A parallel robot optimization method

Abstract

The present invention relates to the technical field of industrial robots, and in particular to a parallel robot optimization method, including determining the configuration of the parallel robot, and selecting visual sensors and visual servo image features for the parallel robot; establishing a visual geometric projection model of the parallel robot based on the visual sensor, and performing first-order differentiation on the projection model to obtain an image Jacobian matrix; constructing a visual servo positioning error model based on the image Jacobian matrix; avoiding the first-class singularity, the second-class singularity and the visual servo controller singularity in the accessible workspace to obtain a smart workspace; based on the visual servo positioning error model and the size of the smart workspace, the visual sensor parameters and the parallel robot geometric size parameters are combined and optimized to obtain an optimal visual sensor plus parallel robot combination scheme. The problem of low control accuracy of the existing parallel robot is solved.

Description

Technical Field

[0001] The invention belongs to the technical field of industrial robots, and in particular relates to a parallel robot optimization method. Background Art

[0002] Compared with serial robots, parallel robots can achieve higher speed, high acceleration movement, higher precision, higher load-to-weight ratio and other advantages. Therefore, parallel robots are widely used in the field of industrial robots. Among them, the most successful parallel configurations in commercial applications are Delta configuration, dual SCARA configuration and Gough-Stewart configuration. Taking into account the industrial manufacturing cost of robots, parallel robots need to reduce the size of the robot as much as possible and obtain the highest possible control accuracy under the premise of meeting the given dexterous workspace. Therefore, it is particularly necessary to optimize the design of industrial parallel robots accordingly.

[0003] Although there are a large number of literatures and research results on the control problem of parallel robots, their high structural complexity and complex nonlinear relationship between control input and output increase the difficulty of precise control. Traditional parallel robot control methods are based on robot model methods with built-in sensors. Their control accuracy is heavily dependent on extremely accurate robot models. In practical applications, due to manufacturing errors, assembly error accumulation and other reasons, accurate robot models cannot guarantee the control accuracy of parallel robots. Therefore, the use of visual servo control methods based on external visual sensors to control parallel robots can bypass the complex structural model of parallel robots and obtain excellent control accuracy.

[0004] However, when using the visual servo control method, the type and position of the visual sensor, the type of visual target features to be observed, etc. will affect the final control accuracy. Therefore, simply optimizing the design of the parallel robot cannot guarantee the best control accuracy. Summary of the invention

[0005] The purpose of the present invention is to provide a parallel robot optimization method, aiming to solve the problem of low control accuracy of existing parallel robots.

[0006] In order to achieve the above technical purpose, the technical solution adopted by the present invention is as follows: comprising the following steps:

[0007] Determine the configuration of the parallel robot and select a visual sensor for the parallel robot;

[0008] A visual geometric projection model of the parallel robot is established based on the visual sensor, and the visual geometric projection model is first differentiated to obtain an image Jacobian matrix and a kinematic equation of the parallel robot based on visual servoing;

[0009] Constructing a visual servo positioning error model based on the image Jacobian matrix;

[0010] The first type singularity, the second type singularity and the visual servo controller singularity are avoided in the achievable workspace to obtain a smart workspace.

[0011] Based on the visual servo positioning error model and the dexterous workspace, a combined optimization design is performed to obtain an optimal visual sensor plus parallel robot combination solution.

[0012] Wherein, the determining of the configuration of the parallel robot and selecting a visual sensor for the parallel robot comprises:

[0013] Determine and optimize the driving joints and degrees of freedom of the parallel robot;

[0014] Determining the type and number of visual servoing image features to be observed based on the degrees of freedom;

[0015] A visual sensor is selected for the parallel robot based on the type and the number.

[0016] The visual geometric projection model of the parallel robot is established based on the visual sensor, and the first-order differentiation of the geometric projection model is performed to obtain the image Jacobian matrix and the visual servo kinematic equation, including:

[0017] Establishing a sensor coordinate system with the visual sensor as the origin of the coordinate system;

[0018] Based on the sensor coordinate system, an imaging image of the image feature in the camera is generated by using the pinhole imaging principle to obtain a visual geometric projection model;

[0019] The visual geometric projection model is differentiated in first order to obtain the image Jacobian matrix and the visual servoing kinematic equation.

[0020] Wherein, constructing a visual servo positioning error model based on the image Jacobian matrix includes:

[0021] Using the image Jacobian matrix to describe the relationship between image noise and displacement, a description relationship is obtained;

[0022] A visual servo positioning error model is constructed based on the description relationship.

[0023] Wherein, the method of avoiding the first type of singularity, the second type of singularity and the visual servo controller singularity in the reachable workspace to obtain a smart workspace includes:

[0024] Calculating a reachable workspace of the parallel robot;

[0025] Through robot forward kinematics, a Rand function is used to generate a random value as a random variable within the motion range of each of the driving joints;

[0026] Based on the random variables, the corresponding position and posture of the terminal are obtained by forward kinematics calculation;

[0027] Obtaining a reachable working space range based on the posture corresponding to the terminal;

[0028] Based on the reachable working space range, the image Jacobian matrix is ​​calculated to obtain the first kind singularity, the second kind singularity and the visual servo controller singularity;

[0029] The first type of singularity, the second type of singularity and the visual servo controller singularity are avoided within the achievable workspace to obtain a dexterous workspace.

[0030] The combined optimization design based on the visual servo positioning error model and the dexterous workspace to obtain the optimal visual sensor plus parallel robot combination scheme includes:

[0031] Establish the objective function;

[0032] Determining optimization design parameters based on the objective function, wherein the optimization design parameters include visual sensor parameters and parallel robot size parameters;

[0033] Establishing a global performance evaluation index based on the optimized design parameters, the visual servo positioning error model and the dexterous workspace;

[0034] Based on the global performance evaluation index, discrete nodes are set in the reachable workspace of the parallel robot, and the average value of the discrete node function value is calculated to obtain the optimal visual sensor plus parallel robot combination solution.

[0035] The present invention provides a parallel robot optimization method, which determines the configuration of the parallel robot and selects a visual sensor for the parallel robot; establishes a visual geometric projection model of the parallel robot based on the visual sensor, and performs first-order differentiation on the visual geometric projection model to obtain an image Jacobian matrix and a kinematic equation of the parallel robot based on visual servoing; constructs a visual servo positioning error model based on the image Jacobian matrix; avoids the first-class singularity, the second-class singularity and the visual servo controller singularity within the reachable workspace to obtain a smart workspace; performs combined optimization design based on the visual servo positioning error model and the smart workspace to obtain an optimal visual sensor plus parallel robot combination scheme. The present invention regards the robot and the sensor as a mutually coordinated whole, and optimizes their parameters accordingly, which can ensure the best control effect and solve the problem of low control accuracy of the existing parallel robot. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The present invention is further illustrated by means of the non-limiting examples given in the accompanying drawings.

[0037] Figure 1 The present invention provides a flowchart of a parallel robot optimization method.

[0038] Figure 2 A flow chart for determining the configuration of a parallel robot and selecting a visual sensor for the parallel robot.

[0039] Figure 3 A visual geometric projection model of the parallel robot is established based on the visual sensor, and the visual geometric projection model is first-order differentiated to obtain an image Jacobian matrix and a flow chart of the kinematic equations of the parallel robot based on visual servoing.

[0040] Figure 4 The present invention is a flowchart for constructing a visual servo positioning error model based on the image Jacobian matrix.

[0041] Figure 5 In order to avoid the first type of singularities, the second type of singularities and the visual servo controller singularities within the achievable workspace, a flow chart of the dexterous workspace is obtained.

[0042] Figure 6 A flowchart of the optimal visual sensor plus parallel robot combination scheme is obtained for combined optimization design based on the visual servo positioning error model and the dexterous workspace.

[0043] Figure 7 A schematic diagram of a flow chart of a parallel robot optimization method provided by the present invention. DETAILED DESCRIPTION

[0044] In order to enable those skilled in the art to better understand the present invention, the technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0045] like Figure 1-7 As shown, the present invention provides a parallel robot optimization method, comprising the following steps:

[0046] S1 determines the configuration of the parallel robot and selects a visual sensor for the parallel robot;

[0047] The specific method is:

[0048] S11 determines the driving joints and degrees of freedom of the generation-optimized parallel robot;

[0049] Specifically, it is necessary to first determine the type of parallel robot to be optimized (how many driven joints and how many degrees of freedom it has).

[0050] S12 determines the type and number of visual servoing image features to be observed based on the degrees of freedom;

[0051] Specifically, according to the optimized degrees of freedom of the parallel robot, the types and numbers of the features of the visual servoing image to be observed are determined, and the number of the image features is ensured to be greater than the degrees of freedom of the parallel robot to ensure that full control can be achieved.

[0052] S13 selects a visual sensor for the parallel robot based on the type and the number.

[0053] S2: establishing a visual geometric projection model of the parallel robot based on the visual sensor, and performing first-order differentiation on the visual geometric projection model to obtain an image Jacobian matrix and a kinematic equation of the parallel robot based on visual servoing;

[0054] The specific method is:

[0055] S21 establishes a sensor coordinate system with the visual sensor as the origin of the coordinate system;

[0056] S22 generates an imaging image of the image feature in the camera based on the sensor coordinate system through the pinhole imaging principle to obtain a visual geometric projection model;

[0057] Specifically, a sensor coordinate system is established with the visual sensor as the origin of the coordinate system, and the imaging image of the image feature in the camera is obtained through the pinhole imaging principle, which is a mapping model from the parallel robot task space to the image feature space.

[0058] S23 performs first-order differentiation on the visual geometric projection model to obtain an image Jacobian matrix and a visual servoing kinematic equation.

[0059] Specifically, by taking the first-order differential of this model, we can obtain the image Jacobian matrix of the robot terminal velocity to the image feature change speed in the image space: (in Represents the speed of image feature change in image space, τ c represents the terminal velocity of the parallel robot in the sensor coordinates, and L is the image Jacobian matrix based on the image s.

[0060] S3 constructs a visual servo positioning error model based on the image Jacobian matrix;

[0061] The specific method is:

[0062] S31 uses the image Jacobian matrix to describe the relationship between image noise and displacement to obtain a description relationship;

[0063] Specifically, a visual servo positioning error model is established. Since the observation noise of the visual sensor is the direct cause of the visual servo positioning error, the performance of the selected visual sensor is analyzed and the noise range (i.e., ±p pixels) is selected. Since the noise of the visual sensor is extremely small, the model equation from the robot terminal speed obtained in step 1 to the image feature change speed in the image space can be used: To describe the relationship between extremely small image noise (i.e., image variation) and the small displacement at the end of the parallel robot (i.e., the parallel robot visual servo positioning error).

[0064] S32 constructs a visual servo positioning error model based on the description relationship.

[0065] Specifically, the positioning error model Δx = L for image observation noise in visual servoing is obtained + Δp (where Δp is the selected image observation noise, here ±p pixels, L + is the pseudo-inverse of the image Jacobian matrix, and Δx is the calculated positioning error of the parallel robot visual servo).

[0066] S4 avoids the first type of singularity, the second type of singularity and the visual servo controller singularity in the reachable workspace and obtains a smart workspace;

[0067] The specific method is:

[0068] S41 calculates the reachable working space of the parallel robot;

[0069] Specifically, the workspace of the parallel robot is divided into a reachable space and a dexterous workspace. First, the reachable workspace of the selected parallel robot to be optimized is calculated. For a parallel robot with low degrees of freedom (degrees of freedom not greater than 3), the geometric analysis method can be used to obtain its reachable workspace, that is, the posture trajectory of a single leg of the parallel robot when it is fully deployed is the reachable workspace of the parallel robot; for a parallel robot with high degrees of freedom (degrees of freedom greater than 3), the Monte Carlo method is used to numerically analyze its reachable workspace.

[0070] S42 generates random values ​​as random variables within the motion range of each driving joint using a Rand function through robot forward kinematics;

[0071] S43 obtains the corresponding position and posture of the terminal through forward kinematics calculation based on the random variable;

[0072] S44 obtains a reachable working space range based on the posture corresponding to the terminal;

[0073] S45 obtains a first-class singularity, a second-class singularity and a visual servo controller singularity based on the reachable workspace range by calculating the image Jacobian matrix;

[0074] Specifically, we first need to establish the kinematic model of the parallel robot in is the driving joint velocity of the parallel robot, and τ is the terminal velocity of the parallel robot. When the determinant of matrix A det(A) = 0, the parallel robot encounters the first type of singular posture, and the parallel robot loses one or more degrees of freedom. When the determinant of matrix B det(B) = 0, the parallel robot encounters the second type of singular posture, and the parallel robot gains one or more degrees of freedom. According to the numerical values ​​of the two determinants, the distance between the current posture and various types of singular postures can be determined. At the same time, in visual servoing, the system loses stability and the control accuracy deteriorates due to the singularity of the controller, so this kind of situation should be avoided. According to the image Jacobian matrix L obtained in step 1, the inverse of the condition number is calculated to obtain the condition for the loss of rank of the matrix L. According to the size of the inverse of its condition number, the distance between the current posture and the singular point of the controller can be evaluated.

[0075] S46 avoids the first type of singularity, the second type of singularity and the visual servo controller singularity within the achievable workspace to obtain a dexterous workspace.

[0076] S5 performs combined optimization design based on the visual servo positioning error model and the dexterous workspace to obtain an optimal visual sensor plus parallel robot combination solution.

[0077] The specific method is:

[0078] S51 establishes the objective function;

[0079] Specifically, an objective function is established. For this optimization method, the purpose is to design the smallest parallel robot, so the objective function is the volume of this parallel robot, and the optimization target is min(V), where V is the volume of the parallel robot to be optimized.

[0080] S52 determines the optimization design parameters based on the objective function;

[0081] Specifically, the optimization design parameter x is determined. In this optimization design, the optimization parameters are divided into two parts. One part is the geometric parameters of the parallel robot (all parameters affecting the geometric dimensions of the parallel robot, such as the bottom radius, the length of each link, and the radius of the end effector), and the second part is the sensor parameters (including the spatial coordinates of the sensor, angle, and characteristic dimensions of the observed image, etc.). The optimization design parameters include visual sensor parameters and parallel robot size parameters.

[0082] S53 establishes a global performance evaluation index based on the optimization design parameters, the visual servo positioning error model and the dexterous workspace;

[0083] Specifically, the global performance evaluation indicators are determined (including the positioning accuracy of the parallel robot at a given position, the distance of the parallel robot from the first kind of singularity, the second kind of singularity, and the controller singular position).

[0084] Δx≤x max (Δx is the value calculated by the visual servo positioning error model in step 2, x max is the maximum positioning error set in advance);

[0085] det(A)>dAmin (det(A) is the determinant value of the current posture matrix A of the parallel robot, dAmin is the set minimum value of the determinant, and the restriction here can ensure that the parallel robot is away from the first type of singular posture);

[0086] det(B)>dBmin (det(B) is the determinant value of the current posture matrix B of the parallel robot, dBmin is the set minimum value of the determinant, and the restriction here can ensure that the parallel robot is away from the second type of singular posture);

[0087] k -1 (L)>10 -3 (where k -1 (L) is the inverse condition number norm of the image Jacobian matrix L. The restriction here can ensure that the visual servo parallel robot stays away from the singular pose of the controller).

[0088] S54 sets discrete nodes in the reachable workspace of the parallel robot based on the global performance evaluation index, calculates the average value of the discrete node function value, and obtains the optimal visual sensor plus parallel robot combination solution.

[0089] The above embodiments are merely illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. Anyone familiar with the technology may modify or change the above embodiments without violating the spirit and scope of the present invention. Therefore, all equivalent modifications or changes made by those with ordinary knowledge in the technical field without departing from the spirit and technical ideas disclosed by the present invention should still be covered by the claims of the present invention.

Claims

1. A parallel robot optimization method, It includes, characterized in that it includes the following steps: Determine the configuration of the parallel robot and select a visual sensor for the parallel robot; A visual geometric projection model of the parallel robot is established based on the visual sensor, and the visual geometric projection model is first differentiated to obtain an image Jacobian matrix and a kinematic equation of the parallel robot based on visual servoing; Constructing a visual servo positioning error model based on the image Jacobian matrix; The first type singularity, the second type singularity and the visual servo controller singularity are avoided in the achievable workspace to obtain a smart workspace. Based on the visual servo positioning error model and the dexterous workspace, a combined optimization design is performed to obtain an optimal visual sensor plus parallel robot combination solution.

2. The parallel robot optimization method according to claim 1, characterized in that: The step of determining the configuration of the parallel robot and selecting a visual sensor for the parallel robot includes: Determine and optimize the driving joints and degrees of freedom of the parallel robot; Determining the type and number of visual servoing image features to be observed based on the degrees of freedom; A visual sensor is selected for the parallel robot based on the type and the number.

3. The parallel robot optimization method according to claim 2, characterized in that: The method of establishing a visual geometric projection model of the parallel robot based on the visual sensor and performing first-order differentiation on the visual geometric projection model to obtain an image Jacobian matrix and a kinematic equation of the parallel robot based on visual servoing includes: Establishing a sensor coordinate system with the visual sensor as the origin of the coordinate system; Based on the sensor coordinate system, an imaging image of the image feature in the camera is generated by using the pinhole imaging principle to obtain a visual geometric projection model; The visual geometric projection model is differentiated in first order to obtain the image Jacobian matrix and the visual servoing kinematic equation.

4. The parallel robot optimization method according to claim 3, characterized in that: The constructing of a visual servo positioning error model based on the image Jacobian matrix includes: Using the image Jacobian matrix to describe the relationship between image noise and displacement, a description relationship is obtained; A visual servo positioning error model is constructed based on the description relationship.

5. The parallel robot optimization method according to claim 4, characterized in that: The method of avoiding the first type of singularity, the second type of singularity and the visual servo controller singularity in the reachable workspace to obtain a smart workspace includes: Calculating a reachable workspace of the parallel robot; Through robot forward kinematics, a Rand function is used to generate a random value as a random variable within the motion range of each of the driving joints; Based on the random variables, the corresponding position and posture of the terminal are obtained by forward kinematics calculation; Obtaining a reachable working space range based on the posture corresponding to the terminal; Based on the reachable working space range, the image Jacobian matrix is ​​calculated to obtain the first kind singularity, the second kind singularity and the visual servo controller singularity; The first type of singularity, the second type of singularity and the visual servo controller singularity are avoided within the achievable workspace to obtain a dexterous workspace.

6. The parallel robot optimization method according to claim 5, characterized in that: The combined optimization design based on the visual servo positioning error model and the dexterous workspace to obtain the optimal visual sensor plus parallel robot combination scheme includes: Establish the objective function; Determining optimization design parameters based on the objective function; Establishing a global performance evaluation index based on the optimized design parameters, the visual servo positioning error model and the dexterous workspace; Based on the global performance evaluation index, discrete nodes are set in the reachable workspace of the parallel robot, and the average value of the discrete node function value is calculated to obtain the optimal visual sensor plus parallel robot combination solution.

7. The parallel robot optimization method according to claim 6, characterized in that: The optimized design parameters include visual sensor parameters and parallel robot size parameters.

Images