Server component assembly method and device, electronic equipment and storage medium
By obtaining the contact force data at the end of the robot arm, using the pose prediction model and the hierarchical search mechanism of the sphere hierarchy, the robot arm movement trajectory is dynamically adjusted, and the problems of stuck and damage caused by posture deviation in the assembly of server components are solved, achieving efficient and accurate automatic assembly.
Patent Information
- Application Number
- CN202510885860.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-27
- Publication Date
- 2025-07-25
- Estimated Expiration
- 2045-06-27
AI Technical Summary
In the prior art, assembly stagnation, component damage and failure risks caused by unpredictable posture deviations during the assembly of server components, and traditional methods are insufficiently applicable in complex shape precision assembly.
By obtaining the contact force data at the end of the robot arm, the pose deviation is inferred using the pre-trained pose prediction model, and combined with the sphere hierarchy and hierarchical search mechanism, the robot arm movement trajectory is dynamically adjusted to achieve precise assembly.
It improves the flexibility and stability of automated assembly of server components, effectively deals with high-dimensional and strong uncertainty complex assembly tasks, and improves assembly efficiency and accuracy.
Smart Images

Figure CN120363222A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the technical field of robot assembly, and particularly to a method and device for assembling server components, an electronic device, and a storage medium. Background Art
[0002] With the surging demand for servers, the efficiency and accuracy of automated assembly of its core components are becoming increasingly crucial. The robotic arm is the core device for realizing automated assembly. However, in the actual assembly process, there are often unpredictable pose deviations between the component and the assembly position, which can easily lead to jams, component damage, or even assembly failure during the assembly process. Summary of the Invention
[0003] This application provides a method and device for assembling server components, an electronic device, and a storage medium, so as to at least solve the problem of assembly reliability caused by unpredictable pose deviations between components and assembly positions in related technologies.
[0004] This application provides a method for assembling server components, including: obtaining contact force data when the end of the robotic arm contacts the server component; inputting the contact force data into a pre-trained pose prediction model to obtain the pose deviation of the server component relative to the component assembly position; performing contact force matching on the pose deviation, and if the matching result indicates that the pose deviation does not match the contact force data, performing hierarchical search starting from the pose deviation to determine the target pose deviation of the server component relative to the component assembly position; and adjusting the motion trajectory of the robotic arm using the target pose deviation so that the robotic arm assembles the server component according to the adjusted motion trajectory.
[0005] This application also provides a device for assembling server components, including: a first acquisition module for obtaining contact force data when the end of the robotic arm contacts the server component; a first determination module for inputting the contact force data into a pre-trained pose prediction model to obtain the pose deviation of the server component relative to the component assembly position; a second determination module for performing contact force matching on the pose deviation, and if the matching result indicates that the pose deviation does not match the contact force data, performing hierarchical search starting from the pose deviation to determine the target pose deviation of the server component relative to the component assembly position; and an assembly module for adjusting the motion trajectory of the robotic arm using the target pose deviation so that the robotic arm assembles the server component according to the adjusted motion trajectory.
[0006] This application also provides an electronic device, including: a memory for storing a computer program; and a processor for implementing the steps of any of the above server component assembly methods when executing the computer program.
[0007] The present application also provides a computer-readable storage medium storing a computer program, where the computer program, when executed by a processor, implements the steps of any of the above server component assembly methods.
[0008] The present application also provides a computer program product including a computer program, where the computer program, when executed by a processor, implements the steps of any of the above server component assembly methods.
[0009] With the present application, when the robotic arm performs an assembly operation, contact force data generated by the contact between its end and the server component is acquired in real time; using a pre-trained pose prediction model, the pose deviation of the server component relative to the target assembly position is inferred based on the contact force; by matching and verifying the pose deviation predicted by the model with the measured contact force, if a mismatch is found, a hierarchical search mechanism is triggered to perform an efficient search starting from the predicted deviation to locate the target pose deviation that more conforms to the actual contact state; finally, the motion trajectory of the robotic arm is dynamically adjusted according to the target pose deviation to guide it to complete precise assembly. Therefore, the technical problem of assembly jams, component damage, and failure risks caused by unpredictable pose deviations in automated assembly can be solved, achieving the technical effect of improving the flexibility and stability of automated assembly of server components. BRIEF DESCRIPTION OF THE DRAWINGS
[0010] To more clearly illustrate the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and those of ordinary skill in the art can obtain other drawings based on these drawings without creative efforts.
[0011] Figure 1 It is a schematic flowchart of a server component assembly method provided by an embodiment of the present application; Figure 2 It is a schematic flowchart of another server component assembly method provided by an embodiment of the present application; Figure 3 It is a schematic flowchart of yet another server component assembly method provided by an embodiment of the present application; FIG. 4(a) is a schematic diagram of the actual relative pose between a memory module and a slot provided by an embodiment of the present application; FIG. 4(b) is a schematic diagram of an inaccurate relative pose between a memory module and a slot provided by an embodiment of the present application; Figure 5 It is a structural block diagram of a server component assembly device according to an embodiment of the present invention; Figure 6 It is a schematic hardware structure diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0012] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.
[0013] It should be noted that in the description of the present application, the terms "include", "comprise" or any other variant thereof are intended to cover a non-exclusive inclusion, such that a process, method, article or device including a series of elements not only includes those elements but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. The terms "first", "second", etc. in the present application are used to distinguish similar objects and not to describe a particular order or sequence.
[0014] With the rapid development of information technology, servers, as the core devices for data storage and computing, have an increasing demand. Server component assembly is a key link in server production, and its efficiency and quality directly affect the overall performance and production cost. Traditional assembly methods mainly rely on manual operations. However, manual assembly is not only inefficient but also prone to human errors, making it difficult to meet the requirements of large-scale production. Therefore, automated assembly technology has gradually become the mainstream trend in server component assembly.
[0015] As the core device of automated assembly technology, the control strategy of the robotic arm plays a crucial role in assembly efficiency and accuracy. Especially when facing uncertain assembly poses, the performance of the robotic arm may be affected, thus hindering the smooth progress of the assembly process. In robotic assembly tasks, two methods are usually adopted to cope with uncertainty: one is to define uncertainty through a probability model, quantify the uncertainty of the pose, and use techniques such as Bayesian estimation, Kalman filtering, and particle filtering to optimize the model by combining sensory inputs to improve accuracy; the other method is to detect the actual error and adopt a compliant motion strategy to handle uncertainty on the basis of defining the uncertainty range, such as passive compliance, active compliance based on force / torque sensing, impedance control, and force / position hybrid control.
[0016] However, these methods also have certain limitations in overcoming uncertainty in robotic assembly and are unable to effectively handle high-dimensional assembly tasks with relatively large uncertainties, especially for objects with complex shapes such as server memory modules that require precise assembly. In addition, they rely too much on assumptions, prior knowledge, or specific initial conditions, lack sufficient flexibility and generality, or are only tailored for specific assembly tasks, resulting in limited applicability in actual applications.
[0017] In view of this, the technical solution of the present invention uses a data-driven neural network model to directly predict the pose deviation of the server component based on the contact force data at the end of the robotic arm, avoiding the dependence on traditional probability models and strong prior knowledge, thereby improving the generalization ability in complex assemblies with high dimensions and strong uncertainties. At the same time, it combines the contact force matching verification based on the sphere hierarchy structure and the hierarchical search dynamic deviation correction mechanism. The theoretical contact force at the target pose is efficiently calculated through the ball tree model for matching verification. When the predicted pose is inconsistent with the actual measured force, a hierarchical space search is performed starting from the predicted value, and candidate poses are sampled and verified in real time until the true target pose deviation is locked. Finally, the robotic arm trajectory is dynamically adjusted. In this way, without the need for task customization design, the problems of real-time disturbance, high-dimensional error, and insufficient adaptability faced by traditional control methods in the assembly of complex-shaped precision components are effectively solved, and the assembly efficiency and accuracy are significantly improved.
[0018] In order to enable those skilled in the art of the present technology to better understand the solution of this application, the following further detailed description of this application will be given in conjunction with the accompanying drawings and specific embodiments.
[0019] According to an embodiment of the present invention, an embodiment of a server component assembly method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.
[0020] In this embodiment, a server component assembly method is provided, which can be used in an electronic device, such as a computing device that works in cooperation with a robotic arm. Figure 1 is a flowchart of the server component assembly method according to an embodiment of the present invention, as Figure 1 shown, the process includes the following steps: Step S101, obtain the contact force data when the end of the robotic arm contacts the server component.
[0021] A robotic arm refers to an industrial robot manipulator with at least 6 degrees of freedom, equipped with a gripper / end effector at its end for grasping server components. The robotic arm is equipped with a force / torque sensor, or a torque sensor is installed at the joint to measure the force and torque in real time when the end contacts the server component. Server components refer to the core parts that need to be assembled into the server, such as memory modules, fans, power supplies, and hard drives. Contact force data refers to the six-dimensional force and torque data measured by the force / torque sensor, which is used to reflect the mechanical state when the end of the robotic arm contacts the server component and can be represented by. Specifically, the robotic arm has at least 6 degrees of freedom, and its end effector is equipped with a force / torque sensor, which can directly measure the six-dimensional force / torque data (including forces in three directions and torques in three directions) generated when the end of the robotic arm contacts the server component; if the force / torque sensor is not directly installed, the contact force data can also be collected through the joint torque sensor and calculated to obtain the contact force data, so as to obtain the mechanical state information during contact in real time.
[0022] Step S102: Input the contact force data into a pre-trained pose prediction model to obtain the pose deviation of the server component relative to the component assembly position.
[0023] The pose prediction model is used to relate the contact force data to the relative contact pose between the server component and the assembly position and predict the pose deviation. The component assembly position refers to the position where the server component finally needs to be correctly inserted. For example, in the memory module assembly scenario, it specifically refers to the position of the memory slot. The pose deviation refers to the difference between the actual pose and the detected pose of the server component relative to the component assembly position and can be represented by . Specifically, after inputting the contact force data into the pre-trained pose prediction model, the pose prediction model analyzes the relationship between the contact force data and the server component relative to the assembly position according to the rules learned during the training process, and predicts the pose deviation between the server component and the component assembly position based on the input contact force data.
[0024] Step S103: Perform contact force matching on the pose deviation. If the matching result indicates that the pose deviation does not match the contact force data, perform a hierarchical search starting from the pose deviation to determine the target pose deviation of the server component relative to the component assembly position.
[0025] The target pose deviation refers to the pose deviation that is finally found and successfully matched through the hierarchical search process. Specifically, verify the contact force matching of the pose deviation according to the preset matching rules. If the pose deviation and the contact force data fail to match successfully, adjust through the hierarchical search strategy. This search process starts from the current pose deviation and performs a hierarchical search. By adjusting the search range, gradually find a target pose deviation that meets the preset matching conditions to ensure that the server component can be accurately docked with the component assembly position.
[0026] Step S104, adjust the motion trajectory of the robotic arm by using the target pose deviation, so that the robotic arm assembles the server component according to the adjusted motion trajectory.
[0027] The motion trajectory refers to the spatial path and pose change that the robotic arm gripper / end effector follows when carrying the server component from the current position to the component assembly position. Specifically, the target pose deviation reflects the real difference between the actual pose and the detected pose of the server component. Based on the target pose deviation, the target pose of the end of the robotic arm is recalculated. Based on this target pose, combined with the inverse kinematics and trajectory planning algorithm of the robotic arm, the motion path of the robotic arm from the current position to the component assembly position is adjusted, so that the robotic arm moves according to the adjusted motion trajectory, ensuring that the server component is accurately inserted into the component assembly position based on real-time force feedback and pose correction to complete the assembly task.
[0028] The server component assembly method provided in this embodiment can sense the mechanical state during the assembly process in real time by obtaining the contact force data when the end of the robotic arm contacts the server component, providing a basis for precise adjustment. Input the contact force data into a pre-trained pose prediction model to obtain the pose deviation in a data-driven manner, without relying on complex physical modeling, improving the versatility and flexibility of the method. Through the contact force matching and hierarchical search mechanism, when the predicted pose deviation is inaccurate, it can search and verify step by step from local to global to ensure finding the accurate target pose deviation, effectively coping with the uncertainty in the assembly. Finally, based on the target pose deviation, the motion trajectory of the robotic arm is dynamically adjusted to realize the closed-loop control from perception, prediction to execution, ensuring the efficiency and accuracy of the server component assembly.
[0029] In this embodiment, a server component assembly method is provided, which can be used in an electronic device, such as a computing device that works in cooperation with a robotic arm. Figure 2 is a flowchart of the server component assembly method according to an embodiment of the present invention, as Figure 2 shown, the process includes the following steps: Step S201, obtain the first surface data of the server component and the second surface data of the component assembly position.
[0030] The first surface data refers to the geometric surface information of the server component, which can be represented in the form of a point cloud or a polygon mesh. The second surface data refers to the geometric surface information of the component assembly position, which can also be represented in the form of a point cloud or a polygon mesh. Specifically, the first surface data and the second surface data can be extracted from a 3D model. For example, a 3D model of the server component or the component assembly position is constructed using CAD software, and the polygon mesh data of its surface is exported. Alternatively, it can be collected through a sensing device, such as using a laser scanner, a 3D camera, etc. to scan the server component or the component assembly position to obtain the point cloud data of its surface.
[0031] Step S202: Use a preset space division method to perform sphere hierarchical division on the first surface data and the second surface data, generating a first sphere hierarchical structure corresponding to the first surface data and a second sphere hierarchical structure corresponding to the second surface data.
[0032] The preset space division method refers to a recursive space division algorithm for constructing a preset ball tree. The first sphere hierarchical structure refers to a sphere tree-like structure generated by performing preset space division on the first surface data of the server component, and can also be called the first ball tree. The second sphere hierarchical structure refers to a tree-like structure formed after performing sphere hierarchical division on the second surface data of the component assembly position, and can also be called the second ball tree. Specifically, the preset space division method uses a sphere tree recursive space division algorithm to simplify complex geometries by dividing the first surface data of the server component and the second surface data of the component assembly position into multiple sphere hierarchical structures. Space division is performed on the first surface data to generate a sphere tree-like structure layer by layer. Each layer of sphere nodes represents a spatial region at a certain level. The spheres recursively decompose a large-scale region into smaller spatial units, thus forming the first sphere hierarchical structure. The same process also applies to the second surface data to form the second sphere hierarchical structure.
[0033] For example, when the server component is a memory module, assuming the memory module is 80mm long, 30mm wide, and 5mm high, first enclose the memory module completely with a sufficiently large sphere (such as a sphere with a radius of 50mm), and this large sphere is the root node of the ball tree. Then, each part of the memory module is subdivided. For example, for the edge part of the memory module, since it is relatively slender, smaller spheres with a radius of about 2 - 5mm can be used to represent it; for the spring pieces of the memory slot, spheres with a radius of 1 - 3mm can be used according to the thickness and shape of the spring pieces. In terms of the hierarchical depth, if the memory module and slot structure are not particularly complex, a ball tree structure with 3 - 5 layers is sufficient to describe its geometric features.
[0034] The server component assembly method provided in this embodiment obtains the first surface data of the server component and the second surface data of the component assembly position, providing accurate original information for subsequent geometric modeling. Using a preset space partitioning method to perform sphere hierarchy partitioning, generating corresponding first and second sphere hierarchy structures. This sphere tree representation method does not rely on an accurate object shape model and can efficiently represent the geometric features of components and assembly positions in a hierarchical structure. During the assembly process, non-collision parts can be quickly excluded, and the collision spheres can be located in real time, effectively improving the processing ability of complex shapes and uncertainties during the assembly process.
[0035] In some alternative embodiments, using a preset space partitioning method to perform sphere hierarchy partitioning on the surface data to generate a sphere hierarchy structure corresponding to the surface data includes: Step a1, obtaining the geometric center of the surface data and determining the geometric center as the root node sphere center.
[0036] The geometric center refers to the coordinate mean of all data points in the surface data (such as point cloud / polygon mesh data of the server component or component assembly position). The root node sphere center refers to the center of the sphere at the topmost layer of the sphere tree. This sphere is used to initially enclose the entire object surface data and is the starting point of the sphere hierarchy structure. Specifically, the mean of the three-dimensional coordinates (x, y, z) of all data points in the surface data is the geometric center. When calculating specifically, the sum of the x coordinates of all points is divided by the number of points to obtain the x mean value, and the y and z mean values are obtained in the same way. This mean coordinate is the geometric center. Using this geometric center as the root node sphere center of the sphere tree, a root node sphere that can initially enclose all surface data is constructed.
[0037] Step a2, determining the data point farthest from the root node sphere center in the surface data as the first child node, and determining the data point farthest from the first child node as the second child node.
[0038] The first child node refers to the data point farthest from the root node sphere center in the surface data, serving as the first child node of the root node. The second child node refers to the data point farthest from the first child node in the surface data, serving as the second child node of the root node. Specifically, calculate the spatial distance from each data point in the surface data to the root node sphere center, and the point with the maximum distance is the first child node. Using the first child node as the new reference point, calculate the distance from all data points to this point again, and the point with the maximum distance is determined as the second child node.
[0039] Step a3, allocating the remaining data points except the first child node and the second child node to the first cluster corresponding to the first child node and the second cluster corresponding to the second child node according to the principle of the nearest distance.
[0040] The remaining data points refer to all data points in the surface data except the first child node and the second child node. The nearest distance principle means that when allocating the remaining data points to the first child node or the second child node, compare the spatial distances of the data points to the two child nodes and assign the data points to the cluster corresponding to the nearer child node. The first cluster refers to the subset composed of data points closer to the first child node. The second cluster refers to the subset composed of data points closer to the second child node. Specifically, for the remaining data points, calculate their spatial distances to the first child node and the second child node respectively. If the distance to the first child node is closer, it is allocated to the first cluster corresponding to the first child node; if the distance to the second child node is closer, it is allocated to the second cluster corresponding to the second child node. In this way, the surface data is divided into two initial subsets.
[0041] Step a4, recursively divide the first cluster and the second cluster until a preset stop condition is met, generating a sphere hierarchy corresponding to the surface data.
[0042] Among them, the surface data is the first surface data, and the sphere hierarchy is the first sphere hierarchy corresponding to the first surface data, or the surface data is the second surface data, and the sphere hierarchy is the second sphere hierarchy corresponding to the second surface data.
[0043] The preset stop condition refers to the termination condition when recursively dividing the clusters. For example, it can include that the radius of the sub-sphere is less than a preset value or the number of data points included in the cluster is less than a specified quantity, etc. Specifically, recursively repeat the above sphere division process for the first cluster and the second cluster, calculate the geometric center of its data points as the sub-sphere center for each cluster; then, among the data points of this cluster, select the point farthest from the sub-sphere center as the first child node, and then select the point farthest from the first child node as the second child node; then calculate the Euclidean distances of the remaining data points to the two child nodes according to the nearest distance principle and allocate them to the cluster corresponding to the nearer child node. The termination conditions of the recursion can include: the radius of the sub-sphere is less than a preset threshold (such as the radius of the sphere at the edge of the memory module ≤ 2 mm), the number of points included in the sub-cluster is too small (such as ≤ 5 points) or the maximum recursion depth is reached (such as 3 - 5 layers). Finally, a hierarchical ball tree structure is generated.
[0044] For example, when the server component is a memory module, the process of converting the geometric shape of the memory module into a ball tree includes: calculating the smallest enclosing ball that can completely contain the cuboid (i.e., the memory module). Measure the length, width, and height of the cuboid, find its geometric center as the center of the enclosing ball, and the distance from the center of the ball to the farthest vertex of the cuboid is the radius of the enclosing ball. Assume the length of the memory module is L, the width is W, and the height is H, and the center coordinates of the ball are , where and are the coordinates of the diagonal vertices of the cuboid, and the radius Select the point A that is farthest from the center of the enclosing sphere within the enclosing sphere, and then select the point B that is farthest from point A. According to these two points, the points within the enclosing sphere (corresponding to the geometric shape points of the memory module) are assigned to the cluster closest to A or B. Calculate the centers of the two clusters respectively, and the minimum radius that can enclose all the points within the cluster, thus obtaining two sub-hyper-spheres. For example, if the distance from point P to A is less than the distance to B, then point P is assigned to the cluster corresponding to A. Repeat the above partitioning steps for the obtained sub-hyper-spheres, continuously recursively. Each recursive partitioning will make the sub-hyper-spheres approximate the local shape of the memory module more precisely. When the sub-hyper-spheres meet specific stopping conditions, such as the radius being less than a certain threshold or the number of points contained being small, stop the partitioning, and finally form a ball tree that describes the geometric shape of the memory module.
[0045] When the component assembly position is the memory slot, the process of converting the geometric shape of the memory slot into a ball tree includes: The memory slot is a slit with spring clips. First, regard it as an irregular shape as a whole, and find the smallest enclosing sphere that can accommodate the entire slot (including the spring clips). Measure the key dimensions of the slot, such as length, width, depth, and the position and shape of the spring clips, to determine the center and radius of the enclosing sphere. If the length of the slot is and the width is and the depth is , the center of the sphere can be determined according to the geometric center of the slot, and the radius is the distance from the center of the sphere to the farthest point of the slot (including the spring clips). Since the slot has spring clips, special consideration should be given to the position and shape of the spring clips during the internal sphere partitioning. Select the point that is farthest from the center of the sphere and located in the area of the spring clip as one of the partitioning points, and then select a suitable far point, and assign the points to different clusters according to the distance. For the spring clip part, more detailed partitioning can be carried out separately because the contact between the spring clip and the memory module during the assembly process is crucial. For example, if the spring clip has two protruding parts, the endpoints of the protruding parts can be used as the partitioning points respectively to form more precise sub-hyper-spheres to describe the shape of the spring clip. Similar to the processing method of the memory module, continuously recursively partition the divided sub-hyper-spheres. During the partitioning process, for the slit part of the slot, make a reasonable partition according to the direction and width change of the slit. When the sub-hyper-spheres meet certain accuracy requirements, such as being able to accurately describe the key geometric features of the slot (slit width change, spring clip position, etc.), stop the partitioning and construct the ball tree of the memory slot.
[0046] In the above embodiments, by obtaining the geometric center of the surface data as the center of the root node sphere, the symmetry and integrity of the initial division of the sphere tree structure are ensured. Determining the point farthest from the root node as the first child node and the point farthest from the first child node as the second child node can maximize the spatial coverage of the initial division and improve the subsequent data grouping efficiency. Allocating the remaining data points to the corresponding clusters according to the principle of the shortest distance makes the distribution of data points within each cluster compact and reduces the ineffective search area. Recursively dividing the clusters and terminating with preset conditions can not only adaptively generate a hierarchical structure according to the geometric complexity of the object, but also avoid over-division while ensuring the model accuracy. The finally generated sphere hierarchical structure can efficiently represent the geometric features of the components and the assembly positions.
[0047] Step S203: Obtain the contact force data when the end of the robotic arm contacts the server component. For details, please refer to Figure 1 Step S101 of the illustrated embodiment, which will not be elaborated here.
[0048] Step S204: Input the contact force data into the pre-trained pose prediction model to obtain the pose deviation of the server component relative to the component assembly position. For details, please refer to Figure 1 Step S102 of the illustrated embodiment, which will not be elaborated here.
[0049] Step S205: Perform contact force matching on the pose deviation. If the matching result indicates that the pose deviation does not match the contact force data, perform hierarchical search starting from the pose deviation to determine the target pose deviation of the server component relative to the component assembly position.
[0050] Specifically, the above step S205 includes: Step S2051: Obtain the initial relative position between the server component and the component assembly position.
[0051] The initial relative position refers to the initial pose of the server component relative to the component assembly position detected through technologies such as vision or auxiliary positioning, and can be represented by 2 Specifically, the initial relative position between the server component and the component assembly position can be obtained through vision perception or auxiliary positioning technologies. For example, after a camera captures an image, it is calculated through a pose detection algorithm. The initial relative position is affected by the accuracy of the image acquisition device and the position detection algorithm, and there are uncertainties in perception and algorithm detection.
[0052] Step S2052: Use the product of the initial relative position and the pose deviation to determine the target contact pose between the server component and the component assembly position.
[0053] The target contact pose refers to the theoretical contact pose adjusted by the pose deviation and can be represented by Representation. Specifically, the target contact pose is determined by the formula = 2 Calculated by substituting into the formula, that is, by combining the initial relative position 2 with the pose deviation the target contact pose can be directly derived.
[0054] Step S2053: Based on the positional relationship between the first sphere hierarchy and the second sphere hierarchy, determine the pair of collision spheres corresponding to the component assembly position of the server component in the target contact pose.
[0055] A pair of collision spheres refers to a pair of spheres that intersect in the first sphere hierarchy of the server component and the second sphere hierarchy of the component assembly position in the target contact pose. Specifically, in the target contact pose, starting from the root node spheres corresponding to the first sphere hierarchy and the second sphere hierarchy respectively, compare layer by layer. If the root node spheres do not intersect, there is no overall collision; if they intersect, recursively check the child node spheres. When comparing to the leaf nodes, if a pair of spheres intersects, it is determined as a pair of collision spheres. For example, when the small spheres on the edge of the memory module intersect with the small spheres of the slot spring piece, this pair of spheres corresponds to the actual contact area.
[0056] Step S2054: Use the pair of collision spheres to perform contact force matching on the pose deviation to obtain a matching result.
[0057] Through the parameter data of the pair of collision spheres and the preset matching rules, perform contact force matching verification on the pose deviation to obtain a matching result. If the matching result indicates that the pose deviation matches the contact force data, then use the pose deviation to adjust the motion trajectory of the robotic arm so that the robotic arm assembles the server component according to the adjusted motion trajectory.
[0058] In some alternative embodiments, step S2054 includes: Step b1: Obtain the penetration depth of the pair of collision spheres, and use the penetration depth and the preset contact stiffness coefficient to determine the target contact force between the server component and the component assembly position.
[0059] The penetration depth refers to the depth value of the intersecting part of the two spheres in the pair of collision spheres of the first sphere hierarchy and the second sphere hierarchy in the target contact pose, which reflects the actual contact deformation amount of the object surface and can be represented by d. The preset contact stiffness coefficient is a physical parameter preset based on the material and geometric parameters of the object corresponding to the server component and the component assembly position, and can be represented by k. The target contact force refers to the theoretical contact force calculated using the penetration depth of the pair of collision spheres and the preset contact stiffness coefficient, and can be represented by Representation. Specifically, at the target contact pose, for the pair of collision spheres that intersect in the first sphere hierarchy of the server component and the second sphere hierarchy of the component assembly position, the penetration depth is the depth value of the intersecting part of the two spheres (that is, when the distance between the two sphere centers is less than the sum of the two radii, the difference between the sum of the radii and the distance between the sphere centers). After obtaining this penetration depth, according to Hooke's law, the target contact force is calculated by the formula = k×d.
[0060] Step b2, if the contact force distance between the target contact force and the contact force data is less than the preset threshold, it is determined that the pose deviation matches the contact force data.
[0061] The contact force distance refers to the distance between the target contact force and the measured contact force data, which is used to quantify the difference degree between the two. For example, it can be the Euclidean distance. The preset threshold is the distance threshold determined by statistical analysis of a large number of measured data. Specifically, calculate the contact force distance between the target contact force and the contact force data. If the contact force distance is less than the preset threshold, it indicates that the theoretical contact force calculated based on the pose deviation and the actually detected contact force are within an acceptable range of difference. Thus, it is determined that the pose deviation matches the contact force data, that is, the current pose deviation can accurately reflect the actual assembly pose of the server component.
[0062] In the above embodiments, the penetration depth of the pair of collision spheres is obtained and combined with the preset contact stiffness coefficient to determine the target contact force. By combining the geometric penetration amount with the physical and mechanical properties, a calculation model of the theoretical contact force is established through Hooke's law, making the calculation of the contact force have clear physical meaning and interpretability. By calculating the contact force distance between the target contact force and the contact force data and comparing it with the preset threshold to determine the matching situation, a quantitative matching determination criterion is formed, realizing an accurate conversion from geometric collision to mechanical verification, ensuring that the accuracy verification process of the pose deviation is objective and quantifiable, thereby providing a reliable matching basis for subsequent hierarchical search and effectively guaranteeing the accuracy and reliability of the judgment of the contact state during the assembly process.
[0063] Step S2055, if the matching result indicates that the pose deviation does not match the contact force data, perform a hierarchical search starting from the pose deviation to determine the target pose deviation of the server component relative to the component assembly position. For details, please refer to Figure 1 Step S103 of the embodiment shown, which will not be elaborated here.
[0064] Step S206, use the target pose deviation to adjust the motion trajectory of the robotic arm so that the robotic arm assembles the server component according to the adjusted motion trajectory. For details, please refer to Figure 1 Step S104 of the embodiment shown, which will not be elaborated here.
[0065] The server component assembly method provided in this embodiment provides a benchmark for the calculation of the target contact pose by obtaining the initial relative position between the server component and the assembly position, ensuring that the superposition calculation of the pose deviation has a clear reference framework. The target contact pose is determined by multiplying the initial relative position by the pose deviation, and the predicted deviation is incorporated into the theoretical pose through operations, making the target contact pose more suitable for the uncertainties in the actual assembly scenario. The collision sphere pairs are determined based on the positional relationship between the first sphere hierarchy and the second sphere hierarchy, enabling the rapid positioning of the contact area with the help of the hierarchical geometric representation of the ball tree, avoiding global search, and significantly improving the collision detection efficiency. The contact force matching is performed using the collision sphere pairs, and the theoretical contact force is calculated through the geometric penetration depth and the physical model, realizing the accurate verification of the predicted pose deviation, ensuring the accuracy of the pose adjustment, and thus forming a closed-loop control from pose prediction to contact verification, effectively improving the robustness and accuracy of the assembly process.
[0066] In this embodiment, a server component assembly method is provided, which can be used in electronic devices, such as computing devices that cooperate with robotic arms. Figure 3 It is a flowchart of the server component assembly method according to an embodiment of the present invention, as Figure 3 shown, and this process includes the following steps: Step S301, obtain the first surface data of the server component and the second surface data of the component assembly position. For details, please refer to Figure 2 Step S201 of the shown embodiment, which will not be elaborated here.
[0067] Step S302, perform sphere hierarchy partitioning on the first surface data and the second surface data using a preset space partitioning method to generate a first sphere hierarchy corresponding to the first surface data and a second sphere hierarchy corresponding to the second surface data. For details, please refer to Figure 2 Step S202 of the shown embodiment, which will not be elaborated here.
[0068] Step S303, obtain the contact force data when the end of the robotic arm contacts the server component. For details, please refer to Figure 2 Step S203 of the shown embodiment, which will not be elaborated here.
[0069] Step S304, input the contact force data into a pre-trained pose prediction model to obtain the pose deviation of the server component relative to the component assembly position.
[0070] Specifically, the pose prediction model includes a feature extraction layer and a deviation prediction layer, and the feature extraction layer is cascaded with the deviation prediction layer. The above step S304 includes: Step S3041, extract the contact force direction feature between the server component and the component assembly position from the contact force data through the feature extraction layer of the pose prediction model.
[0071] The feature extraction layer refers to the first-stage network in the pose prediction model, which adopts a two-hidden-layer structure, with a non-linear function as the activation function and the Adam gradient optimization algorithm. The contact force direction feature refers to the three-dimensional vector extracted from the contact force data by the feature extraction layer, representing the direction of the equivalent force at the contact point. Specifically, the feature extraction layer of the pose prediction model has a two-hidden-layer structure, using a non-linear activation function and the Adam optimization algorithm. After the six-dimensional contact force data is input, the feature extraction layer extracts the key features that can characterize the contact force direction from the data through the calculation of the multi-layer neural network, and finally outputs a three-dimensional vector, which represents the direction of the equivalent force at the contact point and is related to the geometric contact position on the object surface. For example, in the scenario of inserting a memory module, if the contact occurs at the edge or plane, the direction of the 3D vector can reflect the normal direction of the contact surface or the geometric features of the relative pose deviation, providing key direction information for subsequent uncertainty prediction.
[0072] Step S3042: Input the contact force data and the contact force direction feature into the deviation prediction layer of the pose prediction model to obtain the pose deviation of the server component relative to the component assembly position.
[0073] The deviation prediction layer refers to the second-stage network of the pose prediction model. The input is the six-dimensional contact force data and the three-dimensional contact force direction feature output by the feature extraction layer. The activation function is the rectified linear unit ReLU, the loss function is the mean squared error, and the Adam algorithm is used for optimization. Specifically, the input of the deviation prediction layer is the six-dimensional contact force data and the three-dimensional contact force direction feature output by the feature extraction layer. It uses the ReLU activation function and the mean squared error loss function, and optimizes the model parameters through the Adam algorithm. After the input data is processed by the deviation prediction layer, a six-dimensional vector is output , where represents the roll, pitch, and yaw angles of the relative pose, represents the relative position. This six-dimensional vector is the pose deviation of the server component relative to the assembly position.
[0074] In the server component assembly method provided in this embodiment, the pose prediction model adopts a cascaded structure of a feature extraction layer and a deviation prediction layer. By extracting the contact force direction feature from the contact force data through the feature extraction layer, it can effectively refine the key direction information in the contact force data, reduce the data dimension while retaining the core features, and provide more targeted input for subsequent pose deviation prediction. By jointly inputting the contact force data and the contact force direction feature into the deviation prediction layer, combining the original contact force data with the refined direction feature, the mapping relationship between the contact state and the pose deviation can be more comprehensively captured, enabling the pose deviation output by the deviation prediction layer to more accurately reflect the pose deviation of the server component relative to the assembly position, and improving the accuracy and reliability of pose prediction.
[0075] In some alternative embodiments, a high-sensitivity microphone array and vibration sensors can be added at the end of the robotic arm to synchronously collect acoustic fingerprint features (such as component friction spectrum, snap engagement sound pressure) and high-frequency vibration waveforms during the assembly process, and construct an acoustic-vibration-force triple perception system. Design a lightweight Bayesian evaluation network to analyze the contact force signal-to-noise ratio, acoustic fingerprint energy entropy, and model confidence in real time, and output a dynamic uncertainty coefficient κ∈[0,1] (κ>0.7 triggers a high-uncertainty warning). Based on the κ value, intelligently regulate the hierarchical search strategy: in low-uncertainty scenarios (κ≤0.3), maintain the default step size to accelerate convergence; in high-uncertainty scenarios (κ>0.7), activate multimodal fusion, input the acoustic-vibration time-frequency features (Mel cepstral coefficients extracted by wavelet transform) and contact force data into a graph neural network to reconstruct the contact mechanics model, and adaptively expand the search step size according to δ'=δ·(1 + 2κ), and call the historical similar case library to initialize the search direction. After successful deviation correction, automatically bind the acoustic-vibration, contact force, and pose deviation data to construct an enhanced sample, online fine-tune the weights of the fully connected layer of the pose prediction model through the gradient clipping protection mechanism, and distill the key rules (such as the mapping relationship between "1.2kHz vibration peak + 200Hz acoustic fingerprint mutation" and "Z-axis offset dominance") into lightweight rules and store them in the edge node to form a lifelong learning closed loop.
[0076] In the above embodiment, through the acoustic-vibration-force triple perception system and the adaptive decision-making mechanism, the robustness and evolutionary ability of the assembly system under complex disturbances are significantly improved. At the hardware level, expand the acoustic-vibration sensing channels to provide redundant sensing dimensions when the force sensor fails to ensure data integrity. At the algorithm level, quantify the environmental interference intensity with the dynamic uncertainty coefficient κ to achieve intelligent optimization of hierarchical search. In low-κ scenarios, maintain efficient convergence, and in high-κ scenarios, resist noise interference through multimodal fusion and adaptive step size expansion, thereby reducing redundant search paths. At the knowledge level, construct a lifelong learning closed loop, online fine-tune the model to improve the convergence speed of the assembly error of new components, and distill the key rules into a lightweight knowledge base of the edge node, endowing the system with the ability of autonomous evolution to cope with sudden tolerances, and finally forming a "highly elastic assembly brain" that becomes more accurate with use.
[0077] Step S305: Perform contact force matching for the pose deviation. If the matching result indicates that the pose deviation does not match the contact force data, then start a hierarchical search from the pose deviation to determine the target pose deviation of the server component relative to the component assembly position.
[0078] Specifically, the above step S305 includes: Step S3051: Perform contact force matching for the pose deviation. If the matching result indicates that the pose deviation does not match the contact force data, then, with the pose deviation as the center point, expand the search space in ascending order of hierarchy, and sample at a preset step size within each layer to generate candidate pose deviations.
[0079] Candidate pose deviation: When the original pose deviation does not match the contact force data, it is the possible pose deviation value generated by sampling at a preset step size within each layer with the pose deviation as the center point and expanding the search space in ascending order of hierarchy. Specifically, first, based on the pose deviation and the sphere hierarchy structure (i.e., the ball tree model), calculate the theoretical contact force (i.e., the target contact force), and compare its measured force (i.e., the contact force data) with the contact force distance. If the distance is greater than the preset threshold, it is determined to be unmatched. At this time, with the pose deviation as the center point, expand the search space in ascending order of hierarchy: layer 0 only contains , and layer n (n≥1) takes as the center, and generate new candidate pose deviations in its neighborhood according to the preset sampling rule. The search range of each layer expands with the increase of the hierarchy (for example, level = 1 corresponds to a small neighborhood, level = 2 corresponds to a larger range). Among them, the preset sampling rule is that in each dimension, with as the corresponding parameter of the center, generate discrete points at a preset step size (such as Δx±δ, Δy±δ, etc.). For example, when level = 1, the step size is small (such as δ = 0.1mm / 0.01rad), and when level = 2, the step size is doubled (δ = 0.2mm / 0.02rad), and so on. The search range of each layer is the extended neighborhood of the previous layer. By increasing the sampling step size or the number of sampling points in each dimension, the search space is gradually expanded to ensure coverage of the possible range. In addition, the breadth-first order can be adopted. First, search the nearest neighborhood (level = 0). If it fails, expand to a larger range (level = 1, level = 2, etc.) to ensure that the most likely candidate points are verified first and improve the search efficiency.
[0080] Step S3052: Perform contact force matching for the candidate pose deviations to obtain the target matching result corresponding to the candidate pose deviations.
[0081] The target matching result refers to the judgment result obtained by performing contact force matching on the candidate pose deviation. Specifically, for each candidate pose deviation , first use the product of the initial relative position and to determine the target contact pose, then determine the pair of collision spheres based on the positional relationship between the first sphere hierarchy and the second sphere hierarchy, calculate the penetration depth of the pair of collision spheres, and obtain the theoretical contact force through Hooke's law by combining the preset contact stiffness coefficient k . Compare with the measured contact force data to calculate the contact force distance. If the distance is less than the preset threshold, the target matching result indicates that the candidate pose deviation matches the contact force data; otherwise, it does not match.
[0082] Step S3053: When the target matching result indicates that the candidate pose deviation matches the contact force data, determine the candidate pose deviation as the target pose deviation.
[0083] After performing contact force matching on the candidate pose deviation, if the target matching result shows that the corresponding theoretical contact force and the measured force have a contact force distance less than the preset threshold, it is determined that the candidate pose deviation can accurately reflect the actual assembly pose of the server component. At this time, the search is stopped, and the candidate pose deviation is determined as the target pose deviation , which is used for subsequent adjustment of the robotic arm's motion trajectory to ensure precise assembly of the server component.
[0084] Step S306: Use the target pose deviation to adjust the motion trajectory of the robotic arm so that the robotic arm assembles the server component according to the adjusted motion trajectory.
[0085] Substitute the target pose deviation into the formula = 2 (where 2 is the detected relative pose, and is the actual relative pose) to recalculate the target pose of the end of the robotic arm. Based on this target pose, combined with the inverse kinematics and trajectory planning algorithms of the robotic arm, adjust the motion path of the robotic arm from the current position to the assembly position, so that the robotic arm moves according to the adjusted trajectory, ensuring that the server component (such as a memory module) is accurately inserted into the assembly position (such as a memory slot) based on real-time force feedback and pose correction to complete the assembly task.
[0086] The server component assembly method provided in this embodiment takes the pose deviation as the center point and expands the search space hierarchically in an increasing manner. It adopts a search strategy from local to global, preferentially verifies the most likely pose deviation areas, and improves the search efficiency. At each layer, candidate pose deviations are sampled according to a preset step size, and the search accuracy is controlled by the quantified step size to ensure the orderly expansion of the search space. The candidate pose deviations are subjected to contact force matching, and the target pose deviation is determined based on the target matching result, forming a closed-loop mechanism of prediction, verification, and correction. This mechanism can not only find the correct solution through search when the predicted pose deviation is inaccurate, but also ensure the accuracy of the target pose deviation by means of contact force matching, effectively coping with the pose uncertainty in the assembly process and improving the robustness and adaptability of the method.
[0087] In the following of this embodiment, a scenario where the server component is a memory module and the component assembly position is a memory slot will be used to exemplarily illustrate the above server component assembly method. As shown in FIGS. 4(a) and 4(b), FIG. 4(a) shows the actual relative pose between the memory module and the slot, and FIG. 4(b) shows the inaccurate relative pose between the memory module and the slot.
[0088] The first surface data of the memory module and the second surface data of the memory slot (a slit with a spring piece) are obtained by laser scanning, and a ball tree is constructed using a preset space partitioning method: the geometric center of the surface data of the memory module is calculated as the center of the root node sphere, the point farthest from the center of the sphere is selected as the first child node, and then the point farthest from this point is selected as the second child node. The remaining data points are assigned to two clusters according to the nearest distance, and the division is recursively performed until the radius of the sub-sphere is less than 5 mm, generating a first sphere hierarchy structure including edge small spheres (radius 2 - 5 mm). Similarly, a second sphere hierarchy structure including spring piece small spheres (radius 1 - 3 mm) is constructed for the slot. When the robotic arm grabs the memory module, the force / torque sensor obtains six-dimensional contact force data and inputs it into a two-stage pose prediction model: the feature extraction layer extracts the three-dimensional contact force direction feature from the data, and the deviation prediction layer outputs the pose deviation by combining this feature with the contact force data. The target contact pose is obtained based on the product of the initial relative position and the pose deviation. The collision sphere pairs between the edge of the memory module and the spring piece of the slot are determined through the collision detection of the two ball trees, the penetration depth is calculated, and the theoretical contact force is obtained using Hooke's law. Compare with the measured force If the Euclidean distance is greater than a preset threshold (determined by data statistics), the search space is expanded hierarchically with the pose deviation as the center. At layer 0, verification is performed. At layer 1, candidate pose deviations are sampled in the neighborhood of ±0.1 mm / 0.01 rad, and the step size is gradually increased layer by layer until a candidate value that matches is found and determined as the target pose deviation. Finally, is substituted into the formula =2 , according to the calculated Adjust the movement trajectory of the robotic arm so that the memory module is accurately inserted into the slot according to the corrected path.
[0089] Through the description of the above embodiments, those skilled in the art can clearly understand that the method according to the above embodiments can be implemented by means of software plus a necessary general hardware platform. Of course, it can also be implemented by hardware, but in many cases the former is a better implementation method.
[0090] This embodiment provides a server component assembly device, as Figure 5 shown, including: The first acquisition module 501 is used to acquire the contact force data when the end of the robotic arm contacts the server component; The first determination module 502 is used to input the contact force data into a pre-trained pose prediction model to obtain the pose deviation of the server component relative to the component assembly position; The second determination module 503 is used to perform contact force matching on the pose deviation. If the matching result indicates that the pose deviation does not match the contact force data, a hierarchical search is performed starting from the pose deviation to determine the target pose deviation of the server component relative to the component assembly position; The assembly module 504 is used to adjust the movement trajectory of the robotic arm by using the target pose deviation so that the robotic arm assembles the server component according to the adjusted movement trajectory.
[0091] In some alternative embodiments, the above server component assembly device further includes: The second acquisition module is used to acquire the first surface data of the server component and the second surface data of the component assembly position; The division module is used to perform sphere hierarchical division on the first surface data and the second surface data by using a preset space division method to generate a first sphere hierarchical structure corresponding to the first surface data and a second sphere hierarchical structure corresponding to the second surface data.
[0092] In some alternative embodiments, the division module includes: The first acquisition sub-module is used to acquire the geometric center of the surface data and determine the geometric center as the root node sphere center; The first determination sub-module is used to determine the data point farthest from the root node sphere center in the surface data as the first sub-node and the data point farthest from the first sub-node as the second sub-node; The allocation sub-module is used to allocate the remaining data points except the first sub-node and the second sub-node to the first cluster corresponding to the first sub-node and the second cluster corresponding to the second sub-node according to the nearest distance principle; A sub - division module, which is used to recursively divide the first cluster and the second cluster until a preset stop condition is met, and generate a sphere hierarchy corresponding to the surface data; wherein, the surface data is the first surface data, and the sphere hierarchy is the first sphere hierarchy corresponding to the first surface data, or the surface data is the second surface data, and the sphere hierarchy is the second sphere hierarchy corresponding to the second surface data.
[0093] In some alternative embodiments, the second determination module 503 includes: A second acquisition sub - module, which is used to acquire the initial relative position between the server component and the component assembly position; A second determination sub - module, which is used to determine the target contact pose between the server component and the component assembly position by using the product of the initial relative position and the pose deviation; A third determination sub - module, which is used to determine the collision sphere pair corresponding to the server component and the component assembly position in the target contact pose based on the positional relationship between the first sphere hierarchy and the second sphere hierarchy; A first matching sub - module, which is used to perform contact force matching on the pose deviation by using the collision sphere pair to obtain a matching result.
[0094] In some alternative embodiments, the matching sub - module includes: An acquisition unit, which is used to acquire the penetration depth of the collision sphere pair, and determine the target contact force between the server component and the component assembly position by using the penetration depth and a preset contact stiffness coefficient; A determination unit, which is used to determine that the pose deviation matches the contact force data if the contact force distance between the target contact force and the contact force data is less than a preset threshold.
[0095] In some alternative embodiments, the pose prediction model includes a feature extraction layer and a deviation prediction layer; the feature extraction layer is cascaded with the deviation prediction layer; the first determination module 502 includes: An extraction sub - module, which is used to extract the contact force direction feature between the server component and the component assembly position from the contact force data through the feature extraction layer of the pose prediction model; A fourth determination sub - module, which is used to input the contact force data and the contact force direction feature into the deviation prediction layer of the pose prediction model to obtain the pose deviation of the server component relative to the component assembly position.
[0096] In some alternative embodiments, the second determination module 503 further includes: A sampling sub - module, which is used to expand the search space in the order of increasing levels with the pose deviation as the center point, and sample at a preset step size within each level range to generate candidate pose deviations; A second matching sub-module, configured to perform contact force matching on the candidate pose deviation to obtain a target matching result corresponding to the candidate pose deviation; A fifth determination sub-module, configured to determine the candidate pose deviation as the target pose deviation when the target matching result indicates that the candidate pose deviation matches the contact force data.
[0097] For the description of the features in the corresponding embodiments of the server component assembly device, reference can be made to the relevant descriptions in the corresponding embodiments of the server component assembly method, which will not be elaborated here one by one.
[0098] An embodiment of the present application further provides an electronic device, such as Figure 6 as shown, including a memory 10 and a processor 20. A computer program is stored in the memory 10, and the processor 20 is configured to run the computer program to execute the steps in any of the above-mentioned embodiments of the server component assembly method.
[0099] An embodiment of the present application further provides a computer-readable storage medium, in which a computer program is stored. The computer program is configured to execute the steps in any of the above-mentioned embodiments of the server component assembly method when running.
[0100] In an exemplary embodiment, the above-mentioned computer-readable storage medium may include, but is not limited to: USB flash drives, read-only memories (ROM for short), random access memories (RAM for short), mobile hard disks, magnetic disks, or optical discs and other various media that can store computer programs.
[0101] An embodiment of the present application further provides a computer program product. The above-mentioned computer program product includes a computer program, and when the computer program is executed by a processor, the steps in any of the above-mentioned embodiments of the server component assembly method are implemented.
[0102] An embodiment of the present application further provides another computer program product, including a non-volatile computer-readable storage medium. The non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in any of the above-mentioned embodiments of the server component assembly method are implemented.
[0103] Those skilled in the art may further realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described according to functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered as exceeding the scope of this application.
[0104] The above has introduced in detail a server component assembly method, device, electronic device, and storage medium provided by this application. Specific examples have been used herein to elaborate on the principles and implementation manners of this application. The description of the above embodiments is only used to help understand the method and its core idea of this application. It should be noted that for those of ordinary skill in the art, without departing from the principle of this application, several improvements and modifications can be made to this application, and these improvements and modifications also fall within the protection scope of the claims of this application.
Claims
1. A server component assembly method, characterized in that, Including: Obtain the contact force data when the end of the robotic arm contacts the server component; Input the contact force data into a pre-trained pose prediction model to obtain the pose deviation of the server component relative to the component assembly position; Perform contact force matching on the pose deviation. If the matching result indicates that the pose deviation does not match the contact force data, conduct a hierarchical search starting from the pose deviation to determine the target pose deviation of the server component relative to the component assembly position; Adjust the motion trajectory of the robotic arm using the target pose deviation so that the robotic arm assembles the server component according to the adjusted motion trajectory.
2. The server component assembly method according to claim 1, wherein, Also including: Obtain the first surface data of the server component and the second surface data of the component assembly position; Use a preset space partitioning method to perform spherical hierarchical partitioning on the first surface data and the second surface data, generating a first spherical hierarchical structure corresponding to the first surface data and a second spherical hierarchical structure corresponding to the second surface data.
3. The server component assembly method according to claim 2, wherein, Using a preset space partitioning method to perform spherical hierarchical partitioning on the surface data to generate a spherical hierarchical structure corresponding to the surface data, including: Obtain the geometric center of the surface data and determine the geometric center as the root node center; Determine the data point farthest from the root node center in the surface data as the first child node, and determine the data point farthest from the first child node as the second child node; Allocate the remaining data points except the first child node and the second child node to the first cluster corresponding to the first child node and the second cluster corresponding to the second child node according to the principle of the shortest distance; Recursively partition the first cluster and the second cluster until a preset stop condition is met, generating the spherical hierarchical structure corresponding to the surface data; Wherein, the surface data is the first surface data, the spherical hierarchical structure is the first spherical hierarchical structure corresponding to the first surface data, or the surface data is the second surface data, and the spherical hierarchical structure is the second spherical hierarchical structure corresponding to the second surface data.
4. The server component assembly method according to claim 2 or 3, characterized in that, The performing contact force matching on the pose deviation includes: Obtain the initial relative position between the server component and the component assembly position; Use the product of the initial relative position and the pose deviation to determine the target contact pose between the server component and the component assembly position; Based on the positional relationship between the first spherical hierarchical structure and the second spherical hierarchical structure, determine the pair of collision spheres corresponding to the server component and the component assembly position at the target contact pose; Use the pair of collision spheres to perform contact force matching on the pose deviation to obtain a matching result.
5. The server component assembly method according to claim 4, wherein, The using the pair of collision spheres to perform contact force matching on the pose deviation to obtain a matching result includes: Obtain the penetration depth of the pair of collision spheres, and use the penetration depth and a preset contact stiffness coefficient to determine the target contact force between the server component and the component assembly position; If the contact force distance between the target contact force and the contact force data is less than a preset threshold, it is determined that the pose deviation matches the contact force data.
6. The server component assembly method according to claim 1, wherein, The pose prediction model includes a feature extraction layer and a deviation prediction layer; the feature extraction layer is cascaded with the deviation prediction layer; The step of inputting the contact force data into a pre-trained pose prediction model to obtain the pose deviation of the server component relative to the component assembly position includes: Extracting the contact force direction feature between the server component and the component assembly position from the contact force data through the feature extraction layer of the pose prediction model; Inputting the contact force data and the contact force direction feature into the deviation prediction layer of the pose prediction model to obtain the pose deviation of the server component relative to the component assembly position.
7. The server component assembly method according to claim 1, wherein The step of performing hierarchical search starting from the pose deviation to determine the target pose deviation of the server component relative to the component assembly position includes: Taking the pose deviation as the center point, expanding the search space in the order of increasing levels, and sampling to generate candidate pose deviations at a preset step size within each level range; Performing contact force matching on the candidate pose deviations to obtain the target matching result corresponding to the candidate pose deviations; When the target matching result indicates that the candidate pose deviation matches the contact force data, determining the candidate pose deviation as the target pose deviation.
8. A server component assembly device, characterized in that, including: A first acquisition module, configured to acquire the contact force data when the end of the robotic arm contacts the server component; A first determination module, configured to input the contact force data into a pre-trained pose prediction model to obtain the pose deviation of the server component relative to the component assembly position; A second determination module, configured to perform contact force matching on the pose deviation. If the matching result indicates that the pose deviation does not match the contact force data, perform hierarchical search starting from the pose deviation to determine the target pose deviation of the server component relative to the component assembly position; An assembly module, configured to adjust the motion trajectory of the robotic arm by using the target pose deviation, so that the robotic arm assembles the server component according to the adjusted motion trajectory.
9. An electronic device, characterized in that, including: A memory, configured to store a computer program; A processor, configured to implement the steps of the server component assembly method according to any one of claims 1 to 7 when executing the computer program.
10. A computer-readable storage medium, characterized in that, A computer program is stored in the computer-readable storage medium, wherein the computer program, when executed by a processor, implements the steps of the server component assembly method according to any one of claims 1 to 7.
Citation Information
Patent Citations
Mechanical arm assembly control method and device, mechanical arm control equipment and storage medium
CN114310912A
Assembly method and system for installing workpiece
CN118288014A
Control method and device of assembly robot
CN119458327A
Workpiece posture alignment method and device, electronic equipment and storage medium
CN120190610A
Robot system, control device, and control method
JP2021192944A
Cited By
Lens assembly method, device and equipment and storage medium
CN122194410A
Lens assembly method, device, apparatus, and storage medium
CN122194410B
