A human-computer interactive control method based on logarithmic sliding mode observer

Abstract

The invention discloses a human-machine interaction control method based on a logarithmic sliding mode observer, which consists of the following steps: step S1: obtaining dynamic information of human-machine interaction, performing data processing, and calculating the control law of the human-machine interaction process; step S2: calculating the rate of change of the adaptive law according to the data in step S1, and updating the adaptive gain; step S3: calculating the operation behavior estimation of the current cycle according to the observer algorithm and the control law in step S1; step S4: updating the adaptive observation parameters according to the operation behavior estimation in step S3, completing the closed-loop iteration of the algorithm, and ensuring the stability of system tracking control and operation behavior estimation under actual limited time conditions; the invention avoids the necessity of assembling expensive force / torque sensors in human-machine interaction, can quickly respond to the reference trajectory generated by the operation behavior, and at the same time ensures high steady-state accuracy, ensuring that the task process fully reflects the control intention of the operator.

Description

Technical Field

[0001] The present invention belongs to the technical field of human-robot collaborative interaction, and in particular relates to a human-robot interactive control method based on a logarithmic sliding mode observer. Background Art

[0002] Human-computer interaction technology actively adjusts the routine operation process of the robot arm through operation behavior to adapt to specific task requirements, avoid custom programming work, and reduce the cost of industrial processes. Usually, the trajectory formed by the movement of the robot arm needs to meet certain force / position requirements, which are obtained after analysis and design based on the needs of the established task. However, in the industrial process, the uncertainty of the control target or object is difficult to be fully considered in the pre-set task, resulting in the situation that the task effect implemented by the customized program is difficult to meet the requirements.

[0003] Human-machine interaction is an operation mode in which a human-assisted robot performs industrial tasks. The operation behavior is transmitted to the flexible design component through data acquisition devices such as force / torque sensors. After calculation, a new reference trajectory that meets the operator's force / position requirements is generated to guide the robot to perform corresponding movements. The direct purpose of human-machine interaction is to adjust the movement of the end of the robot arm. The commonly used schemes have two shortcomings. First, the tracking performance of the commonly used impedance control method is not suitable for scenarios with fast response and high precision requirements, resulting in a lag in the effect of human operation and unable to meet the task requirements. Second, the measurement of operation behavior relies on force / torque sensors, which increases the cost of the system and reduces the cost-effectiveness of the control system. Summary of the invention

[0004] The purpose of the present invention is to provide a human-computer interaction control method based on a logarithmic sliding mode observer to solve the problems of slow response, low precision and high cost of the existing human-computer interaction control method.

[0005] The present invention adopts the following technical solution: a human-computer interaction control method based on a logarithmic sliding mode observer, which consists of the following steps:

[0006] Step S1: Obtain the dynamic information of human-computer interaction, perform data processing, and calculate the control law of the human-computer interaction process.

[0007] Step S2: Calculate the rate of change of the adaptive law based on the data in step S1 and update the adaptive gain.

[0008] Step S3: Calculate the operating behavior estimate for the current cycle based on the observer algorithm and the control law in step S1,

[0009] Step S4: According to the operation behavior estimation in step S3, the adaptive observation parameters are updated to complete the closed-loop iteration of the algorithm to ensure the stability of system tracking control and operation behavior estimation under actual limited time conditions.

[0010] Furthermore, the operation method of step S1 is:

[0011] According to the information fed back by the system, the state of the system is obtained, including the rotation angle of each joint angle of the robot arm, and the state of the end is solved according to the known DH parameters and dynamic parameters, and the control law is designed according to the following formula:

[0012]

[0013] Among them, f c is the control force generated by the control strategy, M is the inertia matrix at the end of the robot arm, C is the Coriolis force matrix at the end of the robot arm, is the velocity of the end of the robot arm, G is the gravity vector at the end of the robot arm, K 1 , K 3 , and is a diagonal matrix with all diagonal elements being positive. is the time derivative of the tracking error, |e| represents the vector composed of the absolute values ​​of each element of the tracking error vector, 1 c is a vector whose elements are all 1, Indicates that the diagonal elements are K 1 |e|+1 c The inverse matrix of the diagonal matrix of vector elements, Indicates that vectors are placed on the diagonal elements The diagonal matrix of the corresponding elements of , is the estimated gain, x r is the reference trajectory under human-computer interaction conditions, is the acceleration of the reference trajectory, is the observed value of the operating behavior obtained by the observer, is the motion acceleration of the end of the robot arm, f h It is an operational behavior;

[0014] Where sgn(s) is a vector consisting of the sign terms of the elements of s, s is a sliding mode vector, and satisfies

[0015]

[0016] Among them, diag(sgn(e)) represents the identity matrix with the elements of sgn(e) arranged on the diagonal, ln(K 1 |e|+1 c ) indicates that each element is K 1|e|+1 c A vector of the values ​​of the logarithms of the elements of a vector.

[0017] Further, in step S2:

[0018] Update the adaptive parameters of the row control law According to the data in step S1, calculate the rate of change of the adaptive law and update the adaptive gain The specific update basis is as follows:

[0019]

[0020] Among them, K 5 is a diagonal matrix with all diagonal elements being positive. yes The updated adaptive law, ie, its time derivative, |s| is a vector consisting of the absolute values ​​of the elements of the sliding mode vector s.

[0021] Further, in step S3: wherein the observer structure is:

[0022]

[0023]

[0024] where ξ and ζ are observations of the system state and the derivative of the state, respectively. is the dynamics part of the system, is the overall input to the observer, is the equivalent input term of the observer, where and is a diagonal matrix with all diagonal elements being positive. represents the observation error, is the time derivative of the observation error, The diagonal elements are The inverse matrix of the diagonal matrix of vector elements, K 2 is a diagonal matrix with all diagonal elements being positive;

[0025] Among them, the operator behavior estimation in the observer input It is realized by logarithmic sliding mode. The specific description is:

[0026]

[0027] in, Indicates that vectors are placed on the diagonal elements The diagonal matrix of the corresponding elements of , is the estimated gain, and is a diagonal matrix with all diagonal elements being positive. Indicates that the diagonal line is placed The identity matrix of each element, Represents each element as vector of the logarithmic values ​​of each element of the vector, K 7 is a diagonal matrix with all diagonal elements being positive.

[0028] Further, in step S4: adaptively observe the parameters Update, The time derivative satisfies the following relationship:

[0029]

[0030] Among them, K 6 is a diagonal matrix with all diagonal elements being positive.

[0031] The beneficial effects of the present invention are as follows: the present invention can make a more accurate numerical estimate of the operating behavior by reducing the observation error of the observer, thereby avoiding the necessity of assembling expensive force / torque sensors in human-computer interaction, and designs a logarithmic sliding mode control method that can take both dynamic performance and steady-state performance into consideration, which can quickly respond to the reference trajectory generated by the operating behavior while ensuring a high steady-state accuracy, thereby ensuring that the task process fully reflects the operator's control intention. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 The reference trajectory tracking simulation effect of the first motion dimension of the present invention;

[0033] Figure 2 The reference trajectory tracking simulation effect of the second motion dimension of the present invention;

[0034] Figure 3 The reference trajectory tracking simulation effect of the third motion dimension of the present invention;

[0035] Figure 4 It is the operation behavior estimation of the first motion dimension of the present invention;

[0036] Figure 5 The operation behavior estimation of the second motion dimension of the present invention;

[0037] Figure 6 This is the operational behavior estimation of the third motion dimension of the present invention. DETAILED DESCRIPTION

[0038] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0039] The present invention discloses a human-machine interactive control method based on a logarithmic sliding mode observer. The logarithmic sliding mode observer is used to estimate the operation behavior, the motion trajectory of the end of the robot arm is replanned using a flexible design component, and the end motion is controlled according to the driving force generated by the proposed algorithm. The entire interactive process can ensure that the tracking error of the control system converges in an actual finite time.

[0040] Compared with the existing methods, the present invention isolates the operation behavior by using the observer structure, integrates the relevant observer based on the logarithmic sliding mode structure, and ensures the application of the logarithmic control strategy with actual finite time stability in the human-computer interaction process.

[0041] The dynamic model of the human-computer interaction process can be described by the following equation:

[0042]

[0043] in, is the motion acceleration of the end of the robot arm, is the velocity of the end of the robot arm, M is the inertia matrix of the end of the robot arm, C is the Coriolis force matrix of the end of the robot arm, G is the gravity vector of the end of the robot arm, and f c is the control force generated by the control strategy, i.e., the control input, f h The above model describes the three-degree-of-freedom motion of the end of the robot in Cartesian space.

[0044] The tracking error of the controller is expressed as e=xx r , where x is the displacement of the end of the robot, x r It is the reference trajectory under the condition of human-computer interaction. The flexible design component under the action of operation behavior, that is, the reference trajectory generator can be expressed as:

[0045]

[0046] in, is the acceleration of the reference trajectory, is the velocity of the reference trajectory, is the mass matrix of the desired impedance structure, is the damping matrix of the desired impedance structure, is the elastic matrix of the desired impedance structure, and formula (2) can be obtained by solving the differential equation to obtain the reference trajectory x r .

[0047] Obtain the dynamic information of human-computer interaction, perform data processing, and calculate the control law of the human-computer interaction process. First, according to the information fed back by the system, obtain the state of the system, including the rotation angle of each joint angle of the robot arm, and solve the state of the end according to the known DH parameters and dynamic parameters, and design the control law according to the following formula,

[0048]

[0049] Where M is the inertia matrix at the end of the robot arm, C is the Coriolis force matrix at the end of the robot arm, is the velocity of the end of the robot arm, G is the gravity vector at the end of the robot arm, K 1 , K 3 , and is a diagonal matrix with all diagonal elements being positive. is the time derivative of the tracking error, |e| represents the vector composed of the absolute values ​​of each element of the tracking error vector, 1 c is a vector whose elements are all 1, Indicates that the diagonal elements are K 1 |e|+1 c The inverse matrix of the diagonal matrix of vector elements, Indicates that vectors are placed on the diagonal elements The diagonal matrix of the corresponding elements of , is the estimated gain, is the acceleration of the reference trajectory, is the observed value of the operating behavior obtained by the observer, is the motion acceleration of the end of the robot arm, f h is the operation behavior, sgn(s) is a vector whose elements are the symbolic terms of each element of s, and s is a sliding mode vector that satisfies:

[0050]

[0051] Among them, diag(sgn(e)) represents the identity matrix with the elements of sgn(e) arranged on the diagonal, ln(K 1 |e|+1 c ) indicates that each element is K 1 |e|+1 c A vector of the values ​​of the logarithms of the elements of a vector.

[0052] According to the above data, the rate of change of the adaptive law is calculated and the adaptive gain is updated. The specific update basis is as follows:

[0053]

[0054] Among them, K 5is a diagonal matrix with all diagonal elements being positive. yes The updated adaptive law, ie, its time derivative, |s| is a vector consisting of the absolute values ​​of the elements of the sliding mode vector s.

[0055] According to the observer algorithm, the estimation of the operating behavior is calculated. According to the control law, the estimation of the operating behavior of the current cycle can be calculated according to the following observer algorithm. is an estimate of the operating behavior obtained by the logarithmic sliding mode observer. The algorithm of the logarithmic sliding mode observer is as follows:

[0056]

[0057]

[0058] where ξ and ζ are observations of the system state and the derivative of the state, respectively. is the dynamics part of the system, is the overall input to the observer, is the equivalent input term of the observer, where and is a diagonal matrix with all diagonal elements being positive. represents the observation error, is the time derivative of the observation error, The diagonal elements are The inverse matrix of the diagonal matrix of vector elements, K 2 is a diagonal matrix with all diagonal elements being positive, and the operator behavior estimate in the observer input It is realized by logarithmic sliding mode. The specific description is:

[0059]

[0060] in, Indicates that vectors are placed on the diagonal elements The diagonal matrix of the corresponding elements of , is the estimated gain, and is a diagonal matrix with all diagonal elements being positive. Indicates that the diagonal line is placed The identity matrix of each element, Represents each element as vector of the logarithmic values ​​of each element of the vector, K 7 It is a diagonal matrix whose diagonal elements are all positive numbers, and its time derivative satisfies the following relationship

[0061]

[0062] Among them, K 6 is a diagonal matrix with all diagonal elements being positive.

[0063] After completing the above steps, return to the first step to complete the closed-loop iteration of the algorithm, which can ensure the stability of system tracking control and operation behavior estimation under actual finite time conditions.

[0064] Example 1

[0065] The control method of the present invention is used to perform human-machine interactive control of the three-degree-of-freedom robotic arm in three motion dimensions. The motion state of the three-degree-of-freedom robotic arm can be adjusted by introducing the operator's behavior. The motion trajectory of the controlled robotic arm is shown in FIG. Figure 1-Figure 3 As shown. Figure 1-3 It can be seen that compared with the traditional sliding mode control method, the method proposed in this invention can quickly realize the reference trajectory tracking under human-computer interaction conditions, and the tracking error does not exceed 1×10 -4 meters, and under the condition of operational behavior injection, such as Figure 1 15 seconds, Figure 2 As shown in Figure 5, the tracking accuracy can be maintained at 1×10 -4 The error of the traditional method is as high as 0.01 meter. In the case of continuous operation behavior injection, such as Figure 3 As shown in the figure 5-10 seconds, the tracking performance of the traditional method is significantly reduced under the influence of the operation behavior, and the error is generally around 0.01 meters. However, the proposed method can maintain a tracking accuracy of no more than 1×10 -4 Therefore, the method proposed in the present invention is superior to the conventional sliding mode control strategy based on observer.

[0066] In the process of controlling human-machine interaction using the control method of the present invention, the observer can be used to observe the behavior of the operator at the same time. The observed trajectory result is as follows: Figure 4-Figure 6 As shown. Figure 4-6 It can be seen that the force estimation accuracy of the algorithm for input behavior is not less than 0.3N, and the operation behavior injection can be quickly detected within 0.3 seconds and quickly estimated. Therefore, the logarithmic sliding mode based observer proposed in the present invention can accurately estimate the operation behavior with satisfactory estimation accuracy.

[0067] Compared with the existing methods, the present invention uses a logarithmic sliding mode observer to estimate the operating behavior, uses a flexible design component to re-plan the motion trajectory of the end of the robot arm, and controls the end motion according to the driving force generated by the proposed algorithm. The entire interactive process can ensure that the tracking error of the control system converges in an actual finite time.

[0068] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A human-computer interaction control method based on logarithmic sliding mode observer, characterized in that: It consists of the following steps: Step S1: Obtain the dynamic information of human-computer interaction, perform data processing, and calculate the control law of the human-computer interaction process. Step S2: Calculate the rate of change of the adaptive law based on the data in step S1 and update the adaptive gain. Step S3: Calculate the operating behavior estimate for the current cycle based on the observer algorithm and the control law in step S1, Step S4: According to the operation behavior estimation in step S3, the adaptive observation parameters are updated to complete the closed-loop iteration of the algorithm to ensure the stability of system tracking control and operation behavior estimation under actual limited time conditions; The operation method of step S1 is: According to the information fed back by the system, the state of the system is obtained, including the rotation angle of each joint angle of the robot arm, and the state of the end is solved according to the known DH parameters and dynamic parameters, and the control law is designed according to the following formula: Among them, f c is the control force generated by the control strategy, M is the inertia matrix at the end of the robot arm, C is the Coriolis force matrix at the end of the robot arm, is the movement speed of the end of the robot arm, G is the gravity vector at the end of the robot arm, K1, K3, and is a diagonal matrix with all diagonal elements being positive. is the time derivative of the tracking error, |e| represents the vector composed of the absolute values ​​of each element of the tracking error vector, 1 c is a vector whose elements are all 1, Indicates that the diagonal elements are K1|e|+1 c The inverse matrix of the diagonal matrix of vector elements, Indicates that vectors are placed on the diagonal elements The diagonal matrix of the corresponding elements of , is the estimated gain, x r is the reference trajectory under human-computer interaction conditions, is the acceleration of the reference trajectory, is the observed value of the operation behavior obtained by the observer, is the motion acceleration of the end of the robot arm, f h It is an operational behavior; Where sgn(s) is a vector consisting of the sign terms of the elements of s, s is a sliding mode vector, and satisfies Where diag(sgn(e)) represents the identity matrix with the elements of sgn(e) placed on the diagonal, ln(K1|e|+1 c ) means that each element is K1|e|+1 c A vector of the values ​​of the logarithms of each element of the vector; In step S2: Update the adaptive parameters of the row control law According to the data in step S1, calculate the rate of change of the adaptive law and update the adaptive gain The specific update basis is as follows: Among them, K5 is a diagonal matrix with all diagonal elements being positive numbers. yes The updated adaptive law, i.e. its time derivative, |s| is a vector consisting of the absolute values ​​of the elements of the sliding mode vector s; In step S3: the observer structure is: where ξ and ζ are observations of the system state and the derivative of the state, respectively. is the dynamics part of the system, is the overall input to the observer, is the equivalent input term of the observer, where and is a diagonal matrix with all diagonal elements being positive. represents the observation error, is the time derivative of the observation error, The diagonal elements are K2 is the inverse matrix of the diagonal matrix of the vector elements, and all diagonal elements of the matrix are positive numbers. Among them, the operator behavior estimation in the observer input It is realized by logarithmic sliding mode. The specific description is: in, Indicates that vectors are placed on the diagonal elements The diagonal matrix of the corresponding elements of , is the estimated gain, and is a diagonal matrix with all diagonal elements being positive. Indicates that the diagonal line is placed The identity matrix of each element, Represents each element as A vector of the logarithms of each element of a vector. K7 is a diagonal matrix whose diagonal elements are all positive.

2. The human-computer interaction control method based on logarithmic sliding mode observer according to claim 1 is characterized in that: In step S4: Adaptive observation parameters Update, The time derivative satisfies the following relationship: Among them, K6 is a diagonal matrix whose diagonal elements are all positive numbers.

Images