A joint angle selection method for parameter identification of a mechanical arm end six-dimensional force sensor

Abstract

The present application relates to a kind of mechanical arm end six-dimensional force sensor parameter identification joint angle selection method, for improving the parameter identification precision based on geometric model six-dimensional force sensor parameter identification sample selection, belong to parameter identification field.The present application includes: step 1: establishing identification equation;Step 2: according to the corresponding geometric shape, determine the change range of independent variable in identification equation;Step 3: set the first three joint angles in each identification sample;Step 4: under the premise of meeting the corresponding relationship between independent variable and joint angle, fix redundant joint angle;Step 5: parameter identification is carried out using the set sample.The present application solves the problem of expensive time cost caused by real-time acquisition of a large amount of different end posture data of mechanical arm.

Description

Technical Field

[0001] This invention relates to the field of parameter identification, specifically a method for selecting joint angles for parameter identification of a six-dimensional force sensor at the end of a robotic arm. Background Technology

[0002] The six-dimensional force sensor is installed at the front of the robotic arm's end effector and at the rear of the work tool. Even without contact, the six-dimensional force sensor will output force, which is generally determined by the weight of the work tool, temperature drift, and clamping force (see attached diagram). Figure 1 It consists of components such as temperature drift and clamping force. Temperature drift and clamping force are constant values ​​in the sensor coordinate system, denoted by bias. The component of the end-effector's gravity in the sensor coordinate system varies with different robot arm postures. To ensure that the sensor output force is only the contact force when the robot arm completes its task, this component of the force needs to be identified first.

[0003] Six-dimensional force sensor parameter identification is generally performed using the least squares method. Since the parameters to be identified reside in different equations, and the identification of later parameters relies on the results of earlier identifications, a parameter decoupling method is typically employed. Furthermore, obtaining accurate identification results using the least squares method often requires a large number of samples, which incurs significant time costs when the samples are derived from real-time data collected from different postures of the robotic arm. Summary of the Invention

[0004] This invention addresses the problem of high time costs associated with real-time acquisition of large amounts of data on different end-effector postures of a robotic arm. First, based on the geometry corresponding to each identification equation, the optimal selection range of the independent variables in each identification equation is determined. Then, according to the selection methods described in steps 2 and 3, the values ​​of the joint angles corresponding to the independent variables are determined.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] A method for selecting joint angles based on six-dimensional force sensor parameters at the end of a robotic arm is characterized by the following steps.

[0007] Step 1: Establish the identification equation;

[0008] Step 2: Determine the range of variation of the independent variable in the identification equation based on the corresponding geometric shape;

[0009] Step 3: Set the first three joint angles for each identified sample;

[0010] Step 4: Fix redundant joint angles while ensuring the correspondence between independent variables and joint angles;

[0011] Step 5: Parameter identification with the set of samples.

[0012] The identification equation is as follows:

[0013]

[0014] where is the contact force data collected by the six-dimensional force sensor in the sensor coordinate system It is related to the pose of the robot end link coordinate system Force deviation value f bias = (f x_bias , f y_bias , f z_bias , m x_bias , m y_bias , m z_bias ), the gravity G in the robot base coordinate system, the centroid position of the end tool q = (q x , q y , q z ), the sin (denoted by s) and cos (denoted by c) values between the bias angle between the sensor coordinate system and the end link coordinate system, as represented by the above identification equation; wherein the independent variable is the pose of the robot end link coordinate system R, the dependent variable is the contact force in the sensor coordinate system, and the to-be-identified parameter is the variable other than the two.

[0015] The equation of the independent variable R with respect to the 6 joint angles of the robot θ1, θ2, θ3, θ4, θ5, θ6 is as follows:

[0016] R 33 = -s (θ5) * s (θ2+θ3+θ4)

[0017] R 32 = c (θ6) * c (θ2+θ3+θ4) - c (θ5) * s (θ6) * s (θ2+θ3+θ4)

[0018] R 31 = s (θ6) * c (θ2+θ3+θ4) + c (θ5) * c (θ6) * s (θ2+θ3+θ4) (2)

[0019] is the inherent characteristic of the geometric shape type represented by the identification equation, and the optimal independent variable distribution required to uniquely represent a certain specific shape is obtained, wherein the independent variable is the pose component of the robot end link coordinate system.

[0020] The first three joint angles θ1, θ2, θ3 in each identification sample are defined as constant, which is used to ensure that the end position is prevented from fluctuating during the parameter identification process.

[0021] By fixing one of the last three redundant joint angles θ4, θ5, θ6, and keeping it unchanged, the other two redundant joint angles are set; by setting different values of the joint angles, the optimal distribution of the independent variables in each identification equation is obtained using the least number of free joint angles.

[0022] The specific values of the set of joints θ1, θ2, θ3, θ4, θ5, θ6 are brought into the forward kinematics formula (2) of the robot arm, the rotation matrix R is calculated, and then the independent variable R is brought into the identification equation (1), and each parameter is identified in turn.

[0023] The least square method is used to identify it during identification.

[0024] The present application has the following advantages and benefits:

[0025] The present application first determines the optimal selection range of the independent variable in each identification equation based on the corresponding geometric shape of each identification equation, then selects the method to determine the value of the joint angle corresponding to the independent variable, and finally identifies each parameter. The present application solves the problem of expensive time cost caused by real-time collection of a large amount of mechanical arm different end posture data. BRIEF DESCRIPTION OF DRAWINGS

[0026] Figure 1 It is a method flowchart;

[0027] Figure 2 It is a sensor coordinate system and a coordinate system direction force diagram;

[0028] Figure 3 It is a self-variable schematic diagram of circumferential sampling selected according to the geometric shape represented by the plane equation;

[0029] Figure 4 It is a joint angle instance diagram;

[0030] Among them, the first three joint angles are 107.4°, 15.88°, -15.77°, the fourth joint angle is set as a redundant joint angle, the value is -90.61, at the same time, the fifth and sixth joint angles are set to be equal and change in the range of -30° to 70°, the change of the independent variables R32 and R31 of the plane equation. DETAILED DESCRIPTION

[0031] The specific embodiments of the present application will be further described in detail below in combination with the drawings and examples. The following embodiments are used for the purpose of illustrating the present application, but do not limit the scope of the present application.

[0032] The six-dimensional force sensor is installed in front of the mechanical arm end link during use, and a work tool is installed at the back. When no contact occurs, the six-dimensional force sensor will also output force, which is generally composed of work tool gravity, temperature drift, tightening force (attached Figure 2 ) and the like. Among them, the temperature drift and the tightening force are constant values in the sensor coordinate system, represented by f bias , and the component of the end tool gravity in the sensor coordinate system will change with different postures of the mechanical arm. In order to make the sensor output force of the mechanical arm only be the contact force when completing the work, it is necessary to identify this part of the force first. The present application is a six-dimensional force sensor parameter identification sample selection method based on a geometric model, as shown in Figure 1 , and the specific implementation steps are as follows:

[0033] Step 1: Establish identification equation:

[0034] The six-dimensional force sensor collects contact force data in the sensor coordinate system , the posture of the robot end link coordinate system (matrix Θ represents that the position element has no effect on the posture, so the value is arbitrary), the force deviation value f bias =(f x_bias ,f y_bias ,f z_bias ,m x_bias ,m y_bias ,m z_bias ), the gravity G in the robot base coordinate system, the mass center position q of the end tool q=(q x ,q y ,q z ), and the sin (represented by s) and cos (represented by c) values of the angle between the sensor coordinate system and the end link coordinate system are related as follows: identification equation. Among these equations, the independent variable is the robot end link coordinate system posture R, the dependent variable is the contact force in the sensor coordinate system, and the to-be-identified parameter is the initial variable. Since the gravity G accounts for the main part of the sensor collected contact force data, and the gravity is mainly identified by the first three equations in the following equation set, therefore, the present application mainly sets the sample selection method for the first three equations in the six-dimensional force sensor parameter identification, and the least square method is used for identification during identification.

[0035]

[0036] Because the collaborative type mechanical arm is generally 6-axis, and the configuration is similar, UR5 type mechanical arm is taken as an example here, and the equation of the independent variable R with respect to the 6 joint angles of the mechanical arm is given, wherein the 6 joint angles are a group of identification samples actually selected, and are also the samples mentioned in the present application, represented by θ1, θ2, θ3, θ4, θ5, θ6.

[0037] R 33 = -s(θ5) * s(θ2 + θ3 + θ4)

[0038] R 32 = c(θ6) * c(θ2 + θ3 + θ4) - c(θ5) * s(θ6) * s(θ2 + θ3 + θ4)

[0039] R 31 = s(θ6) * c(θ2 + θ3 + θ4) + c(θ5) * c(θ6) * s(θ2 + θ3 + θ4) (2)

[0040] Step 2, determine the range of variation of the independent variable in the identification equation according to the geometric shape corresponding to the identification equation.

[0041] Take a two-dimensional equation corresponding to a two-dimensional shape as an example, explain how to determine the range of variation of the independent variable in the identification equation according to the geometric shape corresponding to the identification equation.

[0042] The two-dimensional equation is of the form y = ax1 + bx2 + c, which represents a specific spatial surface under specific parameters a, b, and c. It is known that a plane can be represented by three non-coplanar points. Therefore, when selecting independent variables, it is necessary to make them as evenly spaced as possible on the circumference, as shown in the following figure. Figure 3

[0043] After step 1, the range of values of the independent variable can be obtained, and it is also known that the independent variable and the actual selected sample, i.e. the joint angle of the robot arm, have a corresponding relationship described in the formula mentioned in step 1. Therefore, the purpose of steps 2 and 3 is to set the joint angle to meet the corresponding relationship while meeting the value of the independent variable obtained in step 1.

[0044] Step 3: Keep the first three joint angles unchanged in each identification sample.

[0045] First, fix the values of the first three joint angles θ1, θ2, and θ3 unchanged, and set fixed values for θ1, θ2, and θ3. Because the independent variable R of the identification equation has the greatest relationship with the last three joint angles of the robot arm, keeping the first three joint angles unchanged in each identification sample will not have a significant impact on the setting of the independent variable, and can also ensure that the end position does not fluctuate greatly during the parameter identification process, ensuring the safety of the movement process.

[0046] Step 4: Find and fix the redundant joint angle.

[0047] ​In order to satisfy the range of the variable R mentioned in step 1, the relationship between the variable mentioned in step 1 and the joint angle of the robot arm, i.e. the actual sample, needs to be satisfied first, and under the premise of satisfying the relationship, the joint angle of the robot arm can have different combinations to realize the value of the variable. In step 2, the values of the first three joint angles are specified, and the setting mode of the last three joint angles θ4, θ5, θ6 of the robot arm is unknown. Then the setting mode of θ4, θ5, θ6 will be selected: if there are multiple setting modes that can satisfy the value range requirement of the variable at the same time, and in one case, a certain redundant joint angle can be fixed to remain unchanged, and the other two redundant joint angles can be set.

[0048] For example, in one actual setting of the joint angle of the robot arm, the first three joint angles θ1, θ2, θ3 are fixed to 107.4°, 15.88°, -15.77° respectively, and the fourth joint angle θ4 is set as a redundant joint angle with a value of -90.61, and at the same time, the fifth and sixth joint angles θ5, θ6 are set to be equal and change in the range of -30° to 70°. At this time, the values of the above six joint angles θ1, θ2, θ3, θ4, θ5, θ6 are brought into formula (2), and the changes of the variables R32 and R31 of the plane equation can satisfy the change range requirement of the variable (R32, R31) ∈ [-1, 1]2, and the variable R33 of the one-dimensional straight line recognition equation also changes at equal intervals. Figure 4

[0049] Step 5: using the set sample to perform parameter identification.

[0050] According to the first four steps, a specific value of a set of joint angles θ1, θ2, θ3, θ4, θ5, θ6 can be set, and then they are brought into the forward kinematics of the robot arm to calculate the rotation matrix R, and then the variable R is brought into the identification equation to identify each parameter in turn.

[0051] The above is only a preferred embodiment of the present application, and does not limit the present application in any way. Any simple modification, change and equivalent structural change made according to the technical essence of the present application to the above embodiment are still within the protection scope of the technical solution of the present application.​

Claims

1.A joint angle selection method for parameter identification of a mechanical arm end six-axis force sensor, characterized by the following steps: Step 1: establishing an identification equation; the identification equation is as follows: （1） wherein is the contact force data collected by the six-dimensional force sensor in the sensor coordinate system , and the posture of the robot end link coordinate system , the force deviation value , the gravity G in the robot base coordinate system and the mass center position of the end tool , the relationship between the sin (denoted by s) and cos (denoted by c) values of the angle of deviation between the sensor coordinate system and the end link coordinate system is expressed as the above identification equation; wherein the independent variable is the posture R of the robot end link coordinate system, the dependent variable is the contact force in the sensor coordinate system, and the to-be-identified parameter is the variable other than the two. The argument R is a function of the 6 joint angles of the robot arm 6 The equation is as follows: （2） Step 2: determining the variation range of the independent variable in the identification equation according to the corresponding geometric shape; including: according to the inherent characteristics of the geometric shape type represented by the identification equation, obtaining the optimal independent variable distribution required to uniquely represent a certain specific shape, wherein the independent variable is the attitude component of the mechanical arm end link coordinate system; Step 3: Set the first three joint angles in each recognition sample; define the first three joint angles in each recognition sample Invariable, to ensure that the end position in the parameter identification process to avoid fluctuations; Step 4: fixing the redundant joint angles under the premise of meeting the correspondence between the independent variables and the joint angles; by fixing one of the last three redundant joint angles, keeping it unchanged, and setting the other two redundant joint angles; for achieving the optimal distribution of the independent variables in each identification equation using the least free joint angles by setting different values of the joint angles; Step 5: Parameter identification is performed using the set sample; a set of joint values is brought into the forward kinematics formula (2) of the mechanical arm, the rotation matrix R is calculated, and then the independent variable R is brought into the identification equation (1), and each parameter is identified in turn; the least square method is used for identification during identification.

Images