Cooperative mechanical arm performance test method and device based on sensitivity optimization and medium

Abstract

The invention relates to the technical field of industrial robot performance testing, and provides a cooperative mechanical arm performance testing method and device based on sensitivity optimization and a medium. The method comprises the steps that D-H parameter sets of all joints in a cooperative mechanical arm are obtained, and a cooperative mechanical arm model is constructed based on the D-H parameter sets of all the joints; according to the cooperative mechanical arm model, performance sensitive area recognition is conducted on the whole working space of the cooperative mechanical arm, and at least one high sensitive area is obtained; based on the recognized high-sensitivity area, a test track corresponding to the cooperative mechanical arm is planned and optimized, and a closed test track covering the high-sensitivity area is generated; and based on the cooperative mechanical arm model and the closed test track, performing a performance test on the cooperative mechanical arm to obtain a performance test result of the cooperative mechanical arm. According to the embodiment of the invention, the systematic evaluation of the performance of the cooperative mechanical arm is realized, and the performance characteristics and potential problems of the mechanical arm are more effectively revealed.

Description

Technical Field

[0001] This disclosure relates to the field of industrial robot performance testing technology, and more specifically, to a collaborative robotic arm performance testing method, apparatus, and medium based on sensitivity optimization. Background Technology

[0002] Collaborative robotic arms, as core equipment for next-generation human-machine collaborative operations, are increasingly widely used in scenarios with extremely high requirements for safety, flexibility, and dynamic precision, such as precision assembly, flexible production lines, and medical assistance. Their performance not only affects the accuracy and efficiency of task execution but also directly relates to the safety and reliable interaction during human-machine collaboration. Therefore, establishing a scientific and comprehensive testing method to evaluate the overall performance of collaborative robotic arms, especially their weaknesses in dynamic, near-human environments, has become a crucial prerequisite for promoting the in-depth implementation and iterative upgrades of this technology.

[0003] In related technologies, performance testing of collaborative robotic arms often draws on traditional industrial robot evaluation systems, especially the GB / T 12642 (equivalent to ISO 9283) standard. This standard provides a unified framework for performance evaluation by setting a regular "cube" test area in the workspace and specifying fixed test points (e.g., 5 points) and standardized trajectories (e.g., straight lines, circles). However, the regular cube in this method is difficult to match the complex shape of the robotic arm's entire workspace, resulting in the inability to effectively detect performance in key areas such as singular configurations and motion boundaries. Furthermore, the selection of test points and trajectories lacks theoretical basis, making it difficult for the evaluation results to fully reflect its true capabilities under complex dynamic working conditions. Summary of the Invention

[0004] This disclosure provides at least one method, apparatus, and medium for testing the performance of a collaborative robotic arm based on sensitivity optimization. By performing error sensitivity analysis across the entire workspace to identify weak performance areas and generating targeted test trajectories accordingly, it achieves a systematic evaluation of the collaborative robotic arm's performance and more effectively reveals the robotic arm's performance characteristics and potential problems.

[0005] This disclosure provides a sensitivity-optimized collaborative robotic arm performance testing method, including: Obtain the DH parameter set of each joint in the collaborative robotic arm, and construct a collaborative robotic arm model based on the DH parameter set of each joint. Based on the collaborative robotic arm model, the performance-sensitive regions of the entire workspace of the collaborative robotic arm are identified to obtain at least one highly sensitive region. Based on the identified highly sensitive area, the test trajectory corresponding to the cooperative robotic arm is planned and optimized to generate a closed test trajectory covering the highly sensitive area; Based on the collaborative robotic arm model and the closed test trajectory, a performance test was conducted on the collaborative robotic arm to obtain the performance test results of the collaborative robotic arm. The step of planning and optimizing the test trajectory corresponding to the collaborative robotic arm based on the identified highly sensitive region includes: If only one highly sensitive region is identified, then a closed test trajectory is planned within that highly sensitive region. If at least two highly sensitive regions are identified, then within each highly sensitive region, multiple sampling points located at the edge of the highly sensitive region are selected as lateral feature points according to a preset rule; and based on the lateral feature points of all highly sensitive regions, a closed test trajectory that can traverse all highly sensitive regions is constructed. Multiple trajectory test points for performance testing are determined on the closed test trajectory.

[0006] In some possible embodiments, the DH parameters include joint angles; the identification of performance-sensitive areas across the entire workspace of the collaborative robotic arm includes: Based on the DH parameter set of each joint, the joint angle range of each joint in the collaborative robotic arm is determined; and, random sampling is performed within the joint angle range of each joint to obtain multiple sets of joint angle combinations. For each set of joint angle combinations, the spatial point corresponding to the end of the collaborative robotic arm is determined using the collaborative robotic arm model, thus obtaining the full workspace point set of the collaborative robotic arm. Uniform sampling is performed on the entire workspace point set to obtain a candidate point set covering the entire workspace of the collaborative robotic arm; Calculate the error sensitivity index for each candidate point in the candidate point set to obtain the sensitivity value corresponding to each candidate point; Based on the sensitivity values ​​and spatial locations of each candidate point in the candidate point set, a clustering analysis method is used to divide the candidate point set into multiple clusters, and at least one highly sensitive region is determined according to the average sensitivity value of each cluster.

[0007] In some possible embodiments, calculating the error sensitivity index for each candidate point in the candidate point set includes: For each candidate point, calculate the Jacobian matrix corresponding to each joint angle in the collaborative robotic arm. The Jacobian matrix represents the relationship between the end-effector pose change and the joint angle change of the collaborative robotic arm. The error sensitivity index of the candidate point is determined based on the determinant value of the Jacobian matrix.

[0008] In some possible embodiments, constructing a closed test trajectory capable of traversing all highly sensitive regions includes: For each highly sensitive region, multiple lateral feature points selected by the preset rules are used to filter out optimized target lateral feature points from them using a genetic optimization algorithm. Connect the target side feature points from different highly sensitive areas with straight line segments; In each highly sensitive region, a cubic Bézier curve is constructed, with the target side feature point in the highly sensitive region as the endpoint and passing through the interior of the highly sensitive region; wherein, the control points of the cubic Bézier curve are selected through optimization, and the optimization objective is to maximize the average sensitivity index value of multiple sampling points evenly distributed on the cubic Bézier curve. The cubic Bézier curves corresponding to each highly sensitive region are smoothly spliced ​​with the connecting straight line segments to form a closed ∞-shaped test trajectory. Accordingly, determining multiple trajectory test points for performance testing on the closed test trajectory includes: Uniform sampling is performed on each segment of the cubic Bézier curve in the ∞-shaped test trajectory to obtain multiple curve test points; The multiple curve test points and the connection points between the connecting straight line segments and the cubic Bézier curve are collectively used as the trajectory test points.

[0009] In some possible embodiments, after generating a closed test trajectory covering the highly sensitive region, the process includes: Based on the collaborative robotic arm model, the torque requirements of each joint of the collaborative robotic arm are predicted when it moves along the closed test trajectory; and, based on the predicted torque requirements of each joint of the collaborative robotic arm when it moves, the end-effector posture of the collaborative robotic arm at each trajectory test point on the closed test trajectory is optimized and adjusted. The optimized and adjusted collaborative robotic arm posture sequence is combined with the closed test trajectory to generate the final test trajectory and posture commands for performance testing.

[0010] In some possible embodiments, the performance test results of the collaborative robotic arm include pose data and trajectory tracking data; obtaining the performance test results of the collaborative robotic arm includes: Based on the pose data at the trajectory test points, the pose accuracy, pose repeatability, and multi-directional pose accuracy variation of the robotic arm are calculated, and the pose accuracy index of the collaborative robotic arm is obtained based on the calculation results. Based on the trajectory tracking data, the trajectory accuracy and trajectory repeatability of the robotic arm are calculated, and the trajectory tracking accuracy index is obtained based on the calculation results. Based on the pose accuracy index and the trajectory tracking accuracy index, the overall performance of the collaborative robotic arm is evaluated, and a performance evaluation report is generated.

[0011] This disclosure provides a sensitivity-optimized collaborative robotic arm performance testing device, comprising: The model building module is used to obtain the DH parameter set of each joint in the collaborative robotic arm and build a collaborative robotic arm model based on the DH parameter set of each joint. The region identification module is used to identify performance-sensitive regions in the entire workspace of the collaborative robotic arm based on the collaborative robotic arm model, and obtain at least one highly sensitive region. The trajectory generation module is used to plan and optimize the test trajectory corresponding to the cooperative robotic arm based on the identified high-sensitivity area, and generate a closed test trajectory covering the high-sensitivity area. The performance testing module is used to conduct performance tests on the collaborative robotic arm based on the collaborative robotic arm model and the closed test trajectory, and to obtain the performance test results of the collaborative robotic arm. Specifically, the trajectory generation module is used for: If only one highly sensitive region is identified, then a closed test trajectory is planned within that highly sensitive region. If at least two highly sensitive regions are identified, then within each highly sensitive region, multiple sampling points located at the edge of the highly sensitive region are selected as lateral feature points according to a preset rule; and based on the lateral feature points of all highly sensitive regions, a closed test trajectory that can traverse all highly sensitive regions is constructed. Multiple trajectory test points for performance testing are determined on the closed test trajectory.

[0012] In some possible embodiments, the DH parameters include joint angles; the region identification module is specifically used for: Based on the DH parameter set of each joint, the joint angle range of each joint in the collaborative robotic arm is determined; and, random sampling is performed within the joint angle range of each joint to obtain multiple sets of joint angle combinations. For each set of joint angle combinations, the spatial point corresponding to the end of the collaborative robotic arm is determined using the collaborative robotic arm model, thus obtaining the full workspace point set of the collaborative robotic arm. Uniform sampling is performed on the entire workspace point set to obtain a candidate point set covering the entire workspace of the collaborative robotic arm; Calculate the error sensitivity index for each candidate point in the candidate point set to obtain the sensitivity value corresponding to each candidate point; Based on the sensitivity values ​​and spatial locations of each candidate point in the candidate point set, a clustering analysis method is used to divide the candidate point set into multiple clusters, and at least one highly sensitive region is determined according to the average sensitivity value of each cluster.

[0013] In some possible embodiments, the region identification module is specifically used for: For each candidate point, calculate the Jacobian matrix corresponding to each joint angle in the collaborative robotic arm. The Jacobian matrix represents the relationship between the end-effector pose change and the joint angle change of the collaborative robotic arm. The error sensitivity index of the candidate point is determined based on the determinant value of the Jacobian matrix.

[0014] In some possible embodiments, the trajectory generation module is specifically used for: For each highly sensitive region, multiple lateral feature points selected by the preset rules are used to filter out optimized target lateral feature points from them using a genetic optimization algorithm. Connect the target side feature points from different highly sensitive areas with straight line segments; In each highly sensitive region, a cubic Bézier curve is constructed, with the target side feature point in the highly sensitive region as the endpoint and passing through the interior of the highly sensitive region; wherein, the control points of the cubic Bézier curve are selected through optimization, and the optimization objective is to maximize the average sensitivity index value of multiple sampling points evenly distributed on the cubic Bézier curve. The cubic Bézier curves corresponding to each highly sensitive region are smoothly spliced ​​with the connecting straight line segments to form a closed ∞-shaped test trajectory. Accordingly, the trajectory generation module is specifically used for: Uniform sampling is performed on each segment of the cubic Bézier curve in the ∞-shaped test trajectory to obtain multiple curve test points; The multiple curve test points and the connection points between the connecting straight line segments and the cubic Bézier curve are collectively used as the trajectory test points.

[0015] In some possible embodiments, the trajectory generation module is further configured to: Based on the collaborative robotic arm model, the torque requirements of each joint of the collaborative robotic arm are predicted when it moves along the closed test trajectory; and, based on the predicted torque requirements of each joint of the collaborative robotic arm when it moves, the end-effector posture of the collaborative robotic arm at each trajectory test point on the closed test trajectory is optimized and adjusted. The optimized and adjusted collaborative robotic arm posture sequence is combined with the closed test trajectory to generate the final test trajectory and posture commands for performance testing.

[0016] In some possible embodiments, the performance test results of the collaborative robotic arm include pose data and trajectory tracking data; the performance test module is also used for: Based on the pose data at the trajectory test points, the pose accuracy, pose repeatability, and multi-directional pose accuracy variation of the robotic arm are calculated, and the pose accuracy index of the collaborative robotic arm is obtained based on the calculation results. Based on the trajectory tracking data, the trajectory accuracy and trajectory repeatability of the robotic arm are calculated, and the trajectory tracking accuracy index is obtained based on the calculation results. Based on the pose accuracy index and the trajectory tracking accuracy index, the overall performance of the collaborative robotic arm is evaluated, and a performance evaluation report is generated.

[0017] This disclosure provides a computer device including a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the computer device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, they perform a sensitivity-optimized collaborative robotic arm performance testing method as described in any of the above possible embodiments.

[0018] This disclosure provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements a sensitivity-optimized collaborative robotic arm performance testing method as described in any of the possible embodiments above.

[0019] The sensitivity-optimized collaborative robotic arm performance testing method, apparatus, and medium provided in this disclosure identify performance-weak areas by performing error-sensitivity analysis across the entire workspace and generating targeted test trajectories accordingly. This allows testing resources to be concentrated on performance-sensitive key areas, enabling a systematic evaluation of the collaborative robotic arm's performance and revealing its performance characteristics and potential defects more efficiently within a limited testing time.

[0020] To make the above-mentioned objects, features and advantages of this disclosure more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0021] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings referenced in the embodiments will be briefly described below. These drawings are incorporated in and constitute a part of this specification. They illustrate embodiments conforming to this disclosure and, together with the specification, serve to explain the technical solutions of this disclosure. It should be understood that the following drawings only show some embodiments of this disclosure and should not be considered as limiting the scope. Those skilled in the art can obtain other related drawings based on these drawings without creative effort.

[0022] Figure 1A flowchart is shown for a sensitivity-optimized collaborative robotic arm performance testing method provided in an embodiment of this disclosure; Figure 2 A flowchart of a performance-sensitive region identification method provided by an embodiment of this disclosure is shown; Figure 3 The flowchart shows a method for generating an ∞-shaped trajectory based on feature point optimization and Bézier curve splicing provided in an embodiment of this disclosure; Figure 4 A flowchart of a collaborative robotic arm end-effector posture optimization method provided in an embodiment of this disclosure is shown; Figure 5 A schematic diagram of the structure of a sensitivity-optimized collaborative robotic arm performance testing device provided in an embodiment of this disclosure is shown. Figure 6 A schematic diagram of the structure of a computer device provided in an embodiment of this disclosure is shown. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this disclosure, and not all of them. The components of the embodiments of this disclosure described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this disclosure provided in the accompanying drawings is not intended to limit the scope of the claimed disclosure, but merely represents selected embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without inventive effort are within the scope of protection of this disclosure.

[0024] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0025] In this document, the term "and / or" merely describes a relationship, indicating that three relationships can exist. For example, A and / or B can represent three cases: A alone, A and B simultaneously, and B alone. Furthermore, the term "at least one" in this document means any combination of at least two of any one or more elements. For example, including at least one of A, B, and C can mean including any one or more elements selected from the set consisting of A, B, and C.

[0026] Industrial robots, as core equipment for automation and intelligence in manufacturing, have profoundly changed modern production models and become a key force driving industrial upgrading due to their high precision, high load capacity, stability, and high efficiency. Their performance directly determines the efficiency, quality, and safety limits of production lines. Therefore, establishing a scientific and comprehensive performance testing system is crucial for robot R&D iteration and large-scale application. Currently, performance testing mainly follows the GB / T 12642 standard (equivalent to ISO 9283), which, based on a cubic test space, specifies the testing methods for pose and trajectory performance, providing a unified evaluation basis for the industry.

[0027] In recent years, with the increasing demand for human-robot collaboration, collaborative robotic arms, as an important extension of industrial robots, have been widely used in precision assembly, flexible production lines, and medical rehabilitation due to their lightweight, flexibility, and ability to share workspace with humans. Compared with traditional industrial robots, collaborative robotic arms place higher demands on motion accuracy, dynamic performance, and environmental adaptability. Especially in operations involving human-robot interaction, frequent starts and stops, and trajectory changes, a comprehensive and reliable evaluation of their performance is crucial.

[0028] Research has found that performance testing for collaborative robotic arms largely draws on evaluation systems used in traditional industrial robots, mostly adopting the GB / T 12642 (equivalent to ISO 9283) standard. This involves setting a regular "cube" test area in the workspace and specifying fixed test points (e.g., 5 points) and standardized trajectories (e.g., straight lines, circles). However, existing standard methods based on fixed cubes and limited test points have inherent limitations, such as limited test areas, difficulty in covering the entire workspace, and inability to accurately identify weak performance areas. Therefore, when used for collaborative robotic arms, these methods often fail to fully reflect their true performance level in complex and dynamic collaborative scenarios.

[0029] Based on the above research, this disclosure provides a sensitivity-optimized collaborative robotic arm performance testing method, device, and medium. Specifically, a dynamic model of the collaborative robotic arm is first established, and the key areas most sensitive to performance changes within its entire workspace are identified. Then, an optimized closed test trajectory that can centrally cover these highly sensitive areas is planned. Finally, the robotic arm is guided to conduct performance testing along this trajectory.

[0030] In this embodiment of the disclosure, by performing error-sensitive analysis across the entire workspace to identify areas with weak performance and generating targeted test trajectories accordingly, test resources can be concentrated on key areas that are sensitive to performance. This enables a systematic evaluation of the collaborative robotic arm's performance and allows for more efficient revelation of the collaborative robotic arm's performance characteristics and potential defects within a limited test time.

[0031] To facilitate understanding of this embodiment, the executing entity of the sensitivity-optimized collaborative robotic arm performance testing method provided in this disclosure will first be described in detail. The executing entity of the sensitivity-optimized collaborative robotic arm performance testing method provided in this disclosure is a computer device. This computer device can be a terminal device or a server. The terminal device can also be a mobile device, user terminal, terminal, handheld device, computing device, vehicle-mounted device, wearable device, etc. The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud storage, big data, and artificial intelligence platforms. Optionally, this method can also be applied to an implementation environment composed of computer devices and servers.

[0032] The sensitivity-optimized collaborative robotic arm performance testing method provided in this application embodiment will be described in detail below with reference to the accompanying drawings. See also Figure 1 The diagram shows a flowchart of a collaborative robotic arm performance testing method based on sensitivity optimization provided in this disclosure. The method includes the following steps S101 to S104: S101, obtain the DH parameter set of each joint in the collaborative robotic arm, and construct a collaborative robotic arm model based on the DH parameter set of each joint.

[0033] As is understandable, collaborative robotic arms are robots capable of direct interaction and collaborative work with users within a shared space. Their typical characteristics include lightweight design, joint torque sensing, collision detection, and compliant control. They are mostly used in scenarios with high requirements for safety and flexibility, such as precision assembly, material handling, laboratory automation, medical assistance, and flexible production lines. The mechanical structure of a collaborative robotic arm typically consists of multiple rotary or translational joints connected in series. Each joint is independently controlled by a driver, and adjacent joints are connected by links, collectively determining the position and orientation of the end effector in three-dimensional space. Here, the DH parameter set for each joint is a standard set of parameters used to describe the geometric relationship of this series linkage mechanism. Typically, each joint includes four core parameters: joint angle, link torsion angle, link length, and link offset. Among these parameters, the joint angle represents the relative rotation angle between two adjacent links about a common axis, and its value changes with joint movement; the link torsion angle represents the torsion angle between the axes of two adjacent links in space, and is usually a fixed value; the link length represents the length of the common perpendicular line from the previous joint axis to the adjacent joint axis along the direction of the previous joint axis; and the link offset represents the distance from the intersection of the previous axis and the common perpendicular line to the intersection of the adjacent axis and the common perpendicular line along the direction of the joint axis. These parameters can be obtained from the design drawings of the robotic arm or through actual measurements.

[0034] Here, after obtaining the DH parameter set for each joint in the collaborative robotic arm, a collaborative robotic arm model can be constructed based on this parameter set. This model includes the kinematic model and the dynamic model of the collaborative robotic arm. In subsequent steps, the kinematic model in the collaborative robotic arm model will be mainly used for calculating the end-effector pose and Jacobian matrix, while its dynamic model can be used for joint torque prediction and analysis. The main steps in constructing the collaborative robotic arm model include: first, constructing the kinematic model of the collaborative robotic arm based on the DH parameter set; then, constructing the dynamic model of the collaborative robotic arm based on the kinematic model and the dynamic parameters such as the mass and inertia of each link.

[0035] First, a forward kinematic model of the collaborative robotic arm can be constructed based on the standard homogeneous transformation matrix recursive formula. This model is a computational model that describes the mathematical relationship between the pose of the robotic arm's end effector and the angles of each joint. Its inputs are the angle values ​​of all joints, and its outputs are the three-dimensional position and three-dimensional orientation of the end effector in the base coordinate system (i.e., the absolute reference coordinate system fixed on the base of the robotic arm, which is the reference for the movement of all joints). For example, given a set of joint angles, the theoretical position that the robotic arm's end effector should reach can be accurately calculated using this model.

[0036] Specifically, when constructing the kinematic model of a collaborative robotic arm based on the DH parameter set of each joint, the following process can be included: By establishing a local coordinate system conforming to the DH convention (Denavit-Hartenberg Convention) for each joint, the homogeneous transformation matrix between adjacent coordinate systems can be obtained; then, based on the four DH parameters (joint angle, link torsion angle, link length, and link offset), the standard transformation matrix corresponding to each joint can be calculated. This matrix comprehensively expresses the rotation about the z-axis of the previous joint, the translation along the z-axis, the translation along the new x-axis, and the rotation about the new x-axis; further, through continuous matrix multiplication from the base coordinate system to the end effector coordinate system, the total transformation matrix describing the position and attitude of the end effector in the base coordinate system can be recursively obtained; finally, the position vector and rotation matrix (or equivalent attitude representation, such as Euler angles or quaternions) are extracted from this total transformation matrix, thus completing the construction of the forward kinematic model.

[0037] Furthermore, after obtaining the kinematic model describing the geometric configuration of the robotic arm, the dynamic parameters of each link can be determined based on the DH parameters and mechanical design data, including link mass, center of mass position, moment of inertia tensor, and joint friction characteristics. At this point, the Lagrange method can be used for dynamic modeling. Based on the principle of energy conservation, and using the kinematic model defined by the DH parameters (providing information such as the position and velocity Jacobian matrix of each link and end effector) and the aforementioned dynamic parameters, the kinetic and potential energy expressions of the robotic arm system are first constructed. Then, the Lagrange function is solved and substituted into the Lagrange equations to derive a complete dynamic model including components of inertia, Coriolis-centrifugal force, gravity, and friction, thus effectively avoiding the problem of accurate quantification of multi-body coupling in direct force analysis. Therefore, the dynamic model of the collaborative robotic arm can be expressed as: ; in, It is represented as a joint drive torque vector, reflecting the output torque of each joint; It is represented as the inertia matrix, which is related to the robotic arm configuration q and characterizes the inertial properties of the system; Represented as a joint position vector; Represented as a velocity vector; Represented as an acceleration vector; Represented as a gravity term, it indicates the joint load of the robotic arm caused by configurational changes in a gravitational field; It is expressed as joint friction torque, which includes nonlinear factors such as static friction and dynamic friction; It is represented as a matrix of Coriolis force and centrifugal force, describing the dynamic effect of the velocity coupling term.

[0038] Here, to optimize the end effector posture of the robotic arm at the test point and dynamically adjust the motion posture during trajectory tracking to achieve smooth joint torque and maximize output, the joint torque can be further... As a key dynamic feature, the joint torque requirements under different postures are then predicted based on the established dynamic model to ensure that each joint of the robotic arm maintains a high torque output level throughout the tracking process, thereby giving full play to the dynamic performance of the robotic arm and avoiding the problems of decreased motion accuracy and trajectory deviation caused by insufficient torque.

[0039] In some other embodiments, a dynamic model of the collaborative robotic arm is constructed based on the DH parameter set of each joint. Besides the aforementioned Lagrange method, the Newton-Euler method, the Kane equations, or numerical methods based on model identification can also be used; no specific limitations are made here. Regardless of the specific method used, the common premise is to utilize the robotic arm's geometry and motion relationships defined by the DH parameters. For example, the efficient recursive Newton-Euler method also relies on the transformation relationship between adjacent link coordinate systems derived from the DH parameters to recursively calculate the velocity, acceleration, and interaction force between the links. Similarly, the Kane equations, with generalized velocity as an independent variable, also require the partial velocities and angular velocities determined by the DH kinematic model for their construction. Furthermore, in practical engineering, the model structure determined by the DH parameters can be combined with the actual joint torque and motion data collected through the excitation trajectory. System identification techniques can then be used to identify and calibrate the model parameters (such as inertia and friction coefficient), thereby obtaining a dynamic model that more closely matches the physical reality. These methods each have their own focus and can be selected based on different needs such as model accuracy, computational efficiency, implementation complexity, and ease of model linearization and controller design.

[0040] S102, Based on the collaborative robotic arm model, identify performance-sensitive areas in the entire workspace of the collaborative robotic arm to obtain at least one highly sensitive area.

[0041] Understandably, in collaborative robotic arms, significant differences exist in the dynamic characteristics and motion accuracy under different configurations or spatial positions during actual operation. These differences mainly manifest in joint torque requirements, end-effector positioning accuracy, dynamic response, and sensitivity to model parameter errors. Therefore, to mitigate risks and optimize performance in trajectory planning and control, a systematic analysis of the entire workspace of the robotic arm can be conducted to identify areas with unstable dynamic performance and particularly sensitive to errors or disturbances—these are performance-sensitive areas. Here, highly sensitive areas specifically refer to the spatial range within the workspace where, due to factors such as the robotic arm being near singular configurations, joints approaching their limits, or significant dynamic coupling effects, the pose error of the end effector is easily amplified by joint angle errors, model parameter mismatches, or external disturbances, thus affecting task accuracy and stability.

[0042] Specifically, since the motion performance of a robotic arm is closely related to its configuration, and the configuration is uniquely determined by the combination of joint angles, in order to systematically identify sensitive areas within the entire reachable space, when implementing performance-sensitive area identification, reference should be made to... Figure 2 As shown, the steps S201~S205 may be included: S201, based on the DH parameter set of each joint, determine the joint angle range of each joint in the collaborative robotic arm; and, randomly sample within the joint angle range of each joint to obtain multiple sets of joint angle combinations.

[0043] Understandably, the DH parameter set of each joint defines the joint's motion type and range. The motion range of different joints collectively determines the physical space boundary reachable by the robotic arm's end effector, and the density distribution of reachable points within this boundary is not uniform. Therefore, in order to comprehensively and unbiasedly explore the entire workspace, a random sampling method can be used to generate a large number of joint angle combinations within the allowable angle range of each joint, resulting in multiple sets of joint angle combinations. Different joint angle combinations correspond to different potential positions of the robotic arm's end effector in space.

[0044] Here, random sampling can use methods such as Monte Carlo sampling or Latin hypercube sampling, without being specifically limited to any particular method. The goal is to ensure that the sampling points can effectively cover the entire joint space.

[0045] S202, for each set of joint angle combinations, the spatial point corresponding to the end of the collaborative robotic arm is determined using the collaborative robotic arm model, thereby obtaining the full workspace point set of the collaborative robotic arm.

[0046] Specifically, after obtaining multiple sets of joint angle combinations, for each combination, the corresponding end-effector pose transformation matrix can be calculated using the kinematic model of the collaborative manipulator in its corresponding collaborative manipulator model. Then, the three-dimensional position coordinates of the end effector in the base coordinate system can be extracted from the matrix. Finally, based on the set of position coordinates calculated from all sampled combinations, the theoretically reachable point cloud of the manipulator can be determined. Here, the entire workspace represents the set of all possible reachable position points of the manipulator end effector, while the entire workspace point set refers to the set of discrete position points obtained through the above sampling and calculation process, used to approximately represent the entire workspace.

[0047] S203, uniformly sample the entire workspace point set to obtain a candidate point set covering the entire workspace of the collaborative robotic arm.

[0048] Here, since the initial random sampling of the entire working space point set may have uneven density or too many points, in order to improve the efficiency of subsequent calculations while ensuring coverage representativeness, the point set can be uniformly resampled to obtain a more spatially uniform and fewer candidate point set for subsequent sensitivity analysis.

[0049] Uniform sampling can employ Fibonacci hierarchical sampling strategies that balance uniformity and coverage, or spatial gridding sampling, farthest point sampling, etc., without specific limitations here.

[0050] For example, taking the Fibonacci hierarchical sampling strategy, the entire workspace of the collaborative robotic arm is uniformly divided using spherical coordinate parameterization, thereby generating a uniformly distributed set of candidate points. The division formula can be expressed as follows: ; ; in, It is represented as the azimuth (longitude) of the nth sampling point, used to determine the projection direction of the point on the horizontal plane; This is represented as the index number of the sampling point, typically taking values ​​of 0, 1, 2, ..., N-1, where N is the preset total number of sampling points; It is expressed as the golden angle constant, which is usually taken as about 137.508° (or about 2.39996 radians). The introduction of this constant can make the azimuth angle interval present a uniform distribution characteristic, ensuring that the sampling points are uniformly covered on the sphere. The zenith angle (co-latitude) of the nth sampling point is the angle between that point and the positive z-axis. Its value is constructed using an inverse cosine function to ensure a uniform distribution density of points on the sphere. By using a fixed radius or combining radial layering mapping, the above spherical sampling points can be extended to the entire three-dimensional workspace of the robotic arm, thereby obtaining a uniformly distributed and comprehensive set of candidate points.

[0051] S204, calculate the error sensitivity index of each candidate point in the candidate point set to obtain the sensitivity value corresponding to each candidate point.

[0052] Specifically, the candidate point set consists of a series of coordinate points representing the spatial position of the end effector. By calculating the error sensitivity index of each candidate point in the candidate point set, the influence of joint angle errors or model parameter perturbations on the end effector pose accuracy when the robotic arm's end effector is located at that point can be quantitatively evaluated, thus obtaining the sensitivity value corresponding to each candidate point. Here, the error sensitivity index is a numerical value used to quantify the performance robustness at that point; a higher value indicates that the point is more sensitive to errors.

[0053] For example, when calculating the error sensitivity index, for each candidate point and its corresponding joint angle, the velocity Jacobian matrix of the collaborative robot arm in that configuration can be calculated. The Jacobian matrix describes the linear mapping relationship between the linear and angular velocities of the robot arm's end effector and the velocities of each joint. Its determinant value directly reflects whether the configuration is close to a singular state. For example, when the determinant value of the Jacobian matrix is ​​close to zero, it means that the robot arm is near a singular configuration. Small changes in joint angles may lead to abnormal changes in the end effector pose or loss of motion capability in a certain direction, i.e., it is in a highly sensitive state.

[0054] In some possible embodiments, when calculating the error sensitivity index of candidate points, the index can also be directly determined based on the reciprocal (or the reciprocal of its absolute value) of the determinant of the Jacobian matrix, defined using the following formulaic error mapping relationship. According to D-optimal design theory, to evaluate the sensitivity of different test configurations to parameter errors, the linear mapping between the robot arm end-effector pose error vector and the joint angle error vector can be defined as the error sensitivity relationship at that point. This relationship can be directly established using the velocity Jacobian matrix, and its expression can be given as: ; in, This is represented as the pose error vector of the robotic arm's end effector in Cartesian space, generally including linear displacement error and angular displacement error, and can be specifically expressed as... ,in, , , These represent the linear displacement errors of the end along the x, y, and z coordinate axes, respectively. These are respectively expressed as angular displacement errors about the corresponding coordinate axes; The velocity Jacobian matrix of the robotic arm in the current configuration is represented as a... The matrix maps joint velocities to end-effector linear and angular velocities, where m is the end-effector degree of freedom (usually 6) and n is the number of robot joints. Represented as a joint angle error vector, that is, the deviation between the actual angle and the theoretical angle of each joint, it can be specifically expressed as: ,in, This is represented as the angular error of the i-th joint. Specifically, in the simplified model that only analyzes positional errors, It can be simplified to At this time, the Jacobian matrix Corresponding to the position Jacobian matrix .

[0055] Furthermore, based on the above mapping relationship, the error sensitivity index under the current configuration can be defined as the reciprocal of the determinant of the Jacobian matrix, that is: ; This metric reflects the degree to which joint space errors are amplified to Cartesian space, when... When the value is small, it indicates that the robotic arm is close to a singular configuration. Small joint errors can lead to large pose deviations at the end effector. At this time, the sensitivity S is high, which corresponds to the high-sensitivity region.

[0056] In this way, by calculating the sensitivity of each candidate point to small changes in joint parameters, we can assess the point's ability to expose potential performance defects in the robot, ensuring that the final set of test points can efficiently reveal the robot's potential performance shortcomings.

[0057] In some other embodiments, the error sensitivity index of candidate points can also be determined based on the condition number or the reciprocal of the minimum singular value of the Jacobian matrix. The larger the condition number or the smaller the minimum singular value, the higher the sensitivity index. No specific limitation is made here.

[0058] S205, based on the sensitivity value and spatial location of each candidate point in the candidate point set, a clustering analysis method is used to divide the candidate point set into multiple clusters, and at least one high-sensitivity region is determined according to the average sensitivity value of each cluster.

[0059] Specifically, after obtaining the sensitivity value of each candidate point, it can be combined with spatial coordinates to form a feature vector. Based on the spatial and sensitivity distribution characteristics of these feature vectors, a clustering analysis algorithm is used to divide the candidate point set into several spatially adjacent clusters with similar sensitivity characteristics. Then, the average sensitivity of all points within each cluster is calculated, and the spatial regions corresponding to clusters with average sensitivity values ​​significantly higher than a preset threshold or significantly higher than other clusters are identified as high-sensitivity regions. For example, the top N clusters with the highest average sensitivity values ​​can be selected, or clusters with average sensitivity values ​​exceeding a certain proportion of the global mean can be identified as high-sensitivity regions. Here, the average sensitivity value of each cluster refers to the arithmetic mean or weighted average of the sensitivity values ​​of all candidate points within that cluster, which can be calculated using conventional statistical methods.

[0060] For example, when performing cluster analysis on a candidate point set, the K-means++ clustering analysis method, which comprehensively considers the spatial coordinates of the sampling points and the sensitivity index value, can be introduced. This method improves upon the standard K-means algorithm, and its process mainly consists of two steps: First, a distance-based probability distribution strategy is used to select higher-quality initial cluster centers to reduce the algorithm's sensitivity to the selection of initial centers and improve convergence performance; second, a standard K-means iterative optimization process is executed, dividing the candidate points into different clusters by alternating "assignment" and "update" operations. The optimization objective of clustering is to minimize the sum of squared distances from all sample points to the center of their respective clusters, and its objective function can be expressed as: ; in, It is represented as the sum of squared total clustering errors, which is the sum of the squared Euclidean distances from all sample points to the center of their respective clusters. The smaller this value is, the higher the similarity of samples within the cluster and the more compact the clustering effect. K represents the preset number of clusters, which is the number of clusters that the candidate point set is expected to be divided into. The feature vector of the i-th candidate point is typically represented by the spatial coordinates (x, y, z) of that point and its sensitivity value S. i Together constitute, that is ; It is represented as the center (centroid) of the k-th cluster, which is calculated during the iteration process by the arithmetic mean of the coordinates and sensitivity values ​​of all sample points belonging to this cluster; It is represented as the k-th cluster, which is the set of all sample points assigned to this cluster.

[0061] Here, the algorithm aims to minimize the sum of squares within clusters. It also repeatedly runs the algorithm to select the most stable and reliable clustering scheme, ensuring that the mined data patterns represent the true inherent characteristics of spatial performance distribution, rather than accidental results caused by random disturbances. In practice, the spatial coordinates of each sampling point and its corresponding sensitivity index value can be fused into a feature vector and input into the clustering algorithm. After determining the optimal number of clusters through stability analysis, continuous regions with similar performance characteristics are grouped into the same cluster. Subsequent optimization focuses on the "high-sensitivity areas" that play a decisive role in performance evaluation, while actively eliminating "low-sensitivity areas" that show weak responses to performance changes. Ultimately, this concentrates testing resources on the key areas that best distinguish the robot's performance.

[0062] In some other embodiments, other clustering methods such as DBSCAN (density-based noise applied spatial clustering) and hierarchical clustering may also be used, without being specifically limited here.

[0063] S103, based on the identified highly sensitive area, the test trajectory corresponding to the cooperative robotic arm is planned and optimized to generate a closed test trajectory covering the highly sensitive area.

[0064] Understandably, after identifying highly sensitive areas, a dedicated test trajectory that can systematically cover these areas needs to be planned to effectively evaluate the dynamic performance and robustness of the collaborative robotic arm in these critical regions. This trajectory should be closed, continuous, smooth, and repeatable to generate a standardized dynamic performance test path for the collaborative robotic arm to track and execute during testing, thereby enabling systematic and comparable performance acquisition and evaluation.

[0065] Here, if only one highly sensitive region is identified, a closed test trajectory can be directly planned within that region. Specifically, a workspace grid-based path search method can be used (e.g.,...). An algorithm or a local trajectory planning method based on artificial potential fields is used to generate a closed loop that traverses typical locations within the region. Then, using equal arc length sampling or equal time interval sampling methods on this closed test trajectory, multiple trajectory test points for performance testing are determined. Based on these test points and their connection order, a closed test trajectory covering the single high-sensitivity region is generated. If at least two high-sensitivity regions are identified, it indicates that the robotic arm has multiple performance weaknesses within the entire workspace. To comprehensively evaluate its overall performance, a test trajectory that connects all high-sensitivity regions needs to be planned, thus examining multiple key regions simultaneously in a single test.

[0066] Specifically, when at least two highly sensitive regions are identified, multiple sampling points located at the edge of the highly sensitive region can be selected as lateral feature points within each highly sensitive region according to preset rules. For example, the preset rules can be set to extract 10% of the sampling points at the edge of each cluster or to extract contour points based on the convex hull boundary or principal component analysis of the region. Based on the lateral feature points of all highly sensitive regions, a closed test trajectory that can traverse all highly sensitive regions is constructed. Then, on the closed test trajectory, isoparametric sampling or adaptive curvature sampling methods are used to determine multiple trajectory test points for performance testing.

[0067] For example, since highly sensitive regions may be scattered and irregularly shaped in space, simply connecting the center or boundary points of the regions directly may result in a trajectory that is not smooth enough, fails to fully cover the highly sensitive locations within the region, and may introduce unnecessary sharp turns, which is detrimental to the smooth tracking and performance evaluation of the robotic arm. Therefore, this disclosure proposes an ∞-shaped trajectory generation method based on feature point optimization and Bézier curve splicing. When constructing a closed test trajectory covering multiple regions, it refers to... Figure 3 As shown, the following steps S301~S304 may be included: S301, for each highly sensitive region, multiple lateral feature points selected by the preset rules are used to select optimized target lateral feature points from them using a genetic optimization algorithm.

[0068] Understandably, given the large number of lateral feature points and their somewhat random distribution, directly using all points to construct a trajectory would lead to redundancy, unevenness, and significant optimization challenges. Therefore, a genetic optimization algorithm can be used to filter feature points within each region, selecting an optimal target lateral feature point that represents the spatial extent of the region and facilitates the subsequent construction of a smooth, highly sensitive coverage trajectory. Here, the optimization objective of the genetic optimization algorithm is to select from multiple lateral feature points in each region the point from which a cubic Bézier curve, using that point as its start and end point and constructed within its region, can maximally cover the points within the highly sensitive region.

[0069] Specifically, the fitness function is designed to evaluate whether, given a candidate point as the endpoint of a Bézier curve, the average sensitivity index value of the sampling points along the path of the Bézier curve generated within that region according to geometric constraints (such as maintaining second-order continuity with the connecting lines between regions) is as high as possible. This ensures that the curve segment can effectively penetrate and cover the most sensitive and core parts of the region that need to be tested. Through iterative optimization using a genetic algorithm (including selection, crossover, and mutation operations), the optimal target lateral feature point can be determined for each highly sensitive region. This point will directly serve as the start and end points for constructing the cubic Bézier curve that traverses the region in subsequent steps.

[0070] S302 connects the target side feature points from different highly sensitive areas using straight line segments.

[0071] Furthermore, after obtaining the target side feature points of each highly sensitive area, these feature points can be connected sequentially using straight line segments according to their spatial relationship (such as the order of connecting the center of the region or the minimum spanning tree principle) to form a basic path framework connecting the regions.

[0072] S303, in each highly sensitive region, a cubic Bézier curve is constructed with the target side feature point in the highly sensitive region as the endpoint and passing through the interior of the highly sensitive region; wherein, the control points of the cubic Bézier curve are selected by optimization, and the optimization objective is to maximize the average sensitivity index value of multiple sampling points evenly distributed on the cubic Bézier curve.

[0073] Specifically, when constructing a local trajectory segment to cover a single highly sensitive region, a cubic Bézier curve can be used as the curve model. Here, a cubic Bézier curve is a parametric smooth curve defined by two endpoints and two control points, possessing advantages such as simple expression, continuous differentiability, and ease of shape adjustment and smoothness control. During construction, the two endpoints are the target lateral feature points of the region, and the two internal control points can be determined through an optimization search within the region. The optimization objective is to maximize the average sensitivity index value of several points uniformly sampled on the Bézier curve, thereby ensuring that the curve segment passes through the most sensitive part of the region as much as possible. Here, several candidate control points are interpolated on the extension line connecting the lateral points. The following formula is used to select the combination of control points that maximizes the average sensitivity index value of a specified number of uniformly sampled points on the generated curve as the optimal solution: ; in, Let be the coordinate vector of a point on the cubic Bézier curve; t represents the parameter variable of the Bézier curve. The starting point of the curve at t=0 The endpoint of the curve at t=1 ; , These are represented as the start and end points of the cubic Bézier curve, respectively. In this specific embodiment, they correspond to the target side feature points selected by the genetic optimization algorithm within the highly sensitive region, that is, the curve enters from one side of the region and exits from the other side. , These are two internal control points of a cubic Bézier curve. They are not located on the curve, but they influence the shape and direction of the curve through weights. They are determined by searching within the region using an optimization algorithm. The goal of selecting their positions is to ensure that the entire curve passes through as many highly sensitive locations as possible when traversing the corresponding highly sensitive region, that is, to maximize the average sensitivity of the uniformly sampled points on the curve.

[0074] For example, the specific steps in the control point optimization process may include: using the position coordinates of the control points as decision variables, using the average sensitivity value of the sampling points on the curve as the fitness function, and using numerical optimization methods such as gradient descent and particle swarm optimization to iteratively solve the problem until the curve shape converges to the position that maximizes the average sensitivity.

[0075] S304, smoothly splice the cubic Bézier curves corresponding to each highly sensitive region with the connecting straight line segments to form a closed ∞-shaped test trajectory.

[0076] Furthermore, after optimizing the Bézier curve segments corresponding to each region, at the connection points, by ensuring that the tangent vector at the curve endpoints is consistent with the direction of the straight line or by applying a slight transition fillet, a smooth G1 continuity (tangent continuity) or higher-order connection between the Bézier curve segments and the connecting straight line segments can be achieved, ultimately forming a closed test trajectory resembling the symbol "∞". This ∞-shaped test trajectory ensures continuous and smooth movement of the robotic arm during tracking, and can systematically traverse the core parts of all highly sensitive areas, enabling concentrated and efficient evaluation of the robotic arm's performance weaknesses in multiple areas.

[0077] In some possible embodiments, the ∞-shaped test trajectory is composed of curve segments and straight line segments with different geometric characteristics. Its curvature and sensitivity distribution are non-uniform, which may lead to insufficient sampling or uneven distribution of data points when directly performing isochronous or isochronous motion control, affecting the representativeness and accuracy of subsequent performance index calculations. Therefore, after obtaining the closed test trajectory, it can be discretized and test points extracted, including the following steps (1)~(2): (1) Uniform sampling is performed on each segment of the cubic Bézier curve in the ∞-shaped test trajectory to obtain multiple curve test points; (2) The multiple curve test points and the connection points between the connecting straight line segments and the cubic Bézier curve are used together as the trajectory test points.

[0078] Here, uniform sampling on each segment of the cubic Bézier curve in the ∞-shaped test trajectory is performed to discretize the continuous smooth curve into a series of representative spatial points. This ensures that the shape and sensitivity distribution of each curve segment can be characterized by a point set of a certain density, guaranteeing that performance data acquisition covers every detail of the trajectory, especially high-curvature or high-sensitivity segments, and avoiding the omission of key dynamic characteristics due to sparse sampling. Simultaneously, including curve test points and connection points in the trajectory test point set ensures that key geometric transitions from one trajectory segment to another are also included in the evaluation, thus comprehensively reflecting the performance of the robotic arm in all typical stages (including straight-line cruising, curve tracking, and mode switching) along the entire closed trajectory.

[0079] Furthermore, after determining the test points for each trajectory segment, online or offline performance data acquisition and processing can be performed based on the sequence of test points to obtain the actual pose data, trajectory tracking error data, and joint state data at each test point during the execution of the closed test trajectory by the collaborative robotic arm.

[0080] Understandably, when a collaborative robotic arm moves along a predetermined spatial trajectory, the orientation of its end effector in three-dimensional space—the end effector attitude—significantly affects the load distribution of each joint actuator. A fixed end effector attitude may place the robotic arm in a dynamically disadvantageous configuration in certain trajectory segments, for example, causing the torque required by one joint to approach its output limit, while other joints fail to fully utilize their efficiency. Therefore, after generating the spatial closed test trajectory, a dynamics-based end effector attitude collaborative optimization step can be introduced to ensure a more reasonable joint load distribution and a more complete and safe demonstration of dynamic performance throughout the tracking test. (Refer to...) Figure 4 As shown, the steps S401~S402 may be included: S401, based on the collaborative robotic arm model, predict the torque requirements of each joint when the collaborative robotic arm moves along the closed test trajectory; and, based on the predicted torque requirements of each joint when the collaborative robotic arm moves, optimize and adjust the end-effector posture of the collaborative robotic arm at each trajectory test point on the closed test trajectory.

[0081] Here, joint torque requirement refers to the torque value required by each joint actuator to drive the robotic arm links, enabling them to overcome factors such as inertia, gravity, Coriolis force, centrifugal force, and joint friction, and to accurately track the predetermined trajectory. This can be predicted using the previously constructed collaborative robotic arm dynamics model, which includes parameters such as mass, moment of inertia, and center of mass position, combined with the spatial position, velocity, and acceleration information of each point on the closed test trajectory, through forward dynamics calculations. This prediction process quantifies the load on each joint when executing the test trajectory under the initial or default end-effector posture. Furthermore, based on issues revealed by the prediction, such as uneven load distribution, excessively high peak values, or severe fluctuations, the spatial rotation state of the robotic arm end-effector at each trajectory test point is iteratively adjusted and optimized in a targeted manner.

[0082] The steps for optimizing the end effector attitude can include: First, defining the end effector attitude parameters at each trajectory test point, such as yaw, pitch, and roll angles expressed in ZYX Euler angles, as a set of decision variables to be optimized; further, constructing one or more objective functions based on the predicted joint torque data, such as minimizing the maximum value of the torque demand of each joint to prevent single joint overload saturation, minimizing the sum of squares of the torque change rates of all joints to improve motion smoothness, or maximizing the average value of the torque output margin of each joint (i.e., the difference between the rated torque and the demand torque) to incentivize overall performance; then, using numerical optimization methods such as sequential quadratic programming or genetic algorithms, iteratively solving the above decision variables under the geometric and kinematic constraints of satisfying the range of motion of the robotic arm joints, avoiding kinematic singular configurations, and maintaining continuous attitude changes, to obtain a new attitude sequence. Finally, substituting the new attitude sequence into the dynamic model to recalculate the joint torque, verifying whether it meets the preset optimization objectives and safety thresholds, to determine the optimized attitude command sequence that matches the highly sensitive spatial trajectory for the final performance test.

[0083] S402, combine the optimized and adjusted collaborative robotic arm posture sequence with the closed test trajectory to generate the final test trajectory and posture commands for performance testing.

[0084] Specifically, after obtaining the optimized collaborative robotic arm posture sequence, this posture sequence can be combined with the coordinate points of the original spatial closed test trajectory in chronological order to generate digital motion commands containing complete spatial position information and optimized posture information, i.e., the final test trajectory and posture commands. These commands are used to directly drive the physical collaborative robotic arm control system to perform high-fidelity performance tests. Here, the final command specifies the position and orientation of the robotic arm's end effector at each moment, where the position component is determined by the closed test trajectory, and the posture component is defined by the optimized collaborative robotic arm posture sequence. Based on this command, the control system can coordinate the movement of each joint, ensuring that the robotic arm strictly follows the predetermined spatial path while maintaining the optimized end effector orientation. This ensures that during the dynamic execution of the performance test, the torque load of each joint remains within the optimized, more reasonable, and efficient working range, allowing for the acquisition of test data that truly reflects its dynamic performance.

[0085] S104. Based on the collaborative robotic arm model and the closed test trajectory, a performance test is conducted on the collaborative robotic arm to obtain the performance test results of the collaborative robotic arm.

[0086] Furthermore, after obtaining the closed test trajectory and its corresponding test points, a collaborative robotic arm model can be used to perform torque demand simulation analysis on the closed test trajectory. This determines the theoretical torque range and peak value of each joint when the robotic arm moves along the trajectory, in order to assess the feasibility of the experiment and set safety thresholds. Simultaneously, a real collaborative robotic arm can be guided to actually execute the closed test trajectory, and motion data can be synchronously collected through its built-in joint encoders, torque sensors, and external measuring devices (such as laser trackers or vision systems). Here, the performance test results of the collaborative robotic arm can include two types of data: pose data, i.e., the actual position and posture of the robotic arm end effector at each test point on the trajectory, and trajectory tracking data, i.e., the real-time position, velocity, acceleration, and deviation from the command value of the robotic arm end effector during the tracking of the entire trajectory.

[0087] Specifically, after obtaining the performance test results, based on the pose data at the trajectory test points and in accordance with standards such as GB / T12642, the pose accuracy (deviation between the commanded pose and the average actual pose), pose repeatability (dispersion of the actual pose around its average value), and multi-directional pose accuracy variation (change in accuracy when approaching the same test point in different directions) of the robotic arm can be calculated. Based on these calculation results, the pose accuracy index of the collaborative robotic arm can be obtained. Simultaneously, based on trajectory tracking data, the trajectory accuracy (spatial deviation between the actual trajectory and the commanded trajectory) and trajectory repeatability (consistency when executing the same trajectory multiple times) of the robotic arm can be calculated, thereby obtaining the trajectory tracking accuracy index. Finally, by combining the pose accuracy index and the trajectory tracking accuracy index, a comprehensive evaluation of the collaborative robotic arm's positioning accuracy, motion stability, and dynamic tracking capabilities in key sensitive areas can be conducted, and a structured performance evaluation report can be automatically generated, indicating performance advantages and potential weaknesses.

[0088] In this way, by combining model-driven sensitive area identification, trajectory optimization and standardized performance index calculation, targeted, efficient and quantitative evaluation of the performance of collaborative robotic arms can be achieved, thereby realizing a paradigm upgrade from universal testing of "fixed areas and fixed points" to precise testing of "weakness guidance and dynamic coverage".

[0089] In some possible embodiments, the sensitivity-optimized performance testing method proposed in this disclosure can also be used to conduct adaptive tests on the collaborative robotic arm under dynamic load conditions or comparative tests under different control parameters. By changing the end load or adjusting the controller gain and repeatedly executing the optimized test trajectory, the final optimal control scheme or load capacity boundary can be determined, thereby providing direct data support for the deployment and application of the robotic arm.

[0090] It should be noted that the sensitivity-optimized collaborative robotic arm performance testing method provided in this disclosure is not only applicable to collaborative robotic arms, but also to multi-axis industrial robots, SCARA (selective compliant assembly) robots, Delta parallel robots, and other industrial robots or precision motion platforms that have serial or parallel joint structures and require comprehensive performance evaluation within the workspace. No specific limitations are made here.

[0091] The sensitivity-optimized collaborative robotic arm performance testing method, apparatus, and medium provided in this disclosure identify performance-weak areas by performing error-sensitivity analysis across the entire workspace and generating targeted test trajectories accordingly. This allows testing resources to be concentrated on performance-sensitive key areas, enabling a systematic evaluation of the collaborative robotic arm's performance and revealing its performance characteristics and potential defects more efficiently within a limited testing time.

[0092] Those skilled in the art will understand that, in the above-described method of the specific implementation, the order in which each step is written does not imply a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of each step should be determined by its function and possible internal logic.

[0093] Based on the same inventive concept, this disclosure also provides a sensitivity-optimized collaborative robotic arm performance testing device corresponding to the sensitivity-optimized collaborative robotic arm performance testing method. Since the principle of the device in this disclosure for solving the problem is similar to the sensitivity-optimized collaborative robotic arm performance testing method described above in this disclosure, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be described again.

[0094] Reference Figure 5 The diagram shown is a schematic of a sensitivity-optimized collaborative robotic arm performance testing device 500 provided in an embodiment of this disclosure. The device includes: The model building module 501 is used to obtain the DH parameter set of each joint in the collaborative robotic arm and build a collaborative robotic arm model based on the DH parameter set of each joint. The region identification module 502 is used to identify performance-sensitive regions in the entire workspace of the collaborative robotic arm based on the collaborative robotic arm model, and obtain at least one highly sensitive region. The trajectory generation module 503 is used to plan and optimize the test trajectory corresponding to the cooperative robotic arm based on the identified high-sensitivity area, and generate a closed test trajectory covering the high-sensitivity area. The performance test module 504 is used to conduct performance tests on the collaborative robotic arm based on the collaborative robotic arm model and the closed test trajectory, and obtain the performance test results of the collaborative robotic arm. Specifically, the trajectory generation module 503 is used for: If only one highly sensitive region is identified, then a closed test trajectory is planned within that highly sensitive region. If at least two highly sensitive regions are identified, then within each highly sensitive region, multiple sampling points located at the edge of the highly sensitive region are selected as lateral feature points according to a preset rule; and based on the lateral feature points of all highly sensitive regions, a closed test trajectory that can traverse all highly sensitive regions is constructed. Multiple trajectory test points for performance testing are determined on the closed test trajectory.

[0095] In some possible embodiments, the DH parameters include joint angles; the region identification module 502 is specifically used for: Based on the DH parameter set of each joint, the joint angle range of each joint in the collaborative robotic arm is determined; and, random sampling is performed within the joint angle range of each joint to obtain multiple sets of joint angle combinations. For each set of joint angle combinations, the spatial point corresponding to the end of the collaborative robotic arm is determined using the collaborative robotic arm model, thus obtaining the full workspace point set of the collaborative robotic arm. Uniform sampling is performed on the entire workspace point set to obtain a candidate point set covering the entire workspace of the collaborative robotic arm; Calculate the error sensitivity index for each candidate point in the candidate point set to obtain the sensitivity value corresponding to each candidate point; Based on the sensitivity values ​​and spatial locations of each candidate point in the candidate point set, a clustering analysis method is used to divide the candidate point set into multiple clusters, and at least one highly sensitive region is determined according to the average sensitivity value of each cluster.

[0096] In some possible embodiments, the region identification module 502 is specifically used for: For each candidate point, calculate the Jacobian matrix corresponding to each joint angle in the collaborative robotic arm. The Jacobian matrix represents the relationship between the end-effector pose change and the joint angle change of the collaborative robotic arm. The error sensitivity index of the candidate point is determined based on the determinant value of the Jacobian matrix.

[0097] In some possible embodiments, the trajectory generation module 503 is specifically used for: For each highly sensitive region, multiple lateral feature points selected by the preset rules are used to filter out optimized target lateral feature points from them using a genetic optimization algorithm. Connect the target side feature points from different highly sensitive areas with straight line segments; In each highly sensitive region, a cubic Bézier curve is constructed, with the target side feature point in the highly sensitive region as the endpoint and passing through the interior of the highly sensitive region; wherein, the control points of the cubic Bézier curve are selected through optimization, and the optimization objective is to maximize the average sensitivity index value of multiple sampling points evenly distributed on the cubic Bézier curve. The cubic Bézier curves corresponding to each highly sensitive region are smoothly spliced ​​with the connecting straight line segments to form a closed ∞-shaped test trajectory. Accordingly, the trajectory generation module 503 is specifically used for: Uniform sampling is performed on each segment of the cubic Bézier curve in the ∞-shaped test trajectory to obtain multiple curve test points; The multiple curve test points and the connection points between the connecting straight line segments and the cubic Bézier curve are collectively used as the trajectory test points.

[0098] In some possible embodiments, the trajectory generation module 503 is further configured to: Based on the collaborative robotic arm model, the torque requirements of each joint of the collaborative robotic arm are predicted when it moves along the closed test trajectory; and, based on the predicted torque requirements of each joint of the collaborative robotic arm when it moves, the end-effector posture of the collaborative robotic arm at each trajectory test point on the closed test trajectory is optimized and adjusted. The optimized and adjusted collaborative robotic arm posture sequence is combined with the closed test trajectory to generate the final test trajectory and posture commands for performance testing.

[0099] In some possible embodiments, the performance test results of the collaborative robotic arm include pose data and trajectory tracking data; the performance test module 504 is also used for: Based on the pose data at the trajectory test points, the pose accuracy, pose repeatability, and multi-directional pose accuracy variation of the robotic arm are calculated, and the pose accuracy index of the collaborative robotic arm is obtained based on the calculation results. Based on the trajectory tracking data, the trajectory accuracy and trajectory repeatability of the robotic arm are calculated, and the trajectory tracking accuracy index is obtained based on the calculation results. Based on the pose accuracy index and the trajectory tracking accuracy index, the overall performance of the collaborative robotic arm is evaluated, and a performance evaluation report is generated.

[0100] Based on the same technical concept, this disclosure also provides a computer device. (See also...) Figure 6 The diagram shows the structure of a computer device 600 provided in this embodiment of the present disclosure, including a processor 601, a memory 602, and a bus 603. The memory 602 stores execution instructions and includes a main memory 6021 and an external memory 6022. The main memory 6021, also called internal memory, is used to temporarily store computational data in the processor 601 and data exchanged with external memory 6022 such as a hard disk. The processor 601 exchanges data with the external memory 6022 through the main memory 6021.

[0101] In this embodiment, the memory 602 is specifically used to store application code that executes the solution of this application, and its execution is controlled by the processor 601. That is, when the computer device 600 is running, the processor 601 communicates with the memory 602 through the bus 603, so that the processor 601 executes the application code stored in the memory 602, and then executes the method described in any of the foregoing embodiments.

[0102] The memory 602 may be, but is not limited to, random access memory (RAM), read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), etc.

[0103] Processor 601 may be an integrated circuit chip with signal processing capabilities. The aforementioned processor can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this invention. The general-purpose processor can be a microprocessor or any conventional processor.

[0104] It is understood that the structures illustrated in the embodiments of this application do not constitute a specific limitation on the computer device 600. In other embodiments of this application, the computer device 600 may include more or fewer components than illustrated, or combine some components, or split some components, or have different component arrangements. The illustrated components may be implemented in hardware, software, or a combination of software and hardware.

[0105] This disclosure also provides a computer-readable storage medium storing a computer program that, when executed by a processor, performs the steps of the sensitivity-optimized collaborative robotic arm performance testing method described in the above-described method embodiments. The storage medium can be volatile or non-volatile computer-readable storage.

[0106] This disclosure also provides a computer program product carrying program code. The program code includes instructions that can be used to execute the steps of the sensitivity-optimized collaborative robotic arm performance testing method described in the above method embodiments. For details, please refer to the above method embodiments, which will not be repeated here.

[0107] The aforementioned computer program product can be implemented through hardware, software, or a combination thereof. In one optional embodiment, the computer program product is specifically embodied in a computer storage medium; in another optional embodiment, the computer program product is specifically embodied in a software product, such as a software development kit (SDK), etc.

[0108] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and devices described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. In the several embodiments provided in this disclosure, it should be understood that the disclosed systems and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division; in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Another point is that the displayed or discussed mutual coupling or direct coupling or communication connection may be through some communication interfaces; the indirect coupling or communication connection of devices or units may be electrical, mechanical, or other forms.

[0109] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0110] In addition, the functional units in the various embodiments of this disclosure can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0111] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this disclosure, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this disclosure. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0112] Finally, it should be noted that the above-described embodiments are merely specific implementations of this disclosure, used to illustrate the technical solutions of this disclosure, and not to limit them. The protection scope of this disclosure is not limited thereto. Although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features within the scope of the technology disclosed in this disclosure. Such modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this disclosure, and should all be covered within the protection scope of this disclosure.

Claims

1. A performance testing method for a collaborative robotic arm based on sensitivity optimization, characterized in that, include: Obtain the DH parameter set of each joint in the collaborative robotic arm, and construct a collaborative robotic arm model based on the DH parameter set of each joint. Based on the collaborative robotic arm model, the performance-sensitive regions of the entire workspace of the collaborative robotic arm are identified to obtain at least one highly sensitive region. Based on the identified highly sensitive area, the test trajectory corresponding to the cooperative robotic arm is planned and optimized to generate a closed test trajectory covering the highly sensitive area; Based on the collaborative robotic arm model and the closed test trajectory, a performance test was conducted on the collaborative robotic arm to obtain the performance test results of the collaborative robotic arm. The step of planning and optimizing the test trajectory corresponding to the collaborative robotic arm based on the identified highly sensitive region includes: If only one highly sensitive region is identified, then a closed test trajectory is planned within that highly sensitive region. If at least two highly sensitive regions are identified, then within each highly sensitive region, multiple sampling points located at the edge of the highly sensitive region are selected as lateral feature points according to a preset rule; and based on the lateral feature points of all highly sensitive regions, a closed test trajectory that can traverse all highly sensitive regions is constructed. Multiple trajectory test points for performance testing are determined on the closed test trajectory.

2. The method according to claim 1, characterized in that, The DH parameters include joint angles; the identification of performance-sensitive areas across the entire workspace of the collaborative robotic arm includes: Based on the DH parameter set of each joint, the joint angle range of each joint in the collaborative robotic arm is determined; and, random sampling is performed within the joint angle range of each joint to obtain multiple sets of joint angle combinations. For each set of joint angle combinations, the spatial point corresponding to the end of the collaborative robotic arm is determined using the collaborative robotic arm model, thus obtaining the full workspace point set of the collaborative robotic arm. Uniform sampling is performed on the entire workspace point set to obtain a candidate point set covering the entire workspace of the collaborative robotic arm; Calculate the error sensitivity index for each candidate point in the candidate point set to obtain the sensitivity value corresponding to each candidate point; Based on the sensitivity values ​​and spatial locations of each candidate point in the candidate point set, a clustering analysis method is used to divide the candidate point set into multiple clusters, and at least one highly sensitive region is determined according to the average sensitivity value of each cluster.

3. The method according to claim 2, characterized in that, The calculation of the error sensitivity index for each candidate point in the candidate point set includes: For each candidate point, calculate the Jacobian matrix corresponding to each joint angle in the collaborative robotic arm. The Jacobian matrix represents the relationship between the end-effector pose change and the joint angle change of the collaborative robotic arm. The error sensitivity index of the candidate point is determined based on the determinant value of the Jacobian matrix.

4. The method according to claim 1, characterized in that, The construction of a closed test trajectory capable of traversing all highly sensitive regions includes: For each highly sensitive region, multiple lateral feature points selected by the preset rules are used to filter out optimized target lateral feature points from them using a genetic optimization algorithm. Connect the target side feature points from different highly sensitive areas with straight line segments; In each highly sensitive region, a cubic Bézier curve is constructed, with the target side feature point in the highly sensitive region as the endpoint and passing through the interior of the highly sensitive region; wherein, the control points of the cubic Bézier curve are selected through optimization, and the optimization objective is to maximize the average sensitivity index value of multiple sampling points evenly distributed on the cubic Bézier curve. The cubic Bézier curves corresponding to each highly sensitive region are smoothly spliced ​​with the connecting straight line segments to form a closed ∞-shaped test trajectory. Accordingly, determining multiple trajectory test points for performance testing on the closed test trajectory includes: Uniform sampling is performed on each segment of the cubic Bézier curve in the ∞-shaped test trajectory to obtain multiple curve test points; The multiple curve test points and the connection points between the connecting straight line segments and the cubic Bézier curve are collectively used as the trajectory test points.

5. The method according to any one of claims 1 to 4, characterized in that, After generating a closed test trajectory covering the highly sensitive region, the process includes: Based on the collaborative robotic arm model, the torque requirements of each joint of the collaborative robotic arm are predicted when it moves along the closed test trajectory; and, based on the predicted torque requirements of each joint of the collaborative robotic arm when it moves, the end-effector posture of the collaborative robotic arm at each trajectory test point on the closed test trajectory is optimized and adjusted. The optimized and adjusted collaborative robotic arm posture sequence is combined with the closed test trajectory to generate the final test trajectory and posture commands for performance testing.

6. The method according to claim 1, characterized in that, The performance test results of the collaborative robotic arm include pose data and trajectory tracking data; obtaining the performance test results of the collaborative robotic arm includes: Based on the pose data at the trajectory test points, the pose accuracy, pose repeatability, and multi-directional pose accuracy variation of the robotic arm are calculated, and the pose accuracy index of the collaborative robotic arm is obtained based on the calculation results. Based on the trajectory tracking data, the trajectory accuracy and trajectory repeatability of the robotic arm are calculated, and the trajectory tracking accuracy index is obtained based on the calculation results. Based on the pose accuracy index and the trajectory tracking accuracy index, the overall performance of the collaborative robotic arm is evaluated, and a performance evaluation report is generated.

7. A performance testing device for a collaborative robotic arm based on sensitivity optimization, characterized in that, include: The model building module is used to obtain the DH parameter set of each joint in the collaborative robotic arm and build a collaborative robotic arm model based on the DH parameter set of each joint. The region identification module is used to identify performance-sensitive regions in the entire workspace of the collaborative robotic arm based on the collaborative robotic arm model, and obtain at least one highly sensitive region. The trajectory generation module is used to plan and optimize the test trajectory corresponding to the cooperative robotic arm based on the identified high-sensitivity area, and generate a closed test trajectory covering the high-sensitivity area. The performance testing module is used to conduct performance tests on the collaborative robotic arm based on the collaborative robotic arm model and the closed test trajectory, and to obtain the performance test results of the collaborative robotic arm. Specifically, the trajectory generation module is used for: If only one highly sensitive region is identified, then a closed test trajectory is planned within that highly sensitive region. If at least two highly sensitive regions are identified, then within each highly sensitive region, multiple sampling points located at the edge of the highly sensitive region are selected as lateral feature points according to a preset rule; and based on the lateral feature points of all highly sensitive regions, a closed test trajectory that can traverse all highly sensitive regions is constructed. Multiple trajectory test points for performance testing are determined on the closed test trajectory.

8. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 6.

9. A computer device, comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 6.

Images