A data-driven interactive process learning method for robot compliant control

Abstract

The present invention discloses a robot compliance control method using data-driven interactive process learning, comprising: a workpiece surface direction motion instruction acquisition step to obtain the robot's motion instructions along the workpiece surface direction; a posture error and posture motion instruction acquisition step to obtain the robot's terminal angular velocity instruction for tracking the desired posture; a force control direction motion control instruction acquisition step to obtain the robot's motion control instruction in the force control direction; and a normalized control instruction construction step to construct a unified robot speed instruction based on the robot's motion instructions along the workpiece surface direction, the terminal angular velocity instruction for tracking the desired posture, and the robot's motion control instruction in the force control direction. By constructing a unified robot speed instruction, the present invention achieves efficient adaptation to the shape of the object's surface set while ensuring sufficient contact force accuracy, thereby realizing high-precision force control.

Description

Technical Field

[0001] The present invention relates to robot control technology, and in particular to a robot compliant control method based on data-driven interactive process learning. Background Art

[0002] With China's increasing emphasis on industrial robots and the growing market demand, demand for robots performing environmentally interactive tasks has surged. In these situations, the robot's end-point interacts with the external environment. Simple end-point position control cannot fully guarantee processing quality and production efficiency. Especially when information about the contact object, such as its position, posture, and material, is uncertain, how can the robot efficiently adapt to the object's surface shape based on limited sensor information while maintaining sufficient contact force accuracy? This is a pressing control issue and poses a significant challenge to the robot's online, efficient, and compliant control.

[0003] Current robot compliance control methods are mostly based on impedance control methods, that is, the interaction process between the robot and the unknown environment is modeled as a mass-spring-damper robot system, and the robot's motion control quantity is corrected based on this reference, so that the robot exhibits certain compliance characteristics.

[0004] However, this model's assumptions have limited adaptability to workpieces made of complex materials or with shape-dependent physical properties. Furthermore, during compliant control, the surface contour of the workpiece changes (i.e., real-time changes in the vertical direction of the workpiece), requiring the robot to not only have good adaptability in force control but also to achieve real-time recognition and tracking of the workpiece's vertical direction. Summary of the Invention

[0005] The purpose of the present invention is to overcome the above-mentioned deficiencies of the prior art and provide a robot compliant control method for data-driven interactive process learning so as to have good adaptability.

[0006] To achieve the above object, the technical solution of the present invention is:

[0007] A robot compliant control method based on data-driven interactive process learning, comprising:

[0008] a workpiece surface direction motion instruction acquisition step, obtaining the robot's motion instruction along the workpiece surface direction;

[0009] The posture error and posture motion instruction acquisition step obtains the end angular velocity instruction of the robot end for tracking the desired posture;

[0010] a motion control instruction acquisition step in the force control direction, obtaining the motion control instruction of the robot in the force control direction;

[0011] The normalized control instruction construction step constructs a unified robot speed instruction based on the robot's motion instruction along the workpiece surface, the end angular velocity instruction of the robot's end tracking the desired posture, and the robot's motion control instruction in the force control direction.

[0012] Optionally, the step of acquiring the posture error and posture motion instruction includes:

[0013] A force sensor installed at the end of the robot is used to obtain the contact force between the current robot end and the uncertain workpiece, and based on this, the normal information of the contact point is obtained to obtain the rotation matrix of the desired posture of the robot end;

[0014] In the posture control space, the terminal angular velocity instruction for tracking the desired posture of the robot terminal is calculated based on the rotation matrix of the desired posture.

[0015] Optionally, the force sensor installed at the end of the robot is used to obtain the contact force between the current end of the robot and the uncertain workpiece, and based on this, the normal information of the contact point is obtained to obtain the rotation matrix of the desired posture of the end of the robot, including:

[0016] In the current robot end force sensor coordinate system, based on the obtained contact force between the current robot end and the uncertain workpiece , normalize the external force data fed back by the force sensor:

[0017] (1);

[0018] Among them, the vector Indicates the rotation angle of the robot end, represents the norm of the contact force at the end of the robot;

[0019] Vector-based Construct an antisymmetric characteristic matrix describing the robot's terminal posture S :

[0020] (2);

[0021] in, n (3) Represents a vector The third element in n (2) Represents a vector The second element in n (1) represents a vector The first element in ;

[0022] The rotation matrix of the desired pose of the robot end R d Expressed as:

[0023] (3);

[0024] in, Represents the three-dimensional identity matrix.

[0025] Optionally, the step of calculating, in the posture control space, the terminal angular velocity instruction for tracking the desired posture of the robot terminal based on the rotation matrix of the desired posture includes:

[0026] In the attitude control space, the rotation matrix based on the desired attitude of the robot end R d To design the angular velocity control, convert the rotation matrix of the desired posture into quaternion form:

[0027] ;

[0028] in, Indicates the expected quaternion, r 11 express The element at row 1 and column 1, r 22 express The element at row 2 and column 2, r 33 express The element at row 3 and column 3, r 32 express The element at row 3 and column 2, r 23 express The element at row 2 and column 3, r 13 express The element in row 1 and column 3, r 31 express The element at row 3 and column 1, r 21 express The element at row 2 and column 1, r 12 express The element at row 1 and column 2; Represent the expected quaternions in order The values ​​of the 1st to 4th elements of the column vector;

[0029] By reading the robot's current posture quaternion The described attitude error is expressed as:

[0030] (4);

[0031] Based on formula (4), the error quaternion The real and imaginary parts of , convert the error quaternion into the form of a rotation matrix:

[0032] (5);

[0033] Represent the error quaternion in order The values ​​of the 1st to 4th elements of the column vector;

[0034] Then convert the rotation matrix into Euler angle form:

[0035] (6);

[0036] Among them, the attitude error Expressed as , e roll represents the roll attitude error vector, e pitch represents the pitch attitude error vector, e yaw represents the yaw attitude error vector;

[0037] express The element at row 2 and column 1 of express The element in the first row and first column of , and the same goes for other elements; R 31 express The element at row 3 and column 1, R 32 express The element at row 3 and column 2, R 33 express The element at row 3 and column 3;

[0038] Then, in the attitude control space, the terminal angular velocity instruction of the robot end tracking the desired attitude is expressed as:

[0039] (7-1);

[0040] in, Represents the gain coefficient of the robot's end-point angular attitude error.

[0041] Optionally, the step of obtaining the motion instruction in the direction of the workpiece surface includes: obtaining the expected trajectory information of the robot, calculating the motion tracking error, and obtaining the motion instruction along the direction of the workpiece surface.

[0042] Optionally, obtaining the desired trajectory information of the robot, calculating the motion tracking error, and obtaining the motion instruction along the workpiece surface direction includes:

[0043] The expected derivatives of the robot's expected trajectory are , , the motion tracking error is:

[0044] ;

[0045] in, x is the actual position of the current robot end;

[0046] Then the motion instruction of the robot along the surface of the workpiece is:

[0047] (7-2);

[0048] in, Indicates the gain factor for position error.

[0049] Optionally, the step of acquiring the motion control instruction in the force control direction includes:

[0050] In the force control direction, a nonlinear discrete system model of the interactive force is constructed, and a compliant control strategy of model-free adaptive control is obtained to obtain the motion control instructions in the force control direction.

[0051] Optionally, the nonlinear discrete system model of the interaction force is:

[0052] (8);

[0053] in, 、 are the contact force and acceleration instructions of the force control direction of the robot system at time k, and are the orders of the robot system output and input, f represents an unknown nonlinear function;

[0054] The following performance index function is introduced into the nonlinear discrete system model of the interaction force for optimization:

[0055] (9);

[0056] in, is a control law penalty factor; For the expected output signal, substitute the data model into the performance index function and find the The derivative of and set it to zero, the design data-driven force control law is:

[0057] (10);

[0058] in, is the control law step size factor; is the pseudo partial derivative of the robot system .

[0059] Optionally, the following performance indicator function is used to estimate the pseudo partial derivative of the robot system online:

[0060] (11);

[0061] in, is the penalty factor for estimating pseudo partial derivatives;

[0062] Solve the equation about The derivative of and set it equal to zero, the following pseudo partial derivative online estimation algorithm is obtained:

[0063] (12);

[0064] in, is the estimated pseudo partial derivative step size factor;

[0065] Set up the following reset mechanism for the estimated pseudo-partial derivatives:

[0066] (13);

[0067] in, It is called the pseudo-partial derivative reset threshold; It is a pseudo partial derivative The initial value of

[0068] Based on the designed data-driven force control law, the motion control instructions in the force control direction are obtained:

[0069] (14);

[0070] Represents the update interval of the motion control instructions in the force control direction.

[0071] Optionally, the unified robot speed instruction is:

[0072] ;

[0073] in, ;

[0074] R d is the rotation matrix of the desired posture of the robot end, yes R dThe transposed moment of It is the movement instruction of the robot along the surface of the workpiece. It is the motion control instruction in the force control direction.

[0075] Compared with the prior art, the present invention has the following beneficial effects:

[0076] The present invention obtains the robot's motion instructions along the workpiece surface, the end angular velocity instructions for tracking the desired posture of the robot end, and the robot's motion control instructions in the force control direction from three aspects, thereby constructing a unified robot speed instruction to achieve efficient adaptation to the shape of the object surface set, while ensuring sufficient contact force accuracy and realizing high-precision force control. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1 This is the main flow chart of the robot compliant control method based on data-driven interactive process learning provided in the embodiment of the present application.

[0078] Figure 2 This is a complete process flow chart of the robot compliant control method based on data-driven interactive process learning provided in an embodiment of the present application. DETAILED DESCRIPTION

[0079] Example:

[0080] The technical solution of the present invention is further described below with reference to the accompanying drawings and embodiments.

[0081] See Figure 1 As shown, the robot compliance control method based on data-driven interactive process learning provided by this embodiment mainly includes the following steps:

[0082] a workpiece surface direction motion instruction acquisition step, obtaining the robot's motion instruction along the workpiece surface direction;

[0083] The posture error and posture motion instruction acquisition steps are used to obtain the end angular velocity instruction of the robot end tracking the desired posture

[0084] a motion control instruction acquisition step in the force control direction, obtaining the motion control instruction of the robot in the force control direction;

[0085] The normalized control instruction construction step constructs a unified robot speed instruction based on the robot's motion instruction along the workpiece surface, the end angular velocity instruction of the robot's end tracking the desired posture, and the robot's motion control instruction in the force control direction.

[0086] In summary, this method obtains the robot's motion instructions along the workpiece surface, the end angular velocity instructions of the robot's end tracking the desired posture, and the robot's motion control instructions in the force control direction from three aspects, thereby constructing a unified robot speed instruction to achieve efficient adaptation to the shape of the object surface set, while ensuring sufficient contact force accuracy and realizing high-precision force control.

[0087] In a specific embodiment, the posture error and posture motion instruction acquisition step includes: using a force sensor installed at the end of the robot to obtain the contact force between the current robot end and the uncertain workpiece, and based on this, obtaining the normal information of the contact point to obtain the rotation matrix of the desired posture of the robot end; in the posture control space, the posture error and posture motion instruction are calculated based on the rotation matrix of the desired posture.

[0088] More specifically, in the current robot end force sensor coordinate system, the contact force information is obtained , normalize the external force data fed back by the force sensor:

[0089] (1);

[0090] Among them, the vector Indicates the rotation angle of the robot end, Represents the norm of the contact force at the end of the robot. Based on the vector The antisymmetric characteristic matrix describing the robot's terminal posture can be constructed:

[0091] (2);

[0092] in, n (3) Represents a vector The third element in n (2) Represents a vector The second element in n (1) represents a vector The first element in ;

[0093] Then the rotation matrix of the desired posture of the robot end can be expressed as:

[0094] (3);

[0095] in, Represents the three-dimensional identity matrix.

[0096] Thus, through the above operations (i.e., formulas (1)-(3)), the contact surface normal information can be obtained and described in quaternion form.

[0097] In the attitude control space, the rotation matrix based on the desired attitude Design the angular velocity control variable. Based on this, convert the rotation matrix of the desired posture into quaternion form:

[0098] ;

[0099] in, Indicates the expected quaternion, r 11 express The element at row 1 and column 1, r 22 express The element at row 2 and column 2, r 33 express The element at row 3 and column 3, r 32 express The element at row 3 and column 2, r 23 express The element at row 2 and column 3, r 13 express The element in row 1 and column 3, r 31 express The element at row 3 and column 1, r 21 express The element at row 2 and column 1, r 12 express The element at row 1 and column 2; Represent the expected quaternions in order Column vector containing the values ​​of the first four elements.

[0100] The traditional method of calculating the rotation matrix may cause singular situations in the robot posture control, making it difficult to calculate usable control instructions in some specific situations. This application uses the above method to convert the rotation matrix of the desired posture into quaternion form, thereby effectively avoiding this problem.

[0101] Furthermore, since the rotation matrix of the desired posture is converted into quaternion form, the robot's current posture quaternion The described attitude error is expressed as:

[0102] (4);

[0103] That is, according to the current posture of the robot end, the error between the posture of the robot end and the normal information of the contact surface is obtained by quaternion operation (i.e., Equation 4).

[0104] Based on formula (4), the error quaternion The real and imaginary parts of , on this basis, the error quaternion is converted into the form of a rotation matrix:

[0105] (5);

[0106] Then convert the rotation matrix into Euler angle form:

[0107] (6);

[0108] Among them, the attitude error Expressed as ；e roll represents the roll attitude error vector, e pitch represents the pitch attitude error vector, e yaw represents the yaw attitude error vector;

[0109] express The element at row 2 and column 1 of express The element in the first row and first column of , and the same goes for other elements; R 31 express The element at row 3 and column 1, R 32 express The element at row 3 and column 2, R 33 express The element at row 3 and column 3;

[0110] Then the terminal angular velocity instruction of the robot's terminal tracking desired posture in space can be expressed as:

[0111] (7-1);

[0112] in, Represents the gain coefficient of the robot's end-point angular attitude error.

[0113] Thus, through the above operations, the attitude error described in the form of Euler angles is used to construct the robot attitude angular velocity instruction, which can align the normal direction of the contact surface with the direction of the robot end.

[0114] As can be seen, in the step of acquiring the posture error and posture motion command, this method uses the collected force information to characterize the normal of the uncertain contact surface. Then, based on the current orientation of the robot's end, it constructs the error between the contact surface normal and the robot's end orientation, and designs the posture control speed to align the two. At the same time, considering that traditional rotation matrix calculation methods may lead to singular situations in robot posture control, this method converts the rotation matrix of the desired posture into quaternion form, effectively avoiding this problem.

[0115] In a specific embodiment, the step of obtaining the workpiece surface direction motion instruction includes:

[0116] Design the robot's motion control strategy along the workpiece surface. Define the expected trajectory of the robot and its expected derivatives as follows: , , and define the motion tracking error as:

[0117] ;

[0118] The robot's motion reference instruction along the workpiece surface is designed as:

[0119] (7-2);

[0120] in, Indicates the gain factor for position error.

[0121] In a specific embodiment, the step of acquiring the motion control instruction in the force control direction includes:

[0122] In the force control direction, a nonlinear discrete system model of the interactive force is constructed, and a compliant control strategy of model-free adaptive control is obtained to obtain the motion control instructions in the force control direction.

[0123] In this step, this method obtains a compliant control strategy for model-free adaptive control (i.e., a data-driven compliant control strategy) by constructing a nonlinear discrete system model of the interaction force. This requires prior information about the robot's interaction object and combines the robot's control command data with the intrinsic information in the actual interaction force data to achieve high-precision force control.

[0124] More specifically, the interaction force contact process between the robot and the unknown workpiece is represented as a nonlinear discrete robot system in the following form:

[0125] (8);

[0126] in, 、 are the contact force and acceleration instructions of the force control direction of the robot system at time k, and are the orders of the robot system output and input, f represents an unknown nonlinear function.

[0127] The following performance indicator functions are introduced for optimization:

[0128] (9);

[0129] in, It is a control law penalty factor. The purpose of introducing this penalty factor is to limit the controller output The sudden change in the robot system is intuitively reflected in ensuring that the input signal curve of the robot system is relatively smooth. For the expected output signal, substitute the data model into the performance index function and find the The derivative of and set it to zero, the design data-driven force control law is:

[0130] (10);

[0131] in, is the control law step factor. The purpose of introducing this parameter is to make the control law algorithm more general and flexible. For nonlinear robot systems whose exact mathematical model is unknown, the pseudo partial derivative of the robot system It is also unknown and time-varying. Therefore, it is necessary to use the input and output data information of the controlled robot system to estimate the pseudo partial derivative of the robot system. The following performance indicator function is used to estimate the pseudo partial derivative of the robot system online:

[0132] (11);

[0133] in, To estimate the penalty factor of the pseudo partial derivative, the purpose of introducing this parameter is to prevent the sudden change of the pseudo partial derivative estimate of the robot system caused by external interference and other factors. The derivative of and set it equal to zero, we can get the following pseudo partial derivative online estimation algorithm:

[0134] (12);

[0135] in, To estimate the pseudo partial derivative step size factor. Finally, the following reset mechanism for the estimated pseudo partial derivative value is set to improve the algorithm's dynamic tracking performance for time-varying robot systems.

[0136] (13);

[0137] in, Is a relatively small number, called the pseudo-partial derivative reset threshold, It is a pseudo partial derivative The initial value of .

[0138] Based on the designed data-driven force control law, the motion control instructions in the force control direction can be obtained:

[0139] (14);

[0140] Represents the update interval of the motion control instructions in the force control direction.

[0141] In this way, through the above operations, a flexible control strategy (Equations (10) and (12)) can be obtained. Without the need for prior information about the robot's interaction object, high-precision force control can be achieved by combining the robot's control command data with the intrinsic information in the actual interaction force data.

[0142] Finally, a unified robot speed command can be constructed based on the robot's motion command along the workpiece surface, the robot's end tracking desired posture end angular velocity command, and the robot's motion control command in the force control direction. :

[0143] ;

[0144] in, ;

[0145] and The operator is an orthogonal decomposition operation.

[0146] It can be seen that this method, based on workpiece contour recognition and orthogonal decomposition of sensor force information, does not require the introduction of other expensive measurement equipment and has the advantages of low cost and easy implementation.

[0147] like Figure 2 As shown in FIG, it is a complete flow chart of the robot compliant control method of data-driven interactive process learning provided by this embodiment.

[0148] The above embodiments are intended only to illustrate the technical concepts and features of the present invention. Their purpose is to enable those skilled in the art to understand the contents of the present invention and implement them accordingly. They are not intended to limit the scope of protection of the present invention. Any equivalent changes or modifications made based on the essence of the present invention are intended to be covered by the scope of protection of the present invention.

Claims

1. A robot compliant control method based on data-driven interactive process learning, characterized in that: include: Workpiece a surface direction motion instruction acquisition step, obtaining a motion instruction of the robot along the surface direction of the workpiece; The posture error and posture motion instruction acquisition step obtains the end angular velocity instruction of the robot end for tracking the desired posture; a motion control instruction acquisition step in the force control direction, obtaining the motion control instruction of the robot in the force control direction; a normalized control instruction construction step, constructing a unified robot speed instruction based on the robot's motion instruction in the direction along the workpiece surface, the robot's end angular velocity instruction for tracking the desired posture, and the robot's motion control instruction in the force control direction; The unified robot speed instruction is: in, , , R d is the rotation matrix of the desired posture of the robot end, yes The transposed moment of It is the movement instruction of the robot along the surface of the workpiece. It is the motion control instruction in the force control direction.

2. The robot compliant control method based on data-driven interactive process learning according to claim 1, characterized in that: The step of acquiring the posture error and the posture motion instruction comprises: A force sensor installed at the end of the robot is used to obtain the contact force between the current robot end and the uncertain workpiece, and based on this, the normal information of the contact point is obtained to obtain the rotation matrix of the desired posture of the robot end; In the posture control space, the terminal angular velocity instruction for tracking the desired posture of the robot terminal is calculated based on the rotation matrix of the desired posture.

3. The robot compliance control method based on data-driven interactive process learning according to claim 2, characterized in that: The force sensor installed at the end of the robot is used to obtain the contact force between the current end of the robot and the uncertain workpiece, and based on this, the normal information of the contact point is obtained to obtain the rotation matrix of the desired posture of the end of the robot, including: In the current robot end force sensor coordinate system, based on the obtained contact force between the current robot end and the uncertain workpiece , normalize the external force data fed back by the force sensor: （1） Among them, the vector Indicates the rotation angle of the robot end, represents the norm of the contact force at the end of the robot; Vector-based Construct the antisymmetric characteristic matrix S that describes the robot's terminal posture: （2） Among them, n(3) represents the vector The third element in n(2) represents the vector The second element in n(1) represents the vector The first element in ; The rotation matrix R of the desired posture of the robot end d Expressed as: （3） in, Represents the three-dimensional identity matrix.

4. The robot compliance control method based on data-driven interactive process learning according to claim 2 or 3, characterized in that: The method of calculating, in the posture control space, the terminal angular velocity instruction for tracking the desired posture of the robot terminal based on the rotation matrix of the desired posture includes: In the attitude control space, the rotation matrix R based on the desired attitude of the robot end d To design the angular velocity control, convert the rotation matrix of the desired posture into quaternion form: in, Indicates the expected quaternion, r 11 express The element in row 1 and column 1, r 22 express The element in row 2 and column 2, r 33 express The element in row 3 and column 3, r 32 express The element in row 3 and column 2, r 23 express The element in row 2 and column 3, r 13 express The element in row 1 and column 3, r 31 express The element in row 3 and column 1, r 21 express The element in row 2 and column 1, r 12 express The element at row 1 and column 2; 、 、 、 Represent the expected quaternions in order The values ​​of the 1st to 4th elements of the column vector; By reading the robot's current posture quaternion The described attitude error is expressed as: （4） Based on formula (4), the error quaternion The real and imaginary parts of , convert the error quaternion into the form of a rotation matrix: （5） Represent the error quaternion in order The values ​​of the 1st to 4th elements of the column vector; Then convert the rotation matrix into Euler angle form: （6） Among them, the attitude error Expressed as , e roll represents the roll attitude error vector, e pitch represents the pitch attitude error vector, e yaw represents the yaw attitude error vector; express The element at row 2 and column 1 of express The element in the first row and first column of R, and the same goes for other elements; 31 express The element in row 3 and column 1, R 32 express The element in row 3 and column 2, R 33 express The element at row 3 and column 3; Then, in the attitude control space, the terminal angular velocity instruction of the robot end tracking the desired attitude is expressed as: （7） in, Represents the gain coefficient of the robot's end-point angular attitude error.

5. The robot compliant control method based on data-driven interactive process learning according to claim 1, characterized in that: The step of obtaining the motion instruction in the direction of the workpiece surface includes: obtaining the expected trajectory information of the robot, calculating the motion tracking error, and obtaining the motion instruction along the direction of the workpiece surface.

6. The robot compliance control method based on data-driven interactive process learning according to claim 5, characterized in that: The method of obtaining the desired trajectory information of the robot, calculating the motion tracking error, and obtaining the motion instruction along the surface direction of the workpiece includes: The expected derivatives of the robot's expected trajectory are , , the motion tracking error is: Among them, x is the actual position of the current robot end; Then the motion instruction of the robot along the surface of the workpiece is: （7-2） in, Indicates the gain factor for position error.

7. The robot compliance control method based on data-driven interactive process learning according to claim 1, characterized in that: The step of acquiring the motion control instruction in the force control direction includes: In the force control direction, a nonlinear discrete system model of the interactive force is constructed, and a compliant control strategy of model-free adaptive control is obtained to obtain the motion control instructions in the force control direction.

8. The robot compliance control method based on data-driven interactive process learning according to claim 7, characterized in that: The nonlinear discrete system model of the interaction force is: （8） in, 、 are the contact force and acceleration instructions of the force control direction of the robot system at time k, and are the orders of the output and input of the robot system, respectively, and f represents an unknown nonlinear function; The following performance index function is introduced into the nonlinear discrete system model of the interaction force for optimization: （9） in, is a control law penalty factor; For the expected output signal, substitute the data model into the performance index function and find the The derivative of and set it to zero, the design data-driven force control law is: （10） in, is the control law step size factor; is the pseudo partial derivative of the robot system .

9. The robot compliance control method based on data-driven interactive process learning according to claim 8, characterized in that: The following performance index function is used to estimate the pseudo partial derivative of the robot system online: （11） in, is the penalty factor for estimating pseudo partial derivatives; Solve the equation about The derivative of and set it equal to zero, the following pseudo partial derivative online estimation algorithm is obtained: （12） in, is the estimated pseudo partial derivative step size factor; Set up the following reset mechanism for the estimated pseudo-partial derivatives: （13） in, It is called the pseudo-partial derivative reset threshold; It is a pseudo partial derivative The initial value of Based on the designed data-driven force control law, the motion control instructions in the force control direction are obtained: （14） Represents the update interval of the motion control instructions in the force control direction.

Images