Robot motion real-time parameter identification and compensation control method and robot

Abstract

The invention discloses a robot motion real-time parameter identification and compensation control method and a robot. Robot motion information is obtained in real time, friction parameters closer to the current moment are calculated in real time, and a high-precision friction model is achieved by continuously updating the friction parameters; the predicted friction torque can be closer to the actual friction torque, a more accurate dynamic model can be constructed, and the response speed and the control precision of a robot control system are improved; in addition, according to the compensation control method, the feedforward compensation torque is calculated once after the real-time parameter identification is executed once, the method is used for a feedback system of the robot to achieve real-time compensation control over robot motion, and the robot motion control precision can be improved.

Description

Technical Field

[0001] The present invention relates to the field of robot technology, and in particular to a robot motion real-time parameter identification and compensation control method and a robot. Background Art

[0002] As people's demand for robots increases, the application areas of robots are also expanding. They have been widely used in grinding, polishing, assembly and other fields. To achieve high-precision motion control, the robot dynamics model must be considered for position control or torque control. By compensating for the dynamic characteristics of the robot, the system can quickly reach a steady state, reduce overshoot, and suppress motion vibration, thereby improving dynamic response speed and control accuracy.

[0003] The joint motion of the robot is transmitted to the robot arm by the servo motor through the RV reducer or harmonic reducer. However, due to the friction between the moving parts of the reducer, the friction in the joint transmission must be considered when establishing the robot dynamics equation. The friction parameters identified by the existing technology are fixed, and need to be identified for different working conditions, which takes a lot of time. In addition, due to the complex characteristics of the friction model, the calculated friction torque is not accurate. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a method and a robot for real-time parameter identification and compensation control of robot motion, which realizes a high-precision friction model by updating friction parameters in real time. The predicted friction torque can be closer to the actual friction torque, which helps to build a more accurate dynamic model and improve the response speed and control accuracy of the robot control system.

[0005] In order to solve the above technical problems, the first aspect of the present invention discloses a method for real-time parameter identification of robot motion, the method comprising:

[0006] S1. According to the trajectory information q(k) of the robot joint at the current time k, To calculate the overall regression matrix Φ at the current moment b (k) and the friction regression matrix Φ f (k): where q, are the position, velocity, and acceleration vector of the robot joints, Φ b (k) Φ f (k)

[0007] S2. Calculate the gain matrix K(k) at the current moment. The formula is as follows:

[0008]

[0009] In the formula, Φ f (k-1) is the friction regression matrix calculated at the last moment, I is the unit diagonal matrix, Φ f The transposed matrix of , P(k-1) is the covariance matrix of the previous moment;

[0010] S3. Calculate the estimated friction torque τ at the current moment f (k), the formula is as follows:

[0011]

[0012] Where τ(k) is the joint torque at the current moment, Φ b (k) is the overall regression matrix at the current moment, β ib is the inertia basic parameter;

[0013] S4. According to the friction parameter β at the previous moment f (k-1) to calculate the estimation error ε:

[0014] ε=τ f (k)-Φ f (k)β f (k-1);

[0015] S5. Update the robot's current friction parameter β by the estimated error ε and the gain matrix K(k) f (k):

[0016] β f (k) = β f (k-1)+K(k)ε;

[0017] S6. Calculate the covariance matrix P(k) at the current moment, which is reserved for calculation at the next moment:

[0018] P(k)=(IK(k)Φ f (k))P(k-1);

[0019] S7. Based on the current friction regression matrix Φ f (k) and the current friction parameter β f (k) Calculate the compensation friction torque τ at the current moment F (k):

[0020] τ F (k) = Φ f (k)β f (k).

[0021] As an optional implementation, the sampling torque of each joint is calculated by sampling the current:

[0022] The sampled current is first filtered:

[0023] I filter (k) = aI filter (k-1)+(1-a)I(k),

[0024] Among them, I(k) is the collected current at the time k, I filter (k-1) is the filter current at time k-1, I filter (k) is the filter current at time k, a is the filter coefficient;

[0025] Then the joint torque τ(k) at time k is obtained by the following conversion formula:

[0026] τ(k)=RK t I filter (k)

[0027] Among them, R is the reduction ratio, K t is the torque coefficient.

[0028] As another optional implementation,

[0029] Φ f (k) is the friction regression matrix at the current moment, where is the velocity of the robot's i-th joint at the current moment, sign is the sign function;

[0030] The overall regression matrix Φ b (k) is calculated by the following formula:

[0031]

[0032] Among them, τ is the joint torque, β b is the minimum inertia parameter set of the dynamic parameters to be identified, including the basic inertia parameters and friction parameters of each connecting rod, Φ ib is the inertial regression matrix determined in advance according to the robot's own parameters, β ib is the basic inertia parameter determined in advance according to the robot's own parameters, β f is the friction parameter β at the current moment f (k).

[0033] As another optional implementation, before step S1, the method further includes:

[0034] Control the robot to run according to the excitation trajectory, collect motion data during the operation and determine the friction parameter β to be identified f The initial value β at the initial time f (0);

[0035] The excitation trajectory is designed by five finite Fourier series, and the excitation trajectory is defined as:

[0036]

[0037] Among them, q i (t) and They represent the position, angular velocity and angular acceleration of the i-th joint at time t, respectively, and q i0 is the joint angle at the initial moment; ω f is the fundamental frequency of the Fourier series excitation trajectory, a k and b k are the coefficients of the Fourier series excitation trajectory.

[0038] As another optional implementation, before step S1, the method further includes:

[0039] The robot is controlled to run according to the excitation trajectory. During the running process, the motion data of each joint of the robot are collected to combine the following linear regression equations:

[0040]

[0041] Wherein, m is the number of dynamic data sets of the robot joints obtained when the robot moves according to the excitation trajectory, each of the dynamic data sets contains the input torque, angle, angular velocity and angular acceleration of all joints of the robot, and m is greater than the minimum inertia parameter set β b The number of parameters in; ε is the error vector of nm×1 least squares, n is the robot's degree of freedom; Φ b is the overall regression matrix corresponding to a single time point, H b is the Φ corresponding to the m time points obtained when the robot runs along the excitation trajectory for a period of time b The overall observation matrix composed of; Γ is the joint torque corresponding to m time points;

[0042] Solve the minimum inertia parameter set β in the linear regression equations b .

[0043] As another optional implementation, in the solving of the linear regression equations, the minimum inertia parameter set β b Afterwards, the method further comprises:

[0044] Get the error vector ε;

[0045] The error vector ε is rearranged to form an m×n error matrix R, whose structure is as follows:

[0046] R=[R 1 R2 …R n ],

[0047] Among them, R i represents the error vector consisting of m sets of torque errors of the i-th joint among n joints;

[0048] Calculate the covariance matrix C, the formula is as follows:

[0049] C=cov(R),

[0050] Where C is an n×n symmetric matrix, and cov is the covariance calculation function;

[0051] Normalize the linear regression equations:

[0052]

[0053] Where W is the weighting matrix;

[0054] Based on the following constraints, the least squares method is used to solve the minimum inertia parameter set β b :

[0055]

[0056] Where n is the number of connecting rods, m is i is the mass of connecting rod i, I i is the inertia tensor matrix of the connecting rod i in the center of mass coordinate system, f ci is the Coulomb friction coefficient of connecting rod i, f vi is the viscous friction coefficient of connecting rod i;

[0057] Determine the minimum inertia parameter set obtained by the solution as the minimum inertia parameter set β b The initial value at the initial time.

[0058] As another optional implementation, the minimum inertia parameter set β in solving the linear regression equations is b ,include:

[0059] The least square method is used to solve the minimum inertia parameter set β in the linear regression equations based on the following constraints: b :

[0060]

[0061] Where n is the number of connecting rods, m is i is the mass of connecting rod i, I i is the inertia tensor matrix of the connecting rod i in the center of mass coordinate system, f ci is the Coulomb friction coefficient of connecting rod i, f vi is the viscous friction coefficient of connecting rod i.

[0062] As another optional implementation, minimizing the observation matrix condition number is used as the optimization index of the excitation trajectory, and the optimization objective function of the excitation trajectory is:

[0063]

[0064] Where cond is the condition number of the observation matrix, q min ,q max , They are the upper and lower limits of the rotational position, velocity and acceleration of each joint of the robot; are the robot joint velocity and acceleration at the initial moment, They are the velocity and acceleration of each joint of the robot at the termination time tf respectively.

[0065] A second aspect of the present invention discloses a real-time compensation control method for robot motion, wherein the robot has a feedback control system, and the method comprises:

[0066] When the robot runs an arbitrary trajectory, the motion data of each joint of the robot in the current time period is collected, and the estimated friction torque τ corresponding to the current time period is calculated according to the real-time parameter identification method of the robot motion as described in the first aspect of the present invention. F (k) and β ib , and then calculate the feedforward compensation torque τ according to the following formula forward (k):

[0067]

[0068] The feedforward compensation torque τ forward (k) The actual value of the friction torque of the robot is provided to the feedback control system.

[0069] The third aspect of the present invention discloses a robot, comprising a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps in the method disclosed in the first aspect or the second aspect of the present invention.

[0070] A fourth aspect of the present invention discloses a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the method disclosed in the first aspect or the second aspect of the present invention.

[0071] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0072] Compared with the prior art, since the robot's joint friction has nonlinear characteristics and the joint movement speed varies greatly, the embodiment of the present invention proposes an online identification method for friction parameters. By acquiring the robot's motion information in real time and calculating the friction parameters that are more suitable for the current moment in real time, a high-precision friction model is achieved by continuously updating the friction parameters, so that the predicted friction torque can be closer to the actual friction torque, which is helpful to build a more accurate dynamic model and improve the response speed and control accuracy of the robot control system. The compensation control method of the present invention calculates a feedforward compensation torque once after executing the above-mentioned real-time parameter identification, which is used in the robot's feedback system to realize real-time compensation control of the robot motion, which can improve the robot motion control accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0074] Figure 1 It is a flow chart of a method for real-time parameter identification of robot motion disclosed in an embodiment of the present invention;

[0075] Figure 2 It is a partial flow chart of a method for real-time parameter identification of robot motion disclosed in an embodiment of the present invention;

[0076] Figure 3 and Figure 4 is a specific parameter composition diagram of the inertial basic parameters disclosed in the embodiment of the present invention;

[0077] Figure 5 It is a flow chart of a real-time compensation control method for robot motion disclosed in an embodiment of the present invention;

[0078] Figure 6 It is a schematic diagram of the structure of a robot disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0079] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0080] Embodiment 1

[0081] See also Figures 1-2 The embodiment of the present invention discloses a method for real-time parameter identification of robot motion, the method comprising:

[0082] S1. According to the trajectory information q(k) of the robot joint at the current time k, To calculate the overall regression matrix Φ at the current moment b (k) and the friction regression matrix Φ f (k): where q, are the position, velocity, and acceleration vector of the robot joints, Φ b (k) Φ f (k)

[0083] S2. Calculate the gain matrix K(k) at the current moment. The formula is as follows:

[0084]

[0085] In the formula, Φ f (k-1) is the friction regression matrix calculated at the last moment, I is the unit diagonal matrix, Φ f The transposed matrix of , P(k-1) is the covariance matrix of the previous moment.

[0086] S3. Calculate the estimated friction torque τ at the current moment f (k), the formula is as follows:

[0087]

[0088] Where τ(k) is the joint torque at the current moment, Φ b (k) is the overall regression matrix at the current moment, β ib is the basic inertia parameter.

[0089] In this step, because the robot joint torque includes inertia torque and friction torque, the inertia torque is subtracted here to obtain the estimated friction torque τ f (k), used to calculate the current new friction parameter β in the subsequent steps f (k).

[0090] S4. According to the friction parameter β at the previous moment f (k-1) to calculate the estimation error ε:

[0091] ε=τ f (k)-Φ f (k)β f (k-1).

[0092] In this step, the estimated error here refers to the error between the friction torque calculated by the friction parameters of the previous moment and the actual friction torque (represented by the estimated friction torque of the current moment calculated in the above step S3, because the friction torque cannot be measured and can only be calculated through step S3).

[0093] S5. Update the robot's current friction parameter β by the estimated error ε and the gain matrix K(k) f (k):

[0094] β f (k) = β f (k-1)+K(k)ε.

[0095] S6. Calculate the covariance matrix P(k) at the current moment, which is reserved for calculation at the next moment:

[0096] P(k)=(IK(k)Φ f (k))P(k-1).

[0097] S7. Based on the current friction regression matrix Φ f (k) and the current friction parameter β f (k) Calculate the compensation friction torque τ at the current moment F (k):

[0098] τ F (k) = Φ f (k)β f (k).

[0099] The method of the present invention can be independently calculated in the robot control system, or it can be run on a host computer such as a PC. Optionally, the interval between two moments can be set to 1-2ms, that is, calculation is performed every 1-2ms. In fact, in the verification experiment, the actual calculation time each time reached about 800 microseconds, that is, the result can be calculated in about 0.8ms.

[0100] Since the joint friction of the robot has nonlinear characteristics and the joint movement speed varies greatly, an embodiment of the present invention proposes an online identification method for friction parameters. By acquiring the robot motion information in real time and calculating the friction parameters that are more suitable for the current moment in real time, a high-precision friction model is implemented by continuously updating the friction parameters. This makes the predicted friction torque closer to the actual friction torque, which helps to build a more accurate dynamic model and improve the response speed and control accuracy of the robot control system.

[0101] In an optional embodiment, the sampled torque of each joint is calculated by sampling the current:

[0102] The sampled current is first filtered:

[0103] I filter (k) = aI filter (k-1)+(1-a)I(k),

[0104] Among them, I(k) is the collected current at the time k, I filter (k-1) is the filter current at time k-1, I filter (k) is the filter current at time k, a is the filter coefficient;

[0105] Then the joint torque τ(k) at time k is obtained by the following conversion formula:

[0106] τ(k)=RK t I filter (k)

[0107] Among them, R is the reduction ratio, K t is the torque coefficient.

[0108] In another optional embodiment, the joint position signal sent by the sensor is processed by first-order low-pass filtering to obtain position information q, and then the angular velocity and angular acceleration are obtained by second-order differentiation of the position information q. The sensor can be an existing position sensor that outputs a current or voltage signal.

[0109] In yet another optional embodiment, Φ f (k) is the friction regression matrix at the current moment, where is the velocity of the robot's i-th joint at the current moment, sign is the sign function;

[0110] The overall regression matrix Φ b (k) is calculated by the following formula:

[0111]

[0112] Among them, τ is the joint torque, β b It is the minimum inertia parameter set including the basic inertia parameters and friction parameters of each connecting rod waiting for the identification of dynamic parameters, Φ ib is the inertial regression matrix determined in advance according to the robot's own parameters, β ib is the basic inertia parameter determined in advance according to the robot's own parameters, β f is the friction parameter β at the current moment f (k). Among them, β f =[f ci f vi b i ] T , f c is the Coulomb friction coefficient, f v is the viscous friction coefficient, b is the friction bias; Figures 3-4 represents the inertia basic parameter β ib There are 40 parameters of the joint, where c is the joint center of mass, B is the motor inertia, m is the joint mass, a is the connecting rod length, d is the joint coordinate system offset, and I is the moment of inertia.

[0113] In this embodiment, a dynamic model of the robot considering friction is established based on the Newton-Euler method:

[0114]

[0115] Where: q, are the position, velocity, and acceleration vectors of the robot joints, M(q) is the inertia matrix of the robot arm, is the velocity term matrix related to centrifugal force and Coriolis force, g(q) is the gravity term, and τ is the joint torque vector; where, represents the friction force / torque, and the friction of joint i is modeled using the Coulomb viscous friction model:

[0116]

[0117] In the formula, f c is the Coulomb friction coefficient, f v is the viscous friction coefficient, b is the friction bias, and sign is the sign function.

[0118] The dynamic model of the robot considering friction is linearized, and the full-rank minimum regression matrix expression is obtained by using QR decomposition parameter reorganization and other methods. and the corresponding minimum inertia parameter set β b , that is, the above formula is obtained

[0119] In yet another optional embodiment, before step S1, the method further includes:

[0120] Control the robot to run according to the excitation trajectory, collect motion data during the operation and determine the friction parameter β to be identified f The initial value β at the initial time f (0);

[0121] The excitation trajectory is designed by five finite Fourier series, and the excitation trajectory is defined as:

[0122]

[0123] Among them, q i (t) and They represent the position, angular velocity and angular acceleration of the i-th joint at time t, respectively, and q i0is the joint angle at the initial moment; ω f is the fundamental frequency of the Fourier series excitation trajectory, a k and b k are the coefficients of the Fourier series excitation trajectory.

[0124] In yet another optional embodiment, before step S1, the method further includes:

[0125] The robot is controlled to run according to the excitation trajectory. During the running process, the motion data of each joint of the robot are collected (the position information q of each joint of the robot and the driving current of each joint are collected; the sampled data needs to be smoothed and filtered first), and the following linear regression equations are combined:

[0126]

[0127] Wherein, m is the number of dynamic data sets of the robot joints obtained when the robot moves according to the excitation trajectory, each of the dynamic data sets contains the input torque, angle, angular velocity and angular acceleration of all joints of the robot, and m is greater than the minimum inertia parameter set β b The number of parameters in; ε is the error vector of nm×1 least squares, n is the robot's degree of freedom; Φ b is the overall regression matrix corresponding to a single time point, H b is the Φ corresponding to the m time points obtained when the robot runs along the excitation trajectory for a period of time b The overall observation matrix composed of; Γ is the joint torque corresponding to m time points;

[0128] Solve the minimum inertia parameter set β in the linear regression equations b .

[0129] In yet another optional embodiment, the minimum inertia parameter set β in solving the linear regression equations is b Afterwards, the method further comprises:

[0130] Get the error vector ε;

[0131] The error vector ε is rearranged to form an m×n error matrix R, whose structure is as follows:

[0132] R=[R 1 R 2 …R n ],

[0133] Among them, R i represents the error vector consisting of m sets of torque errors of the i-th joint among n joints;

[0134] Calculate the covariance matrix C, the formula is as follows:

[0135] C=cov(R),

[0136] Where C is an n×n symmetric matrix, and cov is the covariance calculation function;

[0137] Normalize the linear regression equations:

[0138]

[0139] Where W is the weighting matrix;

[0140] Based on the following constraints, the least squares method is used to solve the minimum inertia parameter set β b :

[0141]

[0142] Where n is the number of connecting rods, m is i is the mass of connecting rod i, I i is the inertia tensor matrix of the connecting rod i in the center of mass coordinate system, f ci is the Coulomb friction coefficient of connecting rod i, f vi is the viscous friction coefficient of connecting rod i;

[0143] Determine the minimum inertia parameter set obtained by the solution as the minimum inertia parameter set β b The initial value at the initial time.

[0144] After solving the minimum inertia parameter set β b On the basis of , considering that when collecting m groups of joint motion data, the measurement noise between joint torques will interfere with each other, in order to reduce the impact of abnormal data on the recognition results, the error vector ε is rearranged in this embodiment, and then the minimum inertia parameter set β is re-arranged. b Solve and obtain the final minimum inertia parameter set β b The initial value of the friction parameter β can provide an initial parameter value that accurately matches the robot's own characteristics when the robot starts running on any trajectory, thus improving the accuracy of the subsequent calculation of the friction parameter β. f (k) Accuracy.

[0145] In yet another optional embodiment, the minimum inertia parameter set β in solving the linear regression equations is b ,include:

[0146] The least square method is used to solve the minimum inertia parameter set β in the linear regression equations based on the following constraints: b :

[0147]

[0148] Where n is the number of connecting rods, m is i is the mass of connecting rod i, I i is the inertia tensor matrix of the connecting rod i in the center of mass coordinate system, f ci is the Coulomb friction coefficient of connecting rod i, f vi is the viscous friction coefficient of connecting rod i.

[0149] For this step, the least squares method can be directly used to solve the minimum inertia parameter set β b However, the unconstrained least squares method cannot guarantee the physical meaning of the minimum inertia parameter set. For example, the mass of the connecting rod obtained may be negative, which is contrary to the actual physical meaning. Therefore, it is necessary to consider the physical meaning of the inertia parameters in the solution process. For connecting rod i, its mass m i and the inertia tensor matrix I in the mass center coordinate system i The following physical constraints must be met: That is, the mass of the connecting rod m i To be a positive number, and the inertia tensor matrix I in the mass center coordinate system i This embodiment takes into account the physical meaning of the inertia parameter during the solution process to avoid being inconsistent with the actual physical meaning.

[0150] In yet another optional embodiment, minimizing the observation matrix condition number is used as the optimization index of the excitation trajectory, and the optimization objective function of the excitation trajectory is:

[0151]

[0152] Where cond is the condition number of the observation matrix, q min ,q max , They are the upper and lower limits of the rotational position, velocity and acceleration of each joint of the robot; are the robot joint velocity and acceleration at the initial moment, They are the velocity and acceleration of each joint of the robot at the termination time tf respectively.

[0153] In yet another optional embodiment, the initial value P(0) of the covariance matrix P at the initial moment is set to be an a×a diagonal matrix:

[0154]

[0155] Where a is equal to the number of friction parameters of all joints, and p is 10^6. For example, each joint has 3 friction parameters, and a 6-DOF robot has 18 parameters. Then P(0) is an 18×18 diagonal matrix, and each value is p.

[0156] Embodiment 2

[0157] See also Figure 5 The embodiment of the present invention discloses a real-time compensation control method for robot motion, wherein the robot has a feedback control system, and the method comprises:

[0158] S100. When the robot runs any trajectory, collect the motion data of each joint of the robot in the current time period, and calculate the estimated friction torque τ corresponding to the current time period according to the robot motion real-time parameter identification method as described in any one of claims 1 to 7 F (k), overall regression matrix and the inertial basis parameter β ib , and then calculate the feedforward compensation torque τ according to the following formula forward (k):

[0159]

[0160] S200. The feedforward compensation torque τ forward (k) The actual value of the friction torque of the robot is provided to the feedback control system.

[0161] The embodiment of the present invention calculates a feedforward compensation torque after performing a parameter identification, which is used in the robot's feedback system to realize real-time compensation control of the robot's motion, improve the robot's motion control accuracy, and can be used in industrial scenarios such as robot mobile handling, polishing and grinding, and heavy-load hydraulics.

[0162] Embodiment 3

[0163] See also Figure 6 An embodiment of the present invention discloses a robot, comprising a memory 201, a processor 202 and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method described in Embodiment 1 or Embodiment 2.

[0164] Embodiment 4

[0165] An embodiment of the present invention discloses a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the method described in Embodiment 1 or Embodiment 2 when executed by a processor.

[0166] The contents disclosed in the embodiments of the present invention only disclose the preferred embodiments of the present invention, which are only used to illustrate the technical solutions of the present invention rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, it should be understood by those skilled in the art that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for real-time parameter identification of robot motion, characterized in that: The method comprises: S1. According to the trajectory information q(k) of the robot joint at the current time k, To calculate the overall regression matrix Φ at the current moment b (k) and the friction regression matrix Φ f (k): where q, are the position, velocity, and acceleration vector of the robot joints, Φ b (k) Φ f (k) S2. Calculate the gain matrix K(k) at the current moment. The formula is as follows: In the formula, Φ f (k-1) is the friction regression matrix calculated at the last moment, I is the unit diagonal matrix, is Φ f The transposed matrix of , P(k-1) is the covariance matrix of the previous moment; S3. Calculate the estimated friction torque τ at the current moment f (k), the formula is as follows: Where τ(k) is the joint torque at the current moment, Φ b (k) is the overall regression matrix at the current moment, β ib is the inertia basic parameter; S4. According to the friction parameter β at the previous moment f (k-1) to calculate the estimation error ε: e=t f (k)-Φ f (k)b f (k-1); S5. Update the robot's current friction parameter β by the estimated error ε and the gain matrix K(k) f (k): b f (k)=β f (k-1)+K(k)e; S6. Calculate the covariance matrix P(k) at the current moment, which is reserved for calculation at the next moment: P(k)=(I-K(k)Φ f (k))P(k-1); S7. According to the current friction regression matrix Φ f (k) and the current friction parameter β f (k) Calculate the compensation friction torque τ at the current moment F (k): t F (k)=Φ f (k)b f (k)。 2. The method for real-time parameter identification of robot motion according to claim 1, characterized in that: The sampling torque of each joint is calculated by sampling current: The sampled current is first filtered: and filter (k)=aI filter (k-1)+(1-a)I(k), Among them, I(k) is the collected current at the time k, I filter (k-1) is the filter current at time k-1, I filter (k) is the filter current at time k, a is the filter coefficient; Then the joint torque τ(k) at time k is obtained by the following conversion formula: τ(k)=RK t I filter (k), Among them, R is the reduction ratio, K t is the torque coefficient.

3. The method for real-time parameter identification of robot motion according to claim 1, characterized in that: Φ f (k) is the friction regression matrix at the current moment, where is the velocity of the robot's i-th joint at the current moment, sign is the sign function; The overall regression matrix Φ b (k) is calculated by the following formula: Among them, τ is the joint torque, β b is the minimum inertia parameter set of the dynamic parameters to be identified, including the basic inertia parameters and friction parameters of each connecting rod, Φ ib is the inertial regression matrix determined in advance according to the robot's own parameters, β ib is the basic inertia parameter determined in advance according to the robot's own parameters, β f is the friction parameter β at the current moment f (k).

4. The method for real-time parameter identification of robot motion according to claim 1, characterized in that: Before step S1, the method further includes: Control the robot to run according to the excitation trajectory, collect motion data during the operation and determine the friction parameter β to be identified f The initial value β at the initial time f (0); The excitation trajectory is designed by five finite Fourier series, and the excitation trajectory is defined as: Among them, q i (t) and They represent the position, angular velocity and angular acceleration of the i-th joint at time t, respectively, and q i0 is the joint angle at the initial moment; ω f is the fundamental frequency of the Fourier series excitation trajectory, a k and b k are the coefficients of the Fourier series excitation trajectory.

5. The method for real-time parameter identification of robot motion according to claim 1, characterized in that: Before step S1, the method further includes: The robot is controlled to run according to the excitation trajectory. During the running process, the motion data of each joint of the robot are collected to combine the following linear regression equations: Wherein, m is the number of dynamic data sets of the robot joints obtained when the robot moves according to the excitation trajectory, each of the dynamic data sets contains the input torque, angle, angular velocity and angular acceleration of all joints of the robot, and m is greater than the minimum inertia parameter set β b The number of parameters in; ε is the error vector of nm×1 least squares, n is the robot's degree of freedom; Φ b is the overall regression matrix corresponding to a single time point, H b is the Φ corresponding to the m time points obtained when the robot runs along the excitation trajectory for a period of time b The overall observation matrix composed of; Γ is the joint torque corresponding to m time points; Solve the minimum inertia parameter set β in the linear regression equations b .

6. The method for real-time parameter identification of robot motion according to claim 5, characterized in that: The minimum inertia parameter set β in solving the linear regression equations b Afterwards, the method further comprises: Get the error vector ε; The error vector ε is rearranged to form an m×n error matrix R, whose structure is as follows: R=[R 1 R 2 …R n ], Among them, R i represents the error vector consisting of m sets of torque errors of the i-th joint among n joints; Calculate the covariance matrix C, the formula is as follows: C=cov(R), Where C is an n×n symmetric matrix, and cov is the covariance calculation function; Normalize the linear regression equations: Where W is the weighting matrix; Based on the following constraints, the least squares method is used to solve the minimum inertia parameter set β b : Where n is the number of connecting rods, m is i is the mass of connecting rod i, I i is the inertia tensor matrix of the connecting rod i in the center of mass coordinate system, f ci is the Coulomb friction coefficient of connecting rod i, f vi is the viscous friction coefficient of connecting rod i; Determine the minimum inertia parameter set obtained by the solution as the minimum inertia parameter set β b The initial value at the initial time.

7. The method for real-time parameter identification of robot motion according to claim 5, characterized in that: The minimum inertia parameter set β in solving the linear regression equations b ,include: The least square method is used to solve the minimum inertia parameter set β in the linear regression equations based on the following constraints: b : Where n is the number of connecting rods, m is i is the mass of connecting rod i, I i is the inertia tensor matrix of the connecting rod i in the center of mass coordinate system, f ci is the Coulomb friction coefficient of connecting rod i, f vi is the viscous friction coefficient of connecting rod i.

8. The method for real-time parameter identification of robot motion according to claim 4, characterized in that: The minimization of the observation matrix condition number is used as the optimization index of the excitation trajectory, and the optimization objective function of the excitation trajectory is: Where cond is the condition number of the observation matrix, q min ,q max , They are the upper and lower limits of the rotational position, velocity and acceleration of each joint of the robot; are the robot joint velocity and acceleration at the initial moment, They are the velocity and acceleration of each joint of the robot at the termination time tf respectively.

9. A real-time compensation control method for robot motion, wherein the robot has a feedback control system, characterized in that: The method comprises: When the robot runs any trajectory, the motion data of each joint of the robot in the current time period is collected, and the estimated friction torque τ corresponding to the current time period is calculated according to the robot motion real-time parameter identification method according to any one of claims 1 to 8 F (k) and β ib , and then calculate the feedforward compensation torque τ according to the following formula forward (k): The feedforward compensation torque τ forward (k) The actual value of the friction torque of the robot is provided to the feedback control system.

10. A robot comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 9.

Images