliary force F pointing to the training target assist,i,n+1 , F assist,i,n+1 =F ideal,i,n+1 ·u target .

[0015] S7. Fusion of the real-time interaction force F corresponding to user i ext,i, variable impedance compensation force F c,i and adaptive assist force F assist,i,n+1 , the total input force F is obtained by vector superposition input,i =F ext,i +F c,i +F assist,i,n+1 , F input,i As the input of the user's corresponding training robot admittance control solution, the desired speed and motion instructions of the robotic arm are generated to control the robotic arm to perform the corresponding operation.

[0016] According to the above-mentioned two-person upper limb robot interaction method based on adaptive assistance of motion performance, preferably, in step S2, the step of generating the motion intention parameter MA includes:

[0017] S201, real-time interactive force F ext Perform threshold processing and limiting, set the dead zone threshold F ext_z , and obtain the effective interaction force F extL ;

[0018] F extL =sgn(F ext )·max(0,|F ext |-F ext_z ).

[0019] S202, instantaneous rate of change of interaction force Perform threshold processing and limiting, set dead zone threshold Get the instantaneous rate of change of effective interaction force

[0020]

[0021] S203, generating movement intention parameter MA:

[0022]

[0023] Among them, F extL is the effective interaction force, F ext For real-time interaction, F ext_z is the set real-time interaction force dead zone threshold,

[0024] is the instantaneous rate of change of the effective interaction force; is the instantaneous rate of change of the real-time interactive force, is the set dead zone threshold of the instantaneous change rate of the real-time interaction force, F ext_max is the preset maximum interaction force, is the instantaneous rate of change of the maximum interaction force, a is the influence factor, and proj(u,v) represents the scalar value of the projection of vector u on vector v.

[0025] According to the above-mentioned two-person upper limb robot interaction method based on adaptive assistance of sports performance, preferably, in step S2, the step of generating a variable impedance compensation force includes:

[0026] S211. Obtain the compensation stiffness diagonal element K according to the motion intention parameter ct :

[0027] K ct =K ct_max ·0.5·(1+tanh(b·(MA-0.5)));

[0028] Among them, K ct is the compensation stiffness diagonal element, MA is the motion intention parameter, K ct_max is the preset maximum compensation stiffness, and b is the adjustment parameter.

[0029] S212. Obtain the compensation damping diagonal element B according to the compensation stiffness diagonal element. ct :

[0030]

[0031] Among them, B ct To compensate for the damping diagonal elements, is the preset weighting factor.

[0032] S213, generate diagonal compensation stiffness matrix K c and the diagonal compensation damping matrix B c ;Take three dimensions as an example;

[0033] K c =diag(K ct ,K ct ,K ct );B c =diag(B ct ,B ct ,B ct );

[0034] Among them, K c is the diagonal compensation stiffness matrix, B c is the diagonal compensation damping matrix.

[0035] S214: Generate a variable impedance compensation force F based on the real-time position and speed of the user's end effector. c :

[0036] F c =-(B c V+K c X);

[0037] Among them, F cis the variable impedance compensation force, and its direction is opposite to the direction of the human body impedance force. X is the real-time position of the user's end effector, and V is the movement speed of the end effector.

[0038] According to the above-mentioned two-person upper limb robot interaction method based on adaptive assistance of sports performance, preferably, in step S3, the step of generating a sports performance score includes:

[0039] S301, in the task cycle, according to the control period ΔT and the number of control frames num consumed to complete the task n , get the exercise time t that the user spends to complete the nth training task n :

[0040] t n =ΔT×num n ;

[0041] Among them, t n The time it takes to complete the current training task, ΔT is the control cycle, num n It is the number of control frames consumed to complete the task; the task cycle refers to the time period that the user takes to complete a complete training task. Its start and end are determined by the logic of the training task. A task cycle can contain hundreds or thousands of control cycles.

[0042] S302: Obtain the average movement speed of the user during the process of completing the nth training task

[0043]

[0044] in, is the user's motion speed of the i-th control frame during the n-th training task, are the velocity vectors of the i-th control frame in the x, y, and z directions during the n-th training task at the end of the robotic arm, The average movement speed of the user when completing the nth training task.

[0045] S303, the movement time t n and average speed As input, a fuzzy logic system is used to process and generate a numerical sports performance score TCF. n .

[0046] According to the above-mentioned two-person upper limb robot interaction method based on adaptive assistance of sports performance, preferably, the fuzzy logic system includes fuzzification, fuzzy reasoning based on preset fuzzy rules and defuzzification, and its specific steps include:

[0047] S311, fuzzification: the input motion time t n and average speed After normalization, it is transformed into a fuzzy linguistic value with membership.

[0048] S312, fuzzy reasoning: The fuzzy language values ​​of the movement time and average speed are reasoned based on the preset fuzzy rules to obtain the output TCF n fuzzy set.

[0049] S313, defuzzification: converting the fuzzy set result obtained by fuzzy inference output into an accurate value y;

[0050]

[0051] Where μ(x) is the membership value of the fuzzy set at x, a and b are the lower and upper limits of the integral respectively. Introducing the scaling factor, we can obtain the training performance score TCF of the nth task. n The actual value of:

[0052] TCF n =f scale y;

[0053] Among them, f scale is the scaling factor.

[0054] According to the above-mentioned two-person upper limb robot interaction method based on adaptive assistance of sports performance, preferably, in step S5, the adaptive adjustment rule is based on the difference degree E n =|TCF1 n -TCF2 n |Adaptive step size function, step size α n It's E n Function, TCF1 n ,TCF2 n The specific steps for scoring the athletic performance of the two users and using the adaptive step length function to dynamically adjust the ideal assist force for each user in the next training cycle are as follows:

[0055] S501, based on the difference E n The piecewise function determines the adaptive step size α n :

[0056]

[0057] Among them, α n is the adaptive step size, is the difference E n The segmentation point, α l ,...,α q is the corresponding step value.

[0058] S502, according to the difference E n, adaptive step size α n and the ideal assist force current value F ideal,n Dynamically adjust the ideal assist force F provided to each user in round n+1 ideal,n+1 , the adjustment rules are:

[0059]

[0060] Among them, F ideal,n+1 is the ideal auxiliary force for the n+1th round, F ideal,n is the ideal auxiliary force for the nth round.

[0061] According to the above-mentioned two-person upper limb robot interaction method based on adaptive assistance of sports performance, preferably, in step S6, the specific steps of generating the adaptive assisting force directed toward the training target are:

[0062] Based on the dynamically adjusted ideal assist force F ideal,i,n+1 The adaptive auxiliary force is the adjusted ideal auxiliary force multiplied by the target direction unit vector to generate an adaptive auxiliary force pointing to the training target. The direction of the auxiliary force always points to the current target point P. ad :

[0063]

[0064] Among them, F assist,n+1 is the adaptive auxiliary force of the n+1th round, F ideal,n+1 is the ideal auxiliary force for the n+1th round, P ad Represents the vector pointing to the target point.

[0065] According to the above-mentioned two-person upper limb robot interaction method based on adaptive assistance of motion performance, preferably, in step S7, the step of generating the desired speed of the robot arm by the admittance controller using the total input force is specifically: substituting the interaction force of the end of the robot arm itself, the variable impedance compensation force and the adaptive auxiliary force into the admittance control model of the admittance controller to obtain the desired speed of the robot arm; the admittance control model is:

[0066]

[0067] Among them, M is the inertia matrix; D is the damping matrix; K is the stiffness matrix; F input,i is the interaction force F at the end of the robotic arm itself ext,i , variable impedance compensation force F c,i and adaptive assist force F assist,i,n+1 The combined force.

[0068] Δx=x0-x d ;

[0069] Among them, x d , is the desired position, velocity and acceleration of the robot, and x0, The position, velocity, and acceleration values ​​that the robot theoretically needs to track when the external force is zero.

[0070] The second aspect of the present invention provides a two-person upper limb robot interaction system based on adaptive assistance of motion performance, including an interaction data and motion state acquisition module, a variable impedance compensation force determination module, a patient motion performance quantitative evaluation module, an adaptive assistance force adjustment module, and a robotic arm execution module.

[0071] The interaction data and motion state acquisition module is used to obtain the current interaction data and current motion state of each user; the interaction data includes the interaction force obtained from the corresponding force sensor, and the real-time position and speed of the corresponding robotic arm end in its own coordinate system.

[0072] The variable impedance compensation force determination module is used to calculate the motion intention parameters based on the real-time interaction force of the patient user and the instantaneous change rate of the interaction force, and dynamically adjust the impedance compensation parameters based on the motion intention parameters to calculate and generate the variable impedance compensation force; the input end of the variable impedance compensation force determination module is connected to the output end of the interaction data and motion state acquisition module.

[0073] The patient sports performance quantitative evaluation module is used to quantify the patient user's sports performance based on the sports status data and generate a sports performance score; the input end of the patient sports performance quantitative evaluation module is connected to the output end of the interaction data and sports status acquisition module.

[0074] The adaptive auxiliary force adjustment module is used to calculate the difference between the sports performance scores of two users, and dynamically adjust the ideal auxiliary force provided to each user in the next training cycle according to the difference through preset adaptive adjustment rules, thereby generating an adaptive auxiliary force directed towards the training goal; the input end of the adaptive auxiliary force adjustment module is connected to the output end of the patient sports performance quantitative evaluation module.

[0075] The input end of the robotic arm execution module is respectively connected to the output end of the variable impedance compensation force determination module and the output end of the adaptive auxiliary force adjustment module; the robotic arm execution module is used to fuse the interaction force, variable impedance compensation force and adaptive auxiliary force of the corresponding user as the input force of the user's corresponding training robot admittance control scheme to generate the desired speed and motion instructions of the robotic arm and control the robotic arm to perform corresponding operations; and execute any step of the two-person upper limb robot interaction method based on adaptive assistance of motion performance as described in the first aspect.

[0076] According to the above-mentioned two-person upper limb robot interaction system based on adaptive assistance of motion performance, preferably, the interaction data and motion state acquisition module includes an interaction force acquisition unit and a motion state acquisition unit; the interaction force acquisition unit acquires the interaction force at the end of the robotic arm through a six-dimensional force sensor; the motion state acquisition unit converts the joint angle of the robotic arm into end position information, converts the joint angular velocity of the robotic arm into the end motion velocity through forward kinematics, and records the number of control frames of the robotic arm in the process of completing the task.

[0077] The variable impedance compensation force determination module includes a motion intention parameter determination unit and a variable impedance compensation force determination unit; the motion intention parameter determination unit is used to calculate and determine the user's motion intention parameters; the variable impedance compensation force determination unit is used to dynamically adjust the impedance compensation parameters according to the motion intention parameters and generate a variable impedance compensation force.

[0078] The patient's sports performance quantitative evaluation module includes a sports performance quantification unit and a sports performance evaluation unit; the sports performance quantification unit is used to quantify the performance based on the sports data; the sports performance evaluation unit is used to use fuzzy logic system reasoning to obtain a sports performance score based on the sports time and average speed obtained after the sports performance is quantified.

[0079] The adaptive auxiliary force adjustment module includes an ideal auxiliary force determination unit and an adaptive auxiliary force determination unit; the ideal auxiliary force determination unit dynamically adjusts the ideal auxiliary force provided to each user in the next training cycle based on the difference between the sports performance scores of two users; the adaptive auxiliary force determination unit is used to determine the final adaptive auxiliary force.

[0080] The robotic arm execution module vector-superimposes the interaction force, variable impedance compensation force and adaptive auxiliary force of the robotic arm end to obtain the total input force, substitutes the total input force into the admittance control model of the admittance controller, generates the desired speed and motion instructions of the robotic arm, and controls the robotic arm to perform corresponding operations.

[0081] Compared with the prior art, the present invention has the following beneficial effects:

[0082] 1. The present invention uses variable impedance compensation technology to dynamically adjust the flexibility of robot interaction according to each user's instantaneous interaction intention. When the user needs to perform low-speed and precise operations, the interaction force and the instantaneous rate of change of the interaction force are reduced, and the compensation is reduced, making the device easier to control accurately. When the user needs to move at high speed, the interaction force and the instantaneous rate of change of the interaction force are increased, and the compensation is increased, making the device easier to drive, improving the naturalness and comfort of the interaction, and effectively taking into account both low-speed precision and high-speed flexibility.

[0083] 2. The present invention achieves adaptive dynamic adjustment of the auxiliary force by quantifying the user's athletic performance in real time and calculating the differences between users, effectively balancing the training intensity and competition level between users with different athletic abilities, avoiding the situation where the strong are too easy and the weak are too difficult, and improving the fairness, challenge and fun of two-person training.

[0084] 3. The present invention integrates the synergistic effects of variable impedance compensation and adaptive assistance. Variable impedance compensation focuses on optimizing the individual interactive experience at the instantaneous level, while the adaptive assistance strategy focuses on the balance between users at the task cycle level. Combining the two ensures that the training intensity can not only ensure fair competition and intensity balance among users, but also provide a personalized, comfortable and smooth interactive experience. This comprehensively improves the performance and user experience of the two-person training system, enhances the attractiveness of the treatment process, improves the patient's active participation, better promotes the remodeling of damaged motor function areas in the brain, improves the efficiency of upper limb motor function training in stroke patients, and enables patients to regain some ability to take care of themselves in daily life.

[0085] 4. The present invention can adaptively adjust the assistance level according to the user's athletic performance, and integrate individualized variable impedance adjustment to optimize the interactive experience. It can intelligently balance the training intensity among users, provide real-time dynamically optimized interactive experience for individual users, and improve the fairness, participation and comfort of two-person rehabilitation training.

[0086] 5. The present invention can dynamically adjust the assistance level according to the user's actual athletic performance, and can balance the competitive level between users of different abilities, so that both users can train in an appropriately challenging environment, meeting the training needs of different users.

[0087] 6. The present invention designs competitive interactive tasks suitable for two-person training. It introduces kinematic evaluation indicators to quantify each user's athletic performance. It uses a fuzzy logic system to comprehensively evaluate user athletic performance and generate a training performance score. It calculates the difference between the two users' performance scores and dynamically adjusts the ideal auxiliary force applied to each user through an adaptive step-size piecewise function to balance the competitive level of both parties. At the same time, the system monitors each user's interactive force and movement intention in real time, and dynamically adjusts the interactive impedance characteristics of the robot end through variable impedance compensation to ensure that individual users can obtain appropriate flexibility and control in different motion states. By balancing the ability differences between users and optimizing the individual interactive experience, the present invention improves the fairness, participation, and comfort of two-person training. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] Figure 1 This is a schematic diagram of the process of Example 1 of the present invention;

[0089] Figure 2 This is a schematic diagram of the control principle of Example 1 of the present invention;

[0090] Figure 3 This is a schematic diagram of generating a target point in Example 1 of the present invention;

[0091] Figure 4 Schematic diagram of the fuzzy logic system in Example 1 of the present invention;

[0092] Figure 5 Schematic diagram of the membership function of input and output variables in Example 1 of the present invention;

[0093] Figure 6 Schematic diagram of the input / output surface of fuzzy reasoning in Example 1 of the present invention;

[0094] Figure 7 Schematic diagram of the model of the admittance controller in Example 1 of the present invention;

[0095] Figure 8 This is a schematic diagram of the process of Example 2 of the present invention;

[0096] Among them, there are an interactive data and motion state acquisition module 801, a variable impedance compensation force determination module 802, a patient motion performance quantitative evaluation module 803, an adaptive auxiliary force adjustment module 804, and a robotic arm execution module 805. DETAILED DESCRIPTION

[0097] The present invention is further described in detail below through specific examples, but the scope of the present invention is not limited thereto.

[0098] Example 1

[0099] A two-person upper limb robot interaction method based on adaptive assistance of motion performance, such as Figure 1-2 As shown, the following steps are included:

[0100] S1. Obtain the interaction data of the robot arm used by each user during training; the interaction data includes the real-time interaction force F obtained from the force sensor at the end of the robot arm. ext and the motion state data of the end effector of the robotic arm; the motion state data includes the current position X and speed V.

[0101] The current real-time position, movement speed and end-to-end interaction force of the robotic arms used by the two users are obtained respectively; among them, the three-dimensional coordinate value of the real-time position is obtained by obtaining the robotic arm joint angle through forward kinematics calculation, the movement speed is obtained by calculating the displacement per unit time, and the interaction force is measured by the six-dimensional force sensor at the end of the robotic arm; the real-time position and interaction force of the robotic arm end are calculated relative to its own coordinate system; among them, the robotic arm's own coordinate system is the robotic arm base coordinate system established with the center point of the robotic arm base as the origin.

[0102] S2. For each user, based on the real-time interaction force F ext and instantaneous rate of change of interaction force Generate a motion intention parameter MA, dynamically adjust the user's impedance compensation parameter based on the motion intention parameter MA, and generate the user's variable impedance compensation force F c The impedance compensation parameters include compensation stiffness diagonal element K ct and the compensating damping diagonal element B ct .

[0103] The steps of generating the movement intention parameter MA include:

[0104] S201, real-time interactive force F ext Perform threshold processing and limiting, set the dead zone threshold F ext_z , and obtain the effective interaction force F extL ;

[0105] F extL =sgn(F ext )·max(0,|F ext |-F ext_z ).

[0106] S202, instantaneous rate of change of interaction force Perform threshold processing and limiting, set dead zone threshold Get the instantaneous rate of change of effective interaction force

[0107]

[0108] S203, generating movement intention parameter MA:

[0109]

[0110] Among them, F extL is the effective interaction force, F ext For real-time interaction, F ext_z is the set real-time interaction force dead zone threshold, is the instantaneous rate of change of the effective interaction force; is the instantaneous rate of change of the real-time interactive force, is the set dead zone threshold of the instantaneous change rate of the real-time interaction force, F ext_max is the preset maximum interaction force, is the instantaneous rate of change of the maximum interaction force, a is the influence factor, and proj(u,v) represents the scalar value of the projection of vector u on vector v.

[0111] The steps for generating variable impedance compensation force include:

[0112] S211, according to the motion intention parameter, obtain the compensation stiffness diagonal element K through the hyperbolic tangent function ct :

[0113] K ct =K ct_max ·0.5·(1+tanh(b·(MA-0.5)));

[0114] Among them, K ct is the compensation stiffness diagonal element, MA is the motion intention parameter, K ct_max is the preset maximum compensation stiffness, and b is the adjustment parameter.

[0115] S212. Obtain the compensation damping diagonal element B according to the compensation stiffness diagonal element. ct :

[0116]

[0117] Among them, B ct To compensate for the damping diagonal elements, is the preset weighting factor.

[0118] S213, generate diagonal compensation stiffness matrix K c and the diagonal compensation damping matrix B c ;

[0119] K c =diag(K ct ,K ct ,K ct );B c =diag(B ct ,B ct ,B ct );

[0120] Among them, K c is the diagonal compensation stiffness matrix, B c is the diagonal compensation damping matrix.

[0121] S214: Generate variable impedance compensation force F based on the real-time position X and velocity V of the user's end effector c :

[0122] F c =-(B c V+K c X);

[0123] Among them, F c is the variable impedance compensation force, and its direction is opposite to the direction of the human body impedance force. X is the real-time position of the user's end effector, and V is the movement speed of the end effector.

[0124] The influencing factor a is the influence of the interaction force change rate on the motion intention parameter MA, and its value is 0.5; the compensation stiffness diagonal element K ct As the speed of change of the movement intention parameter MA is changed, the value is taken as 0.7; the above values ​​are the parameters used in this embodiment, and in actual use, they can be determined according to the user's own conditions and needs.

[0125] like Figure 3 As shown, the participant controls the robot's end effector to move the green or blue cursor along the target motion direction to the red target point position; when the distance between the moving cursor and the target point is less than d, it is considered to have reached the target point; after both participants have reached the target point, a new target point is generated to continue the next round of tasks; in order to eliminate the participants' prejudgment, the target point is generated randomly.

[0126] When the coordinate positions of two participants do not coincide, e.g. Figure 3 As shown in task a, the target point is randomly generated on the perpendicular line between the two points;

[0127]

[0128] At this time, the target point of the two participants is the same, (x t ,y t ) is a randomly generated target point, (x1, y1) is the current coordinate position of user 1, (x2, y2) is the current coordinate position of user 2, and t is a random number.

[0129] When the coordinate positions of two participants coincide, e.g. Figure 3 As shown in task b, the target point is generated on a circle with the point as the center and an arbitrary length as the radius;

[0130]

[0131] At this time, the target points of the two participants are different, (x t1 ,y t1 ) and (x t2 ,y t2 ) are randomly generated target points for user 1 and user 2, (x O ,y O ) is the current overlapping coordinate position of the two participants, θ1 and θ2 are two different random angles, expressed in radians, ranging from 0 to 2π, and r is the radius of the circle.

[0132] This random generation method ensures that the ideal trajectory lengths of the two participants remain consistent in each training session, ensuring the fairness of the competitive task.

[0133] like Figure 3The two conditions of tasks a and b alternated throughout the competitive interaction task, and all randomly generated target points were located within the safe area of ​​the selected range of motion and maximized the range of motion of the participants' upper limbs as much as possible.

[0134] S3. After completing one or more training task cycles, for each user, quantify their sports performance based on their sports status data during the task cycle and generate a sports performance score TCF. n ; Movement state data includes movement time t to complete the task n and average speed

[0135] The steps to generate a performance score include:

[0136] S301, in the task cycle, according to the control period ΔT and the number of control frames num consumed to complete the task n , get the exercise time t that the user spends to complete the nth training task n :

[0137] t n =ΔT×num n ;

[0138] Among them, t n The time it takes to complete the current training task, ΔT is the control cycle, num n It is the number of control frames consumed to complete the task.

[0139] S302: The velocity vectors of the end of the robotic arm used by the i-th user in the x, y, and z directions are respectively Calculate the average movement speed of the user when completing the nth training task

[0140]

[0141] in, is the user's motion speed of the i-th control frame during the n-th training task, are the velocity vectors of the i-th control frame in the x, y, and z directions during the n-th training task at the end of the robotic arm, The average movement speed of the user when completing the nth training task.

[0142] S303, the movement time t n and average speed As input, a fuzzy logic system is used for processing. The fuzzy logic system includes fuzzification, fuzzy reasoning and defuzzification. Its process is as follows Figure 4 , and finally output the numerical sports performance score TCF n .

[0143] The fuzzy logic system includes fuzzification, fuzzy reasoning based on preset fuzzy rules, and defuzzification, and its specific steps include:

[0144] S311, fuzzification: the input motion time t n and average speed After normalization, according to Figure 5 The membership function shown is converted into fuzzy language values ​​with membership, such as: time - bad (B) / medium (M) / good (G), speed - bad (B) / medium (M) / good (G).

[0145] Fuzzification refers to the process of calculating the membership degree of each input variable, such as Figure 5 As shown, the continuous movement time t of the subject in the nth training task is n and the mean speed The two kinematic evaluation indicators are used as inputs of the fuzzy strategy. In addition, in order to define the linguistic variables, the input variables t need to be scaled by a scaling factor. n and The range of variation is converted into a normalized domain. In this embodiment, the input domain and the output domain are both set to [0, 1]. Each input variable consists of three fuzzy linguistic values ​​B (bad), M (medium), and G (good); and the output variable consists of five fuzzy linguistic values ​​L (low), CL (lower), M (medium), CH (higher), and H (high).

[0146] S312, fuzzy reasoning: The fuzzy language values ​​of the movement time and average speed are calculated based on the preset fuzzy rules shown in Table 1 according to “IF t n is AAND is B THEN TCF n is C” and the output TCF is obtained. n fuzzy set.

[0147] Fuzzy reasoning refers to the process of obtaining output language variables through reasoning based on fuzzy rules. This embodiment uses the fuzzy language "IF...AND...THEN" and relies on the feedback data from the subjects after a series of tests to establish a mapping relationship between input and output variables to obtain fuzzy rules. As shown in Table 1, Table 1 lists all combinations of fuzzy input variables and corresponding fuzzy output variables, and obtains the corresponding input / output surface diagram, as shown in Table 1. Figure 6 shown.

[0148] Table 1: Fuzzy rules table;

[0149]

[0150] S313, Defuzzification: Convert the fuzzy set result obtained by fuzzy inference into an accurate value y; use the centroid method to obtain the accurate value y of the output variable:

[0151]

[0152] Where μ(x) is the membership value of the fuzzy set at x, a and b are the lower and upper limits of the integral respectively. By introducing the scaling factor f scale , get the nth task training performance score TCF n The actual value of

[0153] TCF n =f scale y;

[0154] Among them, f scale is the scaling factor, and y is the exact value output by the fuzzy comprehensive evaluation.

[0155] S4, based on the two users' sports performance score TCF1 n ,TCF2 n , calculate the difference between the two, using formula E n =|TCF1 n -TCF2 n |, and get the difference E between the two n ; Among them, TCF1 n represents the sports performance score of user 1, TCF2 n represents the sports performance score of user 2.

[0156] S5. According to the difference E n , based on the preset adaptive adjustment rules, dynamically adjust the ideal auxiliary force F provided to each user i in the next training cycle ideal,i,n+1 .

[0157] Based on the difference E n , the ideal auxiliary force F provided ideal,n Dynamic iterative adjustments are made to gradually balance the differences in exercise ability between patients so that they can maintain a relatively balanced competitive level during training. Specifically, in the nth training session, if user 1’s exercise performance score TCF1 is n Greater than subject 2TCF2 n Therefore, in order to balance the competitive levels of the two subjects, the level of assistance provided to User 1 will be reduced in the next round of training to encourage User 1 to rely more on his own strength for exercise; while the level of assistance provided to Subject 2 will be increased accordingly to help him narrow the ability gap.

[0158] The specific steps for using the adaptive step size function to dynamically adjust the ideal assist force for each user in the next training cycle are:

[0159] S501, based on the difference E n The piecewise function determines the adaptive step size α n :

[0160]

[0161] Among them, α n is the adaptive step size, is the difference E n The segmentation point, α l ,...,α q is the corresponding step value; it is worth noting that the smaller α n The number of training times will be increased to achieve a similar level of competition, which will affect the training efficiency; while a larger α n The auxiliary force provided by the system will change too quickly and may produce oscillations, which will reduce the patient's comfort and safety during training; therefore, an adaptive step length piecewise function based on difference is used to make corresponding changes according to the difference in the sports performance scores of the interacting parties. When the difference is large, a larger step length is used to quickly balance the ability gap, and when the difference decreases, the step length is automatically reduced to achieve smoother auxiliary adjustment.

[0162] S502, according to the difference E n , adaptive step size α n and the ideal assist force current value F ideal,n Dynamically adjust the ideal assist force F provided to each user in round n+1 ideal,n+1 , the adjustment rules are:

[0163]

[0164] Among them, F ideal,n+1 is the ideal auxiliary force for the n+1th round, F ideal,n is the ideal auxiliary force for the nth round.

[0165] S6, based on the ideal auxiliary force F after dynamic adjustment ideal,i,n+1 and the training target position, generating an adaptive auxiliary force directed toward the training target.

[0166]

[0167] Among them, F assist,n+1 is the adaptive auxiliary force of the n+1th round, F ideal,n+1 is the ideal auxiliary force for the n+1th round, P ad Represents the vector pointing to the target point.

[0168] S7. Convert the sum of the current interaction force, the variable impedance compensation force, and the adaptive auxiliary force into a desired speed, and then control the robotic arm to perform a corresponding operation according to the desired speed.

[0169] The control method adopted by the robotic arm is admittance control. The six-dimensional force sensor at the end of the robotic arm measures the input force of the patient's arm as the interactive force, calculates the variable impedance compensation force and the adaptive auxiliary force, and adds the interactive force, variable impedance compensation force and adaptive auxiliary force as the total input force. Figure 7 The desired velocity of the terminal is calculated in the mass-spring-damper model of the admittance controller shown in FIG. The admittance control model is:

[0170]

[0171] Among them, M is the inertia matrix; D is the damping matrix; K is the stiffness matrix; F input,i is the interaction force F at the end of the robotic arm itself ext,i , variable impedance compensation force F c,i and adaptive assist force F assist,i,n+1 The combined force.

[0172] Δx=x0-x d ;

[0173] Among them, x d , is the desired position, velocity and acceleration of the robot, and x0, The position, velocity, and acceleration values ​​that the robot theoretically needs to track when the external force is zero.

[0174] Example 2

[0175] A two-person upper limb robot interaction system based on adaptive assistance of motion performance, such as Figure 8 As shown, it includes an interaction data and motion state acquisition module 801, a variable impedance compensation force determination module 802, a patient motion performance quantitative evaluation module 803, an adaptive auxiliary force adjustment module 804, and a robotic arm execution module 805.

[0176] The interaction data and motion state acquisition module 801 is used to obtain the current interaction data and current motion state of each user; the interaction data includes the interaction force obtained from the corresponding force sensor, and the real-time position and speed of the corresponding robotic arm end in its own coordinate system.

[0177] The interaction data and motion state acquisition module 801 includes an interaction force acquisition unit and a motion state acquisition unit; the interaction force acquisition unit acquires the interaction force at the end of the robotic arm through a six-dimensional force sensor; the motion state acquisition unit converts the joint angle of the robotic arm into end position information and the joint angular velocity of the robotic arm into the end motion velocity through forward kinematics, and records the number of control frames of the robotic arm in the process of completing the task.

[0178] The variable impedance compensation force determination module 802 is used to calculate the motion intention parameters based on the real-time interaction force of the patient user and the instantaneous change rate of the interaction force, and dynamically adjust the impedance compensation parameters based on the motion intention parameters to calculate and generate the variable impedance compensation force; the input end of the variable impedance compensation force determination module 802 is connected to the output end of the interaction data and motion state acquisition module 801.

[0179] The variable impedance compensation force determination module 802 includes a motion intention parameter determination unit and a variable impedance compensation force determination unit; the motion intention parameter determination unit is used to calculate and determine the user's motion intention parameters; the variable impedance compensation force determination unit is used to dynamically adjust the impedance compensation parameters according to the motion intention parameters and generate a variable impedance compensation force.

[0180] The patient sports performance quantitative evaluation module 803 is used to quantify the patient user's sports performance based on the sports status data and generate a sports performance score; the input end of the patient sports performance quantitative evaluation module 803 is connected to the output end of the interaction data and sports status acquisition module 801.

[0181] The patient's sports performance quantitative evaluation module 803 includes a sports performance quantification unit and a sports performance evaluation unit; the sports performance quantification unit is used to quantify the performance based on the sports data; the sports performance evaluation unit is used to use fuzzy logic system reasoning to obtain a sports performance score based on the sports time and average speed obtained after the sports performance is quantified.

[0182] The adaptive auxiliary force adjustment module 804 is used to calculate the difference between the sports performance scores of two users, and dynamically adjust the ideal auxiliary force provided to each user in the next training cycle according to the difference through preset adaptive adjustment rules, thereby generating an adaptive auxiliary force directed towards the training goal; the input end of the adaptive auxiliary force adjustment module 804 is connected to the output end of the patient sports performance quantitative evaluation module 803.

[0183] The adaptive assisting force adjustment module 804 includes an ideal assisting force determination unit and an adaptive assisting force determination unit; the ideal assisting force determination unit dynamically adjusts the ideal assisting force provided to each user in the next training cycle based on the difference between the sports performance scores of two users; the adaptive assisting force determination unit is used to determine the final adaptive assisting force.

[0184] The input end of the robotic arm execution module 805 is respectively connected to the output end of the variable impedance compensation force determination module 802 and the output end of the adaptive auxiliary force adjustment module 804; the robotic arm execution module 805 vector-superimposes the interaction force of the robotic arm end itself, the variable impedance compensation force and the adaptive auxiliary force to obtain the total input force, substitutes the total input force into the admittance control model of the admittance controller, generates the desired speed and motion instructions of the robotic arm, and controls the robotic arm to perform corresponding operations.

[0185] The robotic arm execution module 805 is used to fuse the corresponding user's interaction force, variable impedance compensation force and adaptive auxiliary force as the input force of the user's corresponding training robot admittance control scheme to generate the desired speed and motion instructions of the robotic arm and control the robotic arm to perform corresponding operations; and execute the two-person upper limb robot interaction method based on motion performance adaptive assistance described in Example 1.

[0186] The above embodiments are specific implementation methods of the present invention, but the implementation methods of the present invention are not limited to the above embodiments. Any other combination, change, modification, substitution, and simplification that does not exceed the design concept of the present invention shall fall within the scope of protection of the present invention.

Claims

1. A two-person upper limb robot interaction method based on adaptive assistance of motion performance, characterized in that: The following steps are involved: S1. Obtaining interaction data of each user during training; the interaction data includes real-time interaction force and motion state data obtained from the end of the robot's mechanical arm; the motion state data includes current position and speed; S2. Generate motion intention parameters based on the real-time interaction force and the instantaneous rate of change of the interaction force, and dynamically adjust impedance compensation parameters based on the motion intention parameters to generate variable impedance compensation force; the impedance compensation parameters include compensation stiffness diagonal elements and compensation damping diagonal elements; S3. After completing one or more training task cycles, quantify the user's exercise performance based on the user's exercise status data during the task cycle and generate an exercise performance score; S4. Obtaining the difference in exercise performance between users; S5. Based on the adaptive adjustment rules, dynamically adjust the ideal assist force provided to each user in the next training cycle; S6, generating an adaptive auxiliary force directed toward the training target; S7. Integrate the real-time interactive force, variable impedance compensation force, and adaptive auxiliary force as the input of the admittance control scheme to generate the desired speed and motion instructions of the robotic arm and control the robotic arm to perform the corresponding operation.

2. The method for two-person upper limb robot interaction based on adaptive assistance of motion performance according to claim 1, characterized in that: In step S2, the steps of generating the movement intention parameter MA include: S201, real-time interactive force F ext Perform threshold processing and limiting, set the dead zone threshold F ext_z , and obtain the effective interaction force F extL ; F extL =sgn(F ext )·max(0,|F ext |-F ext_z ); S202, instantaneous rate of change of interaction force Perform threshold processing and limiting, set dead zone threshold Get the instantaneous rate of change of effective interaction force S203, generating movement intention parameter MA: Among them, F extL is the effective interaction force, F ext For real-time interaction, F ext_z is the set real-time interaction force dead zone threshold, is the instantaneous rate of change of the effective interaction force; is the instantaneous rate of change of the real-time interactive force, is the set dead zone threshold of the instantaneous change rate of the real-time interaction force, F ext_max is the preset maximum interaction force, is the instantaneous rate of change of the maximum interaction force, a is the influence factor, and proj(u,v) represents the scalar value of the projection of vector u on vector v.

3. The two-person upper limb robot interaction method based on adaptive assistance of sports performance according to claim 1 is characterized in that: In step S2, the step of generating a variable impedance compensation force includes: S211. Obtain the compensation stiffness diagonal element K according to the motion intention parameter ct : K ct =K ct_max ·0.5·(1+tanh(b·(MA-0.5))); Among them, K ct is the compensation stiffness diagonal element, MA is the motion intention parameter, K ct_max is the preset maximum compensation stiffness, b is the adjustment parameter; S212. Obtain the compensation damping diagonal element B according to the compensation stiffness diagonal element. ct : Among them, B ct To compensate for the damping diagonal elements, is the preset weighting factor; S213, generate diagonal compensation stiffness matrix K c and the diagonal compensation damping matrix B c ; K c =diag(K ct ,K ct ,K ct );B c =diag(B ct ,B ct ,B ct ); Among them, K c is the diagonal compensation stiffness matrix, B c is the diagonal compensation damping matrix; S214: Generate a variable impedance compensation force F based on the real-time position and speed of the user's end effector. c : F c =-(B c V+K c X); Among them, F c is the variable impedance compensation force, X is the real-time position of the end effector of the user, and V is the movement speed of the end effector.

4. The method for two-person upper limb robot interaction based on adaptive assistance of motion performance according to claim 1, characterized in that: In step S3, the steps of generating a sports performance score include: S301, in the task cycle, according to the control period ΔT and the number of control frames num consumed to complete the task n , get the exercise time t that the user spends to complete the nth training task n : t n =ΔT×num n ; Among them, t n The time it takes to complete the current training task, ΔT is the control cycle, num n is the number of control frames consumed to complete the task; S302: Obtain the average movement speed of the user during the process of completing the nth training task in, is the user's motion speed of the i-th control frame during the n-th training task, are the velocity vectors of the i-th control frame in the x, y, and z directions during the n-th training task at the end of the robotic arm, The average movement speed of the user during the nth training task; S303, the movement time t n and average speed As input, a fuzzy logic system is used to process and generate a numerical sports performance score TCF. n .

5. The method for two-person upper limb robot interaction based on adaptive assistance of motion performance according to claim 4, characterized in that: The fuzzy logic system includes fuzzification, fuzzy reasoning based on preset fuzzy rules, and defuzzification, and its specific steps include: S311, fuzzification: the input motion time t n and average speed After normalization, it is transformed into a fuzzy language value with membership degree; S312, fuzzy reasoning: The fuzzy language values ​​of the movement time and average speed are reasoned based on the preset fuzzy rules to obtain the output TCF n fuzzy set of ; S313, defuzzification: converting the fuzzy set result obtained by fuzzy inference output into an accurate value y; Where μ(x) is the membership value of the fuzzy set at x, a and b are the lower and upper limits of the integral respectively; Introduce a scaling factor to obtain the nth task training performance score TCF n The actual value of: TCF n =f scale ·y; Among them, f scale is the scaling factor.

6. The method for two-person upper limb robot interaction based on adaptive assistance of motion performance according to claim 1, characterized in that: In step S5, the adaptive adjustment rule is based on the difference E n =|TCF1 n -TCF2 n |Adaptive step size function, TCF1 n ,TCF2 n The specific steps for scoring the athletic performance of the two users and using the adaptive step length function to dynamically adjust the ideal assist force for each user in the next training cycle are as follows: S501, based on the difference E n The piecewise function determines the adaptive step size α n : Among them, α n is the adaptive step size, is the difference E n The segmentation point, α l ,...,α q is the corresponding step value; S502, according to the difference E n , adaptive step size α n and the ideal assist force current value F ideal,n Dynamically adjust the ideal assist force F provided to each user in round n+1 ideal,n+1 , the adjustment rules are: Among them, F ideal,n+1 is the ideal auxiliary force for the n+1th round, F ideal,n is the ideal auxiliary force for the nth round.

7. The method for two-person upper limb robot interaction based on adaptive assistance of motion performance according to claim 1, characterized in that: In step S6, the specific steps of generating the adaptive auxiliary force directed toward the training target are: Based on the dynamically adjusted ideal assist force F ideal,i,n+1 and the training target position, generating an adaptive auxiliary force pointing to the training target, and the direction of the auxiliary force always points to the current target point P ad : Among them, F assist,n+1 is the adaptive auxiliary force of the n+1th round, F ideal,n+1 is the ideal auxiliary force for the n+1th round, P ad Represents the vector pointing to the target point.

8. The two-person upper limb robot interaction method based on adaptive assistance of motion performance according to claim 1 is characterized in that: In step S7, the step of generating the desired speed of the manipulator by the admittance controller using the total input force is as follows: the interaction force of the end of the manipulator, the variable impedance compensation force, and the adaptive auxiliary force are substituted into the admittance control model of the admittance controller to obtain the desired speed of the manipulator; the admittance control model is: Among them, M is the inertia matrix; D is the damping matrix; K is the stiffness matrix; F input,i is the interaction force F at the end of the robotic arm itself ext,i , variable impedance compensation force F c,i and adaptive assist force F assist,i,n+1 The combined force; Δx=x0-x d ; Among them, x d , is the desired position, velocity and acceleration of the robot, and x0, The position, velocity, and acceleration values ​​that the robot theoretically needs to track when the external force is zero.

9. A two-person upper limb robot interaction system based on adaptive assistance of motion performance, characterized in that: It includes interactive data and motion state acquisition module, variable impedance compensation force determination module, patient motion performance quantitative evaluation module, adaptive auxiliary force adjustment module, and robotic arm execution module; The interaction data and motion state acquisition module is used to obtain the current interaction data and current motion state of each user; the interaction data includes the interaction force obtained from the corresponding force sensor, and the real-time position and speed of the corresponding robot end in its own coordinate system; The variable impedance compensation force determination module is used to calculate the movement intention parameter according to the real-time interaction force of the patient user and the instantaneous change rate of the interaction force, and dynamically adjust the impedance compensation parameter based on the movement intention parameter to calculate and generate the variable impedance compensation force; The input end of the variable impedance compensation force determination module is connected to the output end of the interaction data and motion state acquisition module; The patient sports performance quantification evaluation module is used to quantify the patient's sports performance based on the patient's sports status data and generate a sports performance score; the input end of the patient sports performance quantification evaluation module is connected to the output end of the interaction data and sports status acquisition module; The adaptive assist force adjustment module is used to calculate the difference between the athletic performance scores of two users and dynamically adjust the ideal assist force provided to each user in the next training cycle according to the difference using a preset adaptive adjustment rule, thereby generating an adaptive assist force directed towards the training goal; The input end of the adaptive auxiliary force adjustment module is connected to the output end of the patient's sports performance quantitative evaluation module; The input end of the manipulator execution module is connected to the output end of the variable impedance compensation force determination module and the output end of the adaptive auxiliary force adjustment module respectively; The robotic arm execution module is used to fuse the corresponding user's interaction force, variable impedance compensation force and adaptive auxiliary force as the input force of the user's corresponding training robot admittance control scheme to generate the desired speed and motion instructions of the robotic arm and control the robotic arm to perform corresponding operations; execute the two-person upper limb robot interaction method based on motion performance adaptive assistance as described in any one of claims 1-8.

10. The two-person upper limb robot interaction system based on adaptive assistance of sports performance according to claim 9 is characterized in that: The interaction data and motion state acquisition module includes an interaction force acquisition unit and a motion state acquisition unit; the interaction force acquisition unit acquires the interaction force at the end of the manipulator through a six-dimensional force sensor; the motion state acquisition unit converts the joint angle of the manipulator into end position information and the joint angular velocity of the manipulator into the end motion velocity through forward kinematics, and records the number of control frames of the manipulator in the process of completing the task; The variable impedance compensation force determination module includes a motion intention parameter determination unit and a variable impedance compensation force determination unit; the motion intention parameter determination unit is used to calculate and determine the user's motion intention parameter; the variable impedance compensation force determination unit is used to dynamically adjust the impedance compensation parameter according to the motion intention parameter and generate the variable impedance compensation force; The patient's sports performance quantitative evaluation module includes a sports performance quantification unit and a sports performance evaluation unit; the sports performance quantification unit is used to quantify the performance according to the sports data; the sports performance evaluation unit is used to use fuzzy logic system reasoning to obtain a sports performance score based on the sports time and average speed obtained after the sports performance is quantified; The adaptive assist force adjustment module includes an ideal assist force determination unit and an adaptive assist force determination unit; the ideal assist force determination unit dynamically adjusts the ideal assist force provided to each user in the next training cycle based on the difference between the exercise performance scores of the two users; The adaptive assist force determination unit is used to determine a final adaptive assist force; The robotic arm execution module vector-superimposes the interaction force, variable impedance compensation force and adaptive auxiliary force of the robotic arm end to obtain the total input force, substitutes the total input force into the admittance control model of the admittance controller, generates the desired speed and motion instructions of the robotic arm, and controls the robotic arm to perform corresponding operations.

Images