A six-axis robot control method based on global interior point iteration multi-start solving

Abstract

The application discloses a six-axis manipulator control method based on global interior point iteration multi-start solving, comprising: obtaining initial D-H parameters of the manipulator, performing coordinate transformation, and obtaining a theoretical posture matrix of the manipulator; obtaining an actual end posture matrix of the manipulator, and calculating a pose error between the theoretical posture matrix of the manipulator and the actual end posture matrix of the manipulator; adopting a global interior point iteration multi-start method, and obtaining new D-H parameters according to the initial D-H parameters and the actual end posture matrix of the manipulator, wherein the global interior point iteration multi-start method is to step by step optimize the to-be-optimized variables by adopting an interior point method and a multi-start strategy; and controlling the pose of the manipulator based on the obtained D-H parameters. The application combines the strong capability of the interior point method in local optimization and the advantage of the multi-start global search in global coverage, can efficiently and accurately find an optimal solution meeting a constraint condition, and is applied to the control of the six-axis manipulator, thereby providing a new effective way for improving the absolute positioning accuracy of the six-axis manipulator.

Description

Technical Field

[0001] This invention relates to the field of robotic arm control technology, and in particular to a six-axis robotic arm control method based on global interior point iterative multi-start point solution. Background Technology

[0002] In the rapid development of industrial automation, six-axis robots, as a key component of automated production, have a direct and significant impact on work quality, especially their absolute positioning accuracy. However, in practical applications, due to various factors such as machining errors, assembly deviations, component wear, and rod deformation, the actual kinematic parameters of six-axis robots differ from their nominal dimensions, making the problem of declining absolute positioning accuracy increasingly prominent. Although the repeatability of current industrial six-axis robots is generally high, improving absolute positioning accuracy remains a technical challenge. Among existing geometric parameter correction techniques, traditional methods such as the least squares method and the LM algorithm are prone to getting trapped in local optima when dealing with nonlinear optimization problems, making it difficult to comprehensively and accurately identify the geometric parameters of six-axis robots. Summary of the Invention

[0003] This invention provides a six-axis robot control method based on global interior point iteration with multiple starting points to overcome the above-mentioned technical problems.

[0004] To achieve the above objectives, the technical solution of the present invention is as follows: A six-axis robot control method based on global interior point iterative multi-start point solution includes the following steps: S1. Obtain the initial DH parameters of the robot arm, and perform coordinate transformation on the initial DH parameters to obtain the theoretical posture matrix of the robot arm; The initial DH parameters include link length, link twist angle, link offset, and joint angle; S2. Obtain the actual end-effector posture matrix of the robot, calculate the pose error between the theoretical pose matrix of the robot and the actual end-effector posture matrix of the robot, and determine whether the pose error is less than the preset control error threshold. If so, there is no need to optimize the initial DH parameters and execute S4; otherwise, execute S3. S3. Employ the global interior point iterative multi-starting point method, and optimize the new DH parameters based on the initial DH parameters and the actual end-effector posture matrix of the robot arm, and then execute S4; The global interior point iterative multi-starting point method uses the interior point method and the multi-starting point strategy to optimize the variables to be optimized step by step. S4. Control the pose of the robot arm based on the obtained DH parameters; S5. Verify the new pose of the controlled robot arm, i.e., repeat S2-S4 until the pose error between the actual end effector pose matrix and the theoretical pose matrix of the robot arm is less than the preset control error threshold for N consecutive times. .

[0005] Furthermore, in S1, the specific steps for performing coordinate transformation based on the initial DH parameters to obtain the theoretical posture matrix of the manipulator include: The transformation matrix from coordinate system {i-1} to coordinate system {i} is defined as follows: (1) In the formula, For joint angle, i.e., around the Z-axis of the connecting rod. i-1 Translation of the axis to make X i-1 Axis and X i Angles of collinearity of axes; This is the link offset, i.e., the offset around the link Z. i-1 Translation of the axis to make X i-1 Axis and X i Distance between collinear axes; The link length is the current link X. i Translation of the Z axis to make Z i-1 Axis and Z i The distance between intersecting or parallel axes; For the link twist angle; around the current link X i Rotate the axis to make Z i-1 Axis and Z i Angle of axis alignment; Substituting the initial DH parameters into formula (1), the theoretical attitude matrix of the manipulator, i.e., the forward kinematics equation of the manipulator, is constructed as follows: (2).

[0006] Furthermore, in S2, the actual end-effector posture matrix of the robot arm is represented as: (3) In the formula, , and This is the direction vector of the X-axis of the terminal coordinate system in the base coordinate system; , and This is the direction vector of the Y-axis of the end coordinate system in the base coordinate system; , and This is the direction vector of the Z-axis of the end coordinate system in the base coordinate system; , and The coordinates of the origin of the terminal coordinate system in the base coordinate system.

[0007] Furthermore, in S3, the specific steps for obtaining new DH parameters by using a global interior point iterative multi-starting point method and optimizing based on the initial DH parameters and the actual end-effector pose matrix of the manipulator include: The length and joint variables in the DH parameters are used as optimization variables. The actual end-effector posture matrix of the robot is combined with the interior point method and the multi-starting point strategy to search the solution space of the length variable in order to obtain the optimal length solution. The optimal length solution is substituted into the forward kinematics equations of the manipulator, and the interior point method and multi-start point strategy are used to search the solution space of the angle variable to obtain the optimal angle solution. Among them, the length variable solution space is composed of the link length and the link offset, and the angle variable solution space is composed of the link twist angle and the joint angle. New DH parameters are constructed based on the optimal length solution and the optimal angle solution.

[0008] Furthermore, combining the initial DH parameters and the actual end-effector posture matrix of the manipulator, and employing the interior point method and a multi-starting point strategy to search the solution space of the length variable, the specific steps to obtain the optimal length solution include: The objective function is to minimize the sum of squared pose errors at the end effector of the robotic arm. Set constraints, including: joint angle limits and link length range constraints; Set the algorithm parameters, including: interior point method parameters and multi-start point length search parameters; Multi-starting-point length search: Several sets of initial points satisfying the link length range constraint are randomly generated in the solution space of the length variable. The interior point method is used to independently perform a local search on each set of points to obtain the local optimal solution corresponding to each set of points: Global optimal solution selection: Compare the objective function values ​​of all local optima and select the solution corresponding to the minimum objective function value as the optimal length solution; In the process of performing a local search for each set of points using the interior point method, a barrier function is used to transform the constrained optimization problem into an unconstrained optimization problem, ensuring that the iteration points are always within the feasible region. The barrier function is expressed as: (7) In the formula, To optimize the variable vector, ; and The first Lower and upper bounds of each variable; For the barrier function parameters, ; The gradient of the objective function at the current iteration point is calculated using the central difference method. Based on this gradient, the search direction is iteratively updated using Newton's method. The update formula for Newton's method is as follows: (8) In the formula, To optimize the update step size of variables; The gradient vector, It is a Hessian matrix.

[0009] Furthermore, the specific steps for searching the solution space of the angle variable using the interior point method and the multi-starting point strategy to obtain the optimal angle solution include: The objective function is to minimize the sum of squared pose errors at the end effector of the robotic arm. Set constraints, including: joint angle limit constraints; Set the algorithm parameters, including: interior point method parameters and multi-start point angle search parameters; Multi-starting point angle search: Several sets of initial points satisfying the joint angle limit constraints are randomly generated in the solution space of the angle variable. The interior point method is used to independently perform a local search on each set of points to obtain the local optimal solution corresponding to each set of points: Global optimal solution selection: Compare the objective function values ​​of all local optima and select the solution corresponding to the minimum objective function value as the optimal angle solution.

[0010] Beneficial Effects: This invention obtains the initial DH parameters of the manipulator and performs coordinate transformation based on these parameters to obtain the theoretical posture matrix. Then, it obtains the actual end-effector posture matrix and, based on the error, uses a global interior point iterative multi-starting-point method to optimize and obtain new DH parameters based on the initial DH parameters and the actual end-effector posture matrix. Finally, it controls the manipulator based on the obtained DH parameters. This invention integrates the powerful local optimization capabilities of the interior point method with the global coverage advantages of multi-starting-point global search, enabling it to efficiently and accurately find the optimal solution that satisfies the constraints. Applied to the control of six-axis manipulators, it provides a new and effective way to improve the absolute positioning accuracy of six-axis manipulators. It has significant application value and broad prospects in key areas such as improving the accuracy of industrial six-axis manipulators, aerospace precision manufacturing, precise operation of medical six-axis manipulators, flexible production in electronic manufacturing, and scientific research and education. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 This is a flowchart of a six-axis robot control method based on global interior point iteration and multi-start point solution in this invention; Figure 2 This is a schematic diagram of the linkage coordinate system when the six-axis robot is in the zero position in an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the connection relationship between two adjacent links of the robot in an embodiment of the present invention. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0014] This embodiment provides a six-axis robot control method based on global interior point iteration with multiple starting points, such as... Figure 1 As shown, the specific steps include: S1. Obtain the initial DH parameters of the robot arm according to the robot arm instruction manual, and perform coordinate transformation on the initial DH parameters to obtain the theoretical posture matrix of the robot arm; The initial DH parameters include link length, link twist angle, link offset, and joint angle; S2. Obtain the actual end-effector posture matrix of the robot, calculate the pose error between the theoretical pose matrix of the robot and the actual end-effector posture matrix of the robot, and determine whether the pose error is less than the preset control error threshold. If so, there is no need to optimize the initial DH parameters and execute S4; otherwise, execute S3. Specifically, this embodiment uses multiple sensors for data acquisition, including: using a fusion scheme of laser displacement sensor (measurement range 35-65mm, repeatability 0.05mm) and inertial gyroscope (attitude resolution <0.01°) to acquire 24 sets of pose data at the end of the six-axis manipulator, and generating a measured pose matrix through coordinate transformation.

[0015] Specifically, in this embodiment, the pose error between the theoretical posture matrix and the actual end-effector posture matrix of the robot is defined as the Euclidean distance between the theoretical end-effector position vector and the actual end-effector position vector. When it is less than the preset control error threshold (e.g., 0.1 mm), it is determined that the DH parameter does not need to be optimized.

[0016] S3. A global interior point iterative multi-starting point solution method is adopted, and new DH parameters are obtained by optimization based on the initial DH parameters and the actual end-effector posture matrix of the manipulator, and then S4 is executed; wherein, the global interior point iterative multi-starting point method is to use the interior point method and the multi-starting point strategy to optimize the variables to be optimized step by step; S4. Control the pose of the robot arm based on the obtained DH parameters.

[0017] S5. Verify the new pose of the controlled robot arm, i.e., repeat S2-S4 until the pose error between the actual end effector pose matrix and the theoretical pose matrix of the robot arm is less than the preset control error threshold for N consecutive times. .

[0018] In a specific embodiment, in S1, as follows: Figure 2 (X0, Y0, Z0 are the base coordinate system, X1, Y1, Z1 are the first joint coordinate system, X2, Y2, Z2 are the second joint coordinate system, X3, Y3, Z3 are the third joint coordinate system, X4, Y4, Z4 are the fourth joint coordinate system, X5, Y5, Z5 are the fifth joint coordinate system, and X6, Y6, Z6 are the sixth joint coordinate system) and Figure 3 As shown, the specific steps for obtaining the theoretical posture matrix of the robot arm by performing coordinate transformation based on the initial DH parameters include: The transformation matrix from coordinate system {i-1} to coordinate system {i} is defined as follows: (1) In the formula, For joint angle, i.e., around the Z-axis of the connecting rod. i-1 Translation of the axis to make X i-1 Axis and X i Angles of collinearity of axes; This is the link offset, i.e., the offset around the link Z. i-1 Translation of the axis to make X i-1 Axis and X i Distance between collinear axes; The link length is the current link X. i Translation of the Z axis to make Z i-1 Axis and Z i The distance between intersecting or parallel axes; For the link twist angle; around the current link X i Rotate the axis to make Z i-1 Axis and Z iAngle of axis alignment; Substituting the initial DH parameters into formula (1), the theoretical attitude matrix of the manipulator, i.e., the forward kinematics equation of the manipulator, is constructed as follows: (2).

[0019] In a specific embodiment, in S2, the actual end-effector posture matrix of the robot arm is represented as: (3) In the formula, , and This is the direction vector of the X-axis of the terminal coordinate system in the base coordinate system; , and This is the direction vector of the Y-axis of the end coordinate system in the base coordinate system; , and This is the direction vector of the Z-axis of the end coordinate system in the base coordinate system; , and The coordinates of the origin of the final coordinate system in the base coordinate system; Since the TR10-1310C robot has a six-axis serial structure, the relationship between the forward kinematics equations of the manipulator and the base coordinate system is derived as follows: (4) Based on the above equation, the solution to the forward kinematics equation of the robot is derived as follows: (5) in, P Indicates the position of the robotic arm's end effector. R It is the attitude matrix of the robotic arm's end effector.

[0020] To describe the position of an object in space during robot motion, an XYZ fixed-angle transformation matrix is ​​introduced to represent the end effector pose of the manipulator, as follows: (6) in, α, β, γ These represent the Z, Y, and X rotation angles during attitude transformation.

[0021] In this embodiment, a complete closed-loop control scheme is formed by combining the step-by-step optimization strategy of length and angle variables. In S3, the specific steps of using the global interior point iterative multi-start point solution method and optimizing the new DH parameters based on the initial DH parameters and the actual end-effector posture matrix of the manipulator include: The length and joint variables in the DH parameters are used as optimization variables. The actual end-effector posture matrix of the robot is combined with the interior point method and the multi-starting point strategy to search the solution space of the length variable in order to obtain the optimal length solution. The optimal length solution is substituted into the forward kinematics equations of the manipulator, and the interior point method and multi-start point strategy are used to search the solution space of the angle variable to obtain the optimal angle solution. Among them, the length variable solution space is composed of the link length and the link offset, and the angle variable solution space is composed of the link twist angle and the joint angle. New DH parameters are constructed based on the optimal length solution and the optimal angle solution.

[0022] Specifically, in this embodiment, a six-axis manipulator forward kinematics model is established based on DH parameters to correlate the end-effector pose error with geometric parameters. In practice, except for joint angles, the other geometric parameters are known fixed values, but due to manufacturing and assembly errors, their true values ​​become "unknown." Therefore, the purpose of the method proposed in this embodiment is to make the pose calculated by the robot's theoretical model as close as possible to the actual measured pose. The key geometric parameters that have the greatest impact on the accuracy of the manipulator model and require the most identification and correction are used as optimization variables, including link lengths a3 and a4, and joint offsets d1, d2, d4, and d5. , , Collinear angle , , , There are a total of 13 variables; In a specific embodiment, the steps to obtain the optimal length solution by combining the initial DH parameters and the actual end-effector posture matrix of the manipulator, and by using the interior point method and a multi-starting point strategy to search the solution space of the length variable include: The optimization objective function is to minimize the sum of squared position and pose errors at the end of the robotic arm (e.g., position error ≤ 0.01 mm, attitude error ≤ 0.001 rad). Set constraints, including: joint angle limits and link length range constraints; Specifically, such as a3 [438.5, 444.5] mm, a4 [Nominal value ±3mm]; Set the algorithm parameters, including: interior point method parameters and multi-start point length search parameters; Multi-starting-point length search: Several sets of initial points satisfying the link length range constraint are randomly generated in the solution space of the length variable. The interior point method is used to independently perform a local search on each set of points to obtain the local optimal solution corresponding to each set of points: Specifically, in this embodiment, 20 sets of starting points are selected (length variable ±3mm, angle variable ±0.0175rad). Global optimal solution selection: Compare the objective function values ​​of all local optima and select the solution corresponding to the minimum objective function value as the optimal length solution; In the process of local search using the interior point method, a barrier function is used to transform the constrained optimization problem into an unconstrained optimization problem, ensuring that the iteration point is always within the feasible region. The barrier function is expressed as: (7) In the formula, To optimize the variable vector, ; and The first i Lower and upper bounds of each variable; For the barrier function parameters, In this embodiment, the following is set ; Using central difference (with a step size of 1.4901 × 10⁻⁶) -8 Calculate the gradient of the objective function at the current iteration point, and iteratively update the search direction based on the gradient using Newton's method to achieve sub-millimeter positioning accuracy. The update formula for Newton's method is: (8) In the formula, To optimize the variable update step size, it is set to 0.5; The gradient vector, It is a Hessian matrix.

[0023] In a specific embodiment, the steps for searching the solution space of the angle variable using the interior point method and the multi-starting point strategy to obtain the optimal angle solution include: The objective function is to minimize the sum of squared pose errors at the end effector of the robotic arm. Set constraints, including: joint angle limit constraints; Set the algorithm parameters, including: interior point method parameters and multi-start point angle search parameters; Multi-starting point angle search: Several sets of initial points satisfying the joint angle limit constraints are randomly generated in the solution space of the angle variable. The interior point method is used to independently perform a local search on each set of points to obtain the local optimal solution corresponding to each set of points: Specifically, this embodiment selects 20 sets of starting points (e.g. =441.5±3mm), ensuring that all points meet the link length range constraint.

[0024] Global optimal solution selection: Compare the objective function values ​​of all local optima and select the solution corresponding to the minimum objective function value as the optimal angle solution.

[0025] Specifically, this embodiment applies the interior point method to the optimization of robot parameters. By constructing an obstacle function, the constrained optimization problem is transformed into an unconstrained optimization problem. Within the feasible region, the search direction is iteratively updated using Newton's method to ensure that the iteration points strictly meet the constraints (such as joint angle limits and link length ranges), thus possessing a strong local depth exploration capability.

[0026] Specifically, in this embodiment, 20 sets of initial points are randomly generated in the solution space using a multi-starting point strategy. The interior point method is started independently for each set of initial points to perform local search, covering potential optimal solutions in different regions. When the preset upper limit of the number of iterations is reached or other termination conditions are met (such as the gradient being small enough), the iteration stops.

[0027] Table 1 DH Parameters of TR10-1310C

[0028] Specifically, in the process of solving for the length variable, the angle parameter is fixed, and the values ​​in Table 1 are... 3 4 1 2. Perform a global search on six length variables, including d4 and d5, with initial values ​​set to ±3mm (e.g., ...). 3. The initial value is 441.5 mm, and the upper and lower bounds are 438.5-444.5 mm. The length variable is solved by iterating through the interior point method until the objective function value converges. Specifically, in the process of solving for the angle variable, the result of solving for the length variable is substituted into the solution. , , , , , , Seven angular variables were optimized, with initial values ​​set to ±0.0175 rad (e.g., initial value of θ1 is 0 rad, upper and lower bounds are -0.0175 to -0.0175 rad). The optimal solution was selected after searching through 20 sets of starting points.

[0029] In this embodiment, the optimized DH parameters are input into the robot controller to drive the movement of each joint, and the robot's pose error is monitored in real time. If the error is found to be too large or the constraints are violated, the multi-starting-point global search and interior-point local search processes can be restarted to find a new optimal solution. This process optimizes length and angle variables in stages, combining global search and local depth exploration, to ensure that the six-axis robot achieves high-precision motion control under strict constraints.

[0030] Beneficial effects of this invention: 1. Significantly improved accuracy, breaking through the limitations of traditional algorithms. By integrating the interior-point method with a multi-starting-point global search strategy, the multi-starting-point search covers the global solution space, while the interior-point method ensures local depth optimization. This method effectively solves the problem of traditional algorithms (such as least squares and LM algorithms) easily getting trapped in local optima, improving the absolute positioning accuracy of the six-axis robot by 89.4%. Taking the Turing TR05-900C collaborative six-axis robot as an example, the maximum spatial positioning error was reduced from 3.8150mm to 0.4050mm. The constraint handling capability of the interior-point method ensures that parameter identification conforms to the mechanical structure limitations of the six-axis robot (such as joint angle limits of -175° to 175°), and the length variable identification error is reduced. 1.4×10⁻³ mm, angular variable error With a precision of 4.3×10-9°, it meets the requirements for sub-millimeter level accuracy in precision assembly, welding and other scenarios.

[0031] 2. Strong global search capability, covering multi-dimensional parameter optimization. This embodiment employs a multi-starting-point strategy based on the interior-point method (randomly generating 20 sets of initial points in the solution space, with each set of initial points independently initiating the interior-point method for local search, covering potential optimal solutions in different regions). The multi-starting-point strategy (randomly generating 20 sets of initial points) can cover different regions of the solution space, avoiding optimization deviations caused by initial value selection. For example, in solving for 13 geometric parameters of a six-axis robot (such as link lengths)... Joint offset When (etc.), through 20 independent iterations of starting points, the solution with the minimum objective function (sum of squared pose errors) is finally selected. Compared with the traditional single-starting-point optimization algorithm, the multi-starting-point strategy increases the probability of searching for the global optimal solution by 40%. According to the manufacturer's verification, the deviation between the parameter identification results and the manufacturer's return correction value is ≤5%, which is significantly better than the single-starting-point algorithm.

[0032] 3. Highly efficient constraint handling, adaptable to the mechanical characteristics of a six-axis robot. The interior point method embeds constraints such as joint angle limits (e.g., -175° to 175°) and link length ranges into the optimization process through obstacle functions, ensuring that the iteration point is always within the feasible region.

[0033] For example, when solving for the length variable, by setting... 3. Upper and lower boundaries (438.5~444.5mm) and The upper and lower bounds (-29.5~-23.5mm) prevent the parameter identification results from violating the mechanical structure constraints of the six-axis robot. Compared with the unconstrained optimization algorithm, the parameter effectiveness is improved by 100%.

[0034] 4. Strong data fusion and adaptability, compatible with multiple application scenarios. A multi-sensor fusion solution combining a laser displacement sensor (accuracy 0.05mm) and an inertial gyroscope (attitude resolution <0.01°) can be adapted to complex scenarios such as aerospace assembly and electronic manufacturing. For example, in aircraft structural assembly, through the acquisition and iterative optimization of 24 sets of pose data, precise calibration of the end effector posture of a six-axis robot in a confined space can be achieved, improving efficiency by 30% compared to traditional single-sensor solutions, with a single calibration time of ≤2 hours.

[0035] 5. High engineering practicality, reducing industrial application costs. The method proposed in this embodiment does not rely on expensive equipment such as high-precision laser trackers. By combining low-cost sensors and optimizing algorithms, it can improve the accuracy of industrial six-axis robots. Taking the hot melt adhesive application scenario in automotive assembly as an example, after applying this method, the absolute positioning error of the six-axis robot is reduced by 89.4%, reducing the glue path deviation problem caused by positioning deviation, reducing the product defect rate by 65%, and at the same time reducing the calibration cost by 40% compared with the traditional solution.

[0036] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that the technical solutions described in the foregoing embodiments can still be optimized, or some or all of the technical features can be equivalently replaced. These modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A six-axis robot control method based on global interior point iterative multi-starting-point solution, characterized in that, The specific steps include: S1. Obtain the initial DH parameters of the robot arm, and perform coordinate transformation on the initial DH parameters to obtain the theoretical posture matrix of the robot arm; The initial DH parameters include link length, link twist angle, link offset, and joint angle; S2. Obtain the actual end-effector posture matrix of the robot, calculate the pose error between the theoretical pose matrix of the robot and the actual end-effector posture matrix of the robot, and determine whether the pose error is less than the preset control error threshold. If so, there is no need to optimize the initial DH parameters and execute S4; otherwise, execute S3. S3. Employ the global interior point iterative multi-starting point method, and optimize the new DH parameters based on the initial DH parameters and the actual end-effector posture matrix of the robot arm, and then execute S4; In S3, a global interior point iterative multi-starting point method is adopted, and new DH parameters are obtained by optimization based on the initial DH parameters and the actual end-effector pose matrix of the manipulator, including: The length and joint variables in the DH parameters are used as optimization variables. The actual end-effector posture matrix of the robot is combined with the interior point method and the multi-starting point strategy to search the solution space of the length variable in order to obtain the optimal length solution. Among them, the length variable solution space is composed of the link length and the link offset, and the angle variable solution space is composed of the link twist angle and the joint angle. Construct new DH parameters based on the optimal length solution and the optimal angle solution; The optimal length solution is substituted into the forward kinematics equations of the robot, and the interior point method and multi-start point strategy are used to search the solution space of the angle variable to obtain the optimal angle solution, including: The objective function is to minimize the sum of squared pose errors at the end effector of the robotic arm. Set constraints, including: joint angle limits and link length range constraints; Set the algorithm parameters, including: interior point method parameters and multi-start point length search parameters; Multi-starting-point length search: Several sets of initial points satisfying the link length range constraint are randomly generated in the solution space of the length variable. The interior point method is used to independently perform a local search on each set of points to obtain the local optimal solution corresponding to each set of points: Global optimal solution selection: Compare the objective function values ​​of all local optima and select the solution corresponding to the minimum objective function value as the optimal length solution; In the process of performing a local search for each set of points using the interior point method, a barrier function is used to transform the constrained optimization problem into an unconstrained optimization problem, ensuring that the iteration points are always within the feasible region. The barrier function is expressed as: (7) In the formula, To optimize the variable vector, ; and The first Lower and upper bounds of each variable; For the barrier function parameters, ; The gradient of the objective function at the current iteration point is calculated using the central difference method. Based on this gradient, the search direction is iteratively updated using Newton's method. The update formula for Newton's method is as follows: (8) In the formula, To optimize the update step size of variables; The gradient vector, Let be the Hessian matrix; wherein, the global interior point iterative multi-starting point method adopts the interior point method and the multi-starting point strategy to optimize the variable to be optimized step by step; S4. Control the pose of the robot arm based on the obtained DH parameters; S5. Verify the new pose of the controlled robot arm, i.e., repeat S2-S4 until the pose error between the actual end effector pose matrix and the theoretical pose matrix of the robot arm is less than the preset control error threshold for N consecutive times. .

2. The six-axis robot control method based on global interior point iterative multi-start point solution according to claim 1, characterized in that, In S1, the specific steps for obtaining the theoretical posture matrix of the robot arm by performing coordinate transformation based on the initial DH parameters include: The transformation matrix from coordinate system {i-1} to coordinate system {i} is defined as follows: (1) In the formula, For joint angle, i.e., around the Z-axis of the connecting rod. i-1 Translation of the axis to make X i-1 Axis and X i Angles of collinearity of axes; This is the link offset, i.e., the offset around the link Z. i-1 Translation of the axis to make X i-1 Axis and X i Distance between collinear axes; The link length is the current link X. i Translation of the Z axis to make Z i-1 Axis and Z i The distance between intersecting or parallel axes; This is the link twist angle, i.e., the angle around the current link X. i Rotate the axis to make Z i-1 Axis and Z i Angle of axis alignment; Substituting the initial DH parameters into formula (1), the theoretical attitude matrix of the manipulator, i.e., the forward kinematics equation of the manipulator, is constructed as follows: (2)。 3. The six-axis robot control method based on global interior point iterative multi-start point solution according to claim 2, characterized in that, In S2, the actual end-effector posture matrix of the robot arm is represented as follows: (3) In the formula, , and This is the direction vector of the X-axis of the terminal coordinate system in the base coordinate system; , and This is the direction vector of the Y-axis of the end coordinate system in the base coordinate system; , and This is the direction vector of the Z-axis of the end coordinate system in the base coordinate system; , and The coordinates of the origin of the terminal coordinate system in the base coordinate system.

4. The six-axis robot control method based on global interior point iterative multi-start point solution according to claim 1, characterized in that, The specific steps for searching the solution space of the angle variable using the interior point method and the multi-starting point strategy to obtain the optimal angle solution include: The objective function is to minimize the sum of squared pose errors at the end effector of the robotic arm. Set constraints, including: joint angle limit constraints; Set the algorithm parameters, including: interior point method parameters and multi-start point angle search parameters; Multi-starting point angle search: Several sets of initial points satisfying the joint angle limit constraints are randomly generated in the solution space of the angle variable. The interior point method is used to independently perform a local search on each set of points to obtain the local optimal solution corresponding to each set of points: Global optimal solution selection: Compare the objective function values ​​of all local optima and select the solution corresponding to the minimum objective function value as the optimal angle solution.

Images