A test tube thread assembly simulation method and system for robot body operation data construction

Abstract

The application discloses a test tube thread assembly simulation method and system for robot body operation data construction, aiming to reduce the calculation overhead of traditional discrete grid collision and trajectory shock phenomenon. The thread center line is parameterized as a three-dimensional cylindrical spiral curve, and a continuous symbolic distance field is constructed in combination with a tubular radius, and then is bound to a simulation rigid body. Thanks to the continuous differentiable characteristics of the implicit geometric field, the contact torque is accurately fed back by calculating the overlapping area and the spatial gradient; in view of the natural fit of the distance field to the spiral topological structure, the system does not need to configure explicit spiral pair joint constraints, and can naturally drive the relative motion and friction self-locking. In addition, a multi-dimensional state machine cooperates with a mask control, automatically jumps to an assembly stage according to real-time collision distance, torque threshold and axial displacement, and supports parallel environment initialization and trajectory recording. In summary, the scheme replaces discrete grid operation with continuous field analysis, realizes non-penetration high-fidelity simulation, and improves the efficiency of body assembly data construction.

Description

Technical Field

[0001] This invention relates to the field of robot simulation and embodied intelligent data acquisition technology, and in particular to a simulation method and system for test tube thread assembly oriented towards the construction of embodied robot operation data. Background Technology

[0002] Threaded assembly is widely used in chemical experimental equipment operation and industrial manufacturing scenarios. This process essentially involves complex geometric contact, friction, and motion coupling problems. In existing robot simulation data acquisition systems, contact simulation is typically performed using geometric modeling and a physics engine. The mainstream technical solutions mainly cover the following three paths: First, explicit thread modeling based on high-precision triangular meshes, with collision detection between meshes performed by a physics engine; second, approximate simulation of thread behavior based on a reward function, without explicitly constructing the thread geometry, but guiding the virtual object to screw in along a preset direction through algorithmic rewards; third, using a multi-threaded architecture to independently run simulation instances for data acquisition.

[0003] Given the inherent limitations of the existing technical solutions in practical applications, the field has long faced the following technical bottlenecks: First, due to the high-frequency geometric details of threaded structures, explicit modeling using high-precision triangular meshes significantly increases the computational complexity and resource overhead of collision detection. Furthermore, complex discrete surfaces are prone to contact oscillations and geometric penetration during dynamic contact, leading to highly unstable simulation responses. Second, while approximation methods based on reward functions can reduce geometric construction costs, these methods rely solely on algorithmic scalars to guide object motion and cannot realistically constrain the relative motion relationships of the helical pairs within the underlying physics engine. Consequently, the generated assembly trajectories often lack strict physical consistency. Moreover, even though existing multi-threaded data acquisition architectures can increase concurrency, the independent operation of each thread simulation instance and the lack of a unified state scheduling mechanism make it difficult to maintain structural consistency among the acquired data, thus hindering effective control of key stages in the assembly process.

[0004] In summary, to overcome the inherent defects of existing simulation modeling such as computational redundancy, contact response instability, lack of physical constraints, and unstructured data acquisition, this invention aims to provide a computationally efficient, physically stable simulation method for test tube thread assembly with structured parallel acquisition capabilities. Summary of the Invention

[0005] This invention aims to propose a thread modeling method based on parametric spiral and sign distance function (SDF), construct an implicit thread contact modeling mechanism that does not require explicit joint constraints, and design a parallel data acquisition method based on multidimensional state machine, thereby effectively improving simulation stability, computational efficiency and data quality.

[0006] This invention provides a simulation method for test tube thread assembly based on robot embodied operation data construction, including: The thread centerline is modeled as a three-dimensional cylindrical helical curve, and the distance from the spatial point to the helix is ​​calculated based on the thread radius, pitch, and parameter variables. The thread geometry is determined using the radius and parameter set, and a signed distance function is constructed for both the nut and the bolt. Bind the symbolic distance function to the simulated rigid body; The contact overlap region and gradient of the symbolic distance function are calculated, and the contact force and torque are solved to perform helical relative motion and frictional self-locking.

[0007] In one embodiment of the present invention, the three-dimensional cylindrical helical curve is: ; in, Where is the thread radius. For pitch, For parameter variables; The formula for calculating the distance from a spatial point to the spiral is: ; in, , ; The radius of the tube is [missing information]. Let be any point in three-dimensional space from which the distance needs to be calculated, with coordinates as... , For point The initial angle parameters of the spiral corresponding to the projection. For point The shortest distance to the center line of the thread helix For the point of departure The most recent spiral coil number sequence, This is a four-quadrant arctangent function; the input is the ordinate and abscissa of a point, and the output is the polar angle of the plane. The coordinates of the matching point on the spiral line that matches the distance point P; The thread geometry profile is a set of parameters , This represents the total axial length of the thread along the Z-axis. This represents the height of the thread.

[0008] In one embodiment of the present invention, the method further includes: constructing a multi-dimensional state machine to control the assembly process, wherein the multi-dimensional state machine defines a state vector, and the state vector is... This includes the approach state, alignment state, contact state, screw-in state, and end state; The steps for constructing a multi-dimensional state machine to control the assembly process include: The contact state is determined based on the real-time collision distance. The determination of entering the screw-in state is based on the contact torque threshold. When the axial displacement reaches the preset stroke, it is determined to enter the end state.

[0009] In one embodiment of the present invention, the method further includes: configuring a masking mechanism, wherein the masking mechanism selectively masks or activates the input channel and action space of the parallel instance through bit operations.

[0010] In one embodiment of the present invention, the mask formula is: .

[0011] In one embodiment of the present invention, the masking mechanism triggers a state transition when a preset geometric or mechanical threshold is met, and simultaneously releases the pose constraints of the next stage to lock the action space of the previous stage. The formula for the target pose is: ; The state transition formula is: ; in This is the difference between the current pose of the robotic arm and the target pose. This is the difference between the current contact force between the test tube cap and the test tube itself and the target contact force. The pose difference threshold, The contact force difference threshold, This is an element-wise multiplication operation. For the k-th independent subtask, the subtarget pose is... is the weighted weight coefficient corresponding to the k-th sub-objective.

[0012] In one embodiment of the present invention, the method further includes: constructing a vectorized parallel environment in a simulation platform, wherein each parallel environment independently and randomly initializes the initial pose of the assembly, and calls a unified inverse kinematics solver to drive the end effector. Automatically record state vectors and corresponding action command sequences, and generate a structured assembly trajectory dataset based on the recorded data.

[0013] In one embodiment of the present invention, the symbolic distance function is constructed using an explicit analytical expression or a neural network fitting function.

[0014] This invention also provides a simulation system for test tube thread assembly based on robot embodied operation data construction, comprising: The thread geometry modeling module is used to model the thread centerline as a three-dimensional cylindrical helical curve, calculate the distance from a spatial point to the helix, and introduce a tubular radius and parameter set to uniquely determine the thread geometry profile. The symbolic distance function construction module is used to construct symbolic distance functions for nuts and bolts respectively; A rigid body binding module is used to bind the signed distance function to a simulated rigid body; The contact mechanics solution module is used to calculate the contact overlap region and gradient of the symbolic distance function, solve the contact force and torque, realize the relative motion of the helical joints and frictional self-locking, and the system is not configured with explicit helical joint constraints.

[0015] In one embodiment of the present invention, the thread geometry modeling module includes a distance mapping unit for performing parametric calculations based on the Euclidean distance mapping between spatial point coordinates and the helix centerline.

[0016] In one embodiment of the present invention, a multi-dimensional state machine control module is provided, the multi-dimensional state machine control module is configured with a state vector register, the state vector register sequentially stores the approach state, alignment state, contact state, screw-in state and end state. The multidimensional state machine control module includes: The distance determination unit is used to determine whether the contact state has been entered based on the real-time collision distance; Torque threshold unit, used to determine entry into the screw-in state based on the contact torque threshold; The stroke detection unit is used to determine the end state when the axial displacement reaches the preset stroke. A mask control module is used to selectively mask or activate the input channels and action space of parallel instances through bit operations.

[0017] The present invention has the following beneficial effects: (1) This invention parameterizes the thread centerline as a three-dimensional cylindrical helical curve and introduces a tubular radius to construct a continuous implicit geometric field. Thanks to the good continuity characteristics of SDF, the distance calculation from a spatial point to the object surface is simplified, thereby avoiding complex mesh collision calculations between discrete surfaces. When the simulated entity enters the contact range, the analytical calculation based on the continuous distance field can accurately feed back the contact force and position deviation. Therefore, the generated assembly trajectory data has no obvious penetration or oscillation phenomenon, and the computational cost is greatly reduced.

[0018] (2) This invention binds the corresponding SDF fields to the rigid bodies of the nut and bolt respectively, and directly performs SDF contact calculations in the physics engine. Since the implicit geometric field can accurately characterize the spatial interference and normal distribution between the thread helical surfaces, the relative displacement and frictional self-locking behavior that conform to the laws of helical kinematics are naturally derived through contact calculation. Therefore, even without preset explicit helical pair constraints, the virtual component can strictly follow the laws of physical contact mechanics to complete the screwing action, and the resulting trajectory has strict physical consistency.

[0019] (3) This invention defines state vectors that include stages such as approach, alignment, contact, spin-in, and termination, and configures corresponding target poses and state transition conditions. In a vectorized parallel simulation environment, each instance is randomly initialized and driven by a unified inverse kinematics (IK) solver, and a mask mechanism is used to uniformly issue instructions and mask states in the parallel environment. As a result, each simulation thread can synchronously follow the preset state machine transition logic, which not only ensures the synchronous execution efficiency of parallel acquisition, but also the automatically recorded action and state data have highly structured features, which is convenient for subsequent training of embodied intelligent models. Attached Figure Description

[0020] Figure 1 A flowchart of a simulation method for test tube thread assembly based on robot embodied operation data is shown in one embodiment of the present invention; Figure 2 A schematic diagram of a threaded assembly cross section based on a sign distance function is shown in one embodiment of the present invention; Figure 3 This diagram illustrates the effect of thread assembly based on the sign distance function in a simulator according to an embodiment of the present invention. Figure 4 The diagram illustrates a data acquisition scenario for a test tube assembly task based on a multi-dimensional state machine, according to an embodiment of the present invention. Detailed Implementation

[0021] In the following description, the invention is described with reference to various embodiments. However, those skilled in the art will recognize that the embodiments may be practiced without one or more specific details or with other alternatives and / or additional methods, materials, or components. In other instances, well-known structures, materials, or operations are not shown or described in detail so as not to obscure the inventive points of the invention. Similarly, for illustrative purposes, specific quantities, materials, and configurations are set forth to provide a comprehensive understanding of embodiments of the invention. However, the invention is not limited to these specific details.

[0022] In this invention, the various embodiments are merely intended to illustrate the solutions of the invention and should not be construed as limiting.

[0023] In this specification, references to "an embodiment" or "this embodiment" mean that a particular feature, structure, or characteristic described in connection with that embodiment is included in at least one embodiment of the invention. The phrase "in one embodiment" appearing throughout this specification does not necessarily refer to the same embodiment in all instances.

[0024] Furthermore, the numbering of the steps in the methods of the present invention does not limit the execution order of the method steps. Unless otherwise specified, the method steps may be executed in different orders.

[0025] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0026] Figure 1 A flowchart of a simulation method for test tube thread assembly based on robot embodied operation data is shown in one embodiment of the present invention.

[0027] like Figure 1 As shown, in this embodiment, the simulation method for test tube thread assembly based on robot embodied operation data includes: S1. Parametric modeling of thread geometry.

[0028] The centerline of the thread to be assembled is abstracted as a three-dimensional cylindrical helical curve. Given the high-frequency spatial detail features of the thread structure, explicit triangular mesh discretization would drastically increase the complexity of collision detection. Therefore, this invention defines the three-dimensional cylindrical helical curve as follows: ; in, Where is the thread radius. For pitch, For parameter variables.

[0029] A continuously differentiable parametric equation for a spiral curve is constructed. This parametric expression not only avoids the computational redundancy caused by massive surface patches, but also provides an accurate mathematical benchmark for subsequent spatial distance mapping.

[0030] S2, Distance Mapping and Contour Determination.

[0031] like Figure 2 As shown, for any point P to be calculated in three-dimensional space, the system performs parametric calculations based on the Euclidean distance between its spatial coordinates and the helical centerline. Specifically, a tubular radius r2 and a parameter set {L, H} (where L is the thread length and H is the thread height) are introduced to uniquely determine the thread geometry. During this process, the shortest distance d from point P to the helical centerline is calculated, and combined with the index n of the nearest helical coil to point P, the precise distance field value is obtained through analytical relationships. Thanks to this distance mapping mechanism, the system can accurately capture the spatial topological distribution of the thread profile with a low sampling density, thus laying the data foundation for implicit geometry construction.

[0032] The formula for calculating the distance from a spatial point to the spiral is: ; in, , ; The radius of the tube is [missing information]. Let be any point in three-dimensional space from which the distance needs to be calculated, with coordinates as... , For point The initial angle parameters of the spiral corresponding to the projection. For point The shortest distance to the center line of the thread helix For the point of departure The most recent spiral coil number sequence, This is a four-quadrant arctangent function; the input is the ordinate and abscissa of a point, and the output is the polar angle of the plane. The coordinates of the matching point on the spiral line that matches the distance point P; The thread geometry profile is a set of parameters , This represents the total axial length of the thread along the Z-axis. This represents the height of the thread.

[0033] Based on this, symbolic distance functions are constructed for nuts and bolts respectively. Since the symbolic distance field possesses continuous implicit characteristics, its zero isosurface naturally corresponds to the physical contact surface; therefore, the constructed symbolic distance functions are directly bound to the corresponding simulated rigid body objects. Preferably, the symbolic distance functions are constructed using explicit analytical expressions to ensure the numerical stability and solution speed of gradient calculations; alternatively, when the thread morphology exhibits localized wear or non-standard features, a neural network fitting function can be used to approximate the complex distance distribution. Figure 3 As shown, through rigid body binding operations, the implicit distance field is fixed in the motion coordinate system, enabling seamless integration of geometric representation and dynamic solution.

[0034] S3, Solution of contact mechanics.

[0035] When the assembly enters the contact interaction stage, the contact mechanics solution step is executed. The system calculates the contact overlap region and its spatial gradient of the two sign distance functions in real time, and solves for the contact force and contact torque based on the gradient direction and overlap depth. Since the system does not configure explicit helical joint constraints, the helical relative motion and frictional self-locking phenomenon between the nut and bolt are entirely driven by the contact mechanics response of the sign distance field. Once the contact torque exceeds the critical threshold or the axial displacement meets the mechanical equilibrium condition, the rigid body will autonomously generate a screw-in action along the helical trajectory, thus achieving a high-fidelity physical assembly process without the need for manual intervention in path planning.

[0036] S4, Multidimensional state machine control logic.

[0037] like Figure 4 As shown, to achieve structured management of the entire assembly lifecycle, this method further introduces multi-dimensional state machine control logic. The system defines a state vector, which is... This includes the approach state, alignment state, contact state, screw-in state, and end state.

[0038] During operation, the distance determination unit continuously monitors the real-time collision distance. When the collision distance is less than a preset safety threshold, the system determines that it has entered the contact state. Subsequently, the torque threshold unit intervenes to calculate and determine whether to trigger the screw-in state based on the contact torque threshold. If the axial displacement reaches the preset stroke, the stroke detection unit determines that the assembly is complete and jumps to the end state. This state machine mechanism connects the various stages of assembly through clear physical determination conditions, effectively avoiding disordered exploration of the action space and strategy drift.

[0039] During the parallel simulation execution phase, a masking mechanism is configured to selectively mask or activate the input channels and action spaces of multiple instances. Preferably, the masking mechanism is executed through bitwise operation logic, triggering a state transition when a preset geometric or mechanical threshold is met, and simultaneously releasing the pose constraints of the next stage to lock the action space of the previous stage. The target pose and state transition process follow a predefined mapping relationship to ensure that the action spaces of each parallel instance are strictly decoupled during state transitions. The masking formula is: .

[0040] The masking mechanism triggers a state transition when a preset geometric or mechanical threshold is met, and simultaneously releases the pose constraints of the next stage to lock the action space of the previous stage. The formula for the target pose is: ; The state transition formula is: ; in This is the difference between the current pose of the robotic arm and the target pose. This is the difference between the current contact force between the test tube cap and the test tube itself and the target contact force. The pose difference threshold, The contact force difference threshold, This is an element-wise multiplication operation. For the k-th independent subtask, the subtarget pose is... is the weighted weight coefficient corresponding to the k-th sub-objective.

[0041] Furthermore, the system constructs a vectorized parallel environment, where each environment independently and randomly initializes the initial pose of the assembly and calls a unified inverse kinematics solver to drive the end effector. To improve data reuse, the system automatically records each state vector and its corresponding action command sequence, generating a structured assembly trajectory dataset based on the recorded data. In summary, this embodiment achieves a computationally efficient, physically stable, and structurally consistent thread assembly simulation and data acquisition process through the synergistic effect of parametric spiral modeling, implicit distance field contact solving, and parallel state mask control.

[0042] In another embodiment of the invention, given the high-frequency detailed features of the thread geometry, explicit triangular mesh discretization would lead to an exponential increase in collision detection complexity. Therefore, the invention employs a parametric implicit representation strategy to balance physical realism and computational efficiency. Specifically, the system first calls the thread geometry modeling module to model the thread centerline as a three-dimensional cylindrical helical curve. To accurately describe the spatial geometric relationships, this module introduces distance mapping units, performing parametric calculations based on the Euclidean distance mapping between spatial point coordinates and the helical centerline.

[0043] Since the distance calculation from a spatial point to the helix directly determines the accuracy of subsequent contact solutions, the distance mapping unit further introduces a tubular radius and parameter set to uniquely determine the thread geometry. When any spatial point is obtained, its shortest distance to the helix centerline is calculated. Thanks to the above parametric modeling mechanism, the system not only avoids the discretization error of a massive number of triangular facets, but also can flexibly adapt to test tubes and test tube racks of different specifications by adjusting the length and height parameters.

[0044] After completing the geometric contour construction, the symbolic distance function construction module constructs symbolic distance functions for the nut and bolt respectively. Since the symbolic distance field possesses continuous differentiability, its zero isosurface naturally corresponds to the physical surface of the object. Therefore, binding the aforementioned analytical distance field to the simulated rigid body can directly replace traditional discrete collision geometry. Next, the rigid body binding module binds the constructed symbolic distance functions to the corresponding simulated rigid body objects. When the assembly process enters the contact phase, the contact mechanics solution module calculates the contact overlap region and its spatial gradient of the symbolic distance function in real time. Through the gradient direction and overlap depth, the system can accurately solve for contact force and contact torque. Because the symbolic distance field implicitly includes the topological continuity of the helical geometry, the system can still naturally drive the nut and bolt to produce helical relative motion and frictional self-locking phenomena without configuring explicit helical joint constraints when solving for the contact mechanics response. In summary, this embodiment achieves high-fidelity threaded contact modeling without explicit constraints through the deep fusion of parameterized helices and symbolic distance functions, significantly improving the computational efficiency and physical stability of the simulation system.

[0045] Building upon this embodiment, a multi-dimensional state machine control module and a parallel acquisition mechanism are further integrated to address the technical shortcomings of existing data acquisition methods, such as lack of structured scheduling and poor data consistency. Given the distinct stage characteristics of the thread assembly process, a single continuous control strategy is highly susceptible to motion space redundancy or state drift. Therefore, the system introduces a multi-dimensional state machine to provide structured control over the assembly process. Specifically, the multi-dimensional state machine control module is equipped with a state vector register, which sequentially stores the approach state, alignment state, contact state, screw-in state, and end state. After the simulation environment is initialized, the distance determination unit continuously monitors the real-time collision distance. Once the collision distance is less than a preset safety threshold, the system immediately determines that it has entered the contact state. Subsequently, the torque threshold unit intervenes, determining whether to enter the screw-in state based on the contact torque threshold. Since the screw-in process is accompanied by significant axial and angular coupled motion, the stroke detection unit synchronously tracks the axial displacement. When the axial displacement reaches a preset stroke, the system determines that the assembly task is complete and jumps to the end state.

[0046] To achieve unified scheduling of multiple instances in a parallel environment, the masking control module selectively masks or activates the input channels and action spaces of parallel instances through bit operations. Preferably, the masking mechanism triggers a state transition when a preset geometric or mechanical threshold is met, and simultaneously releases the pose constraints of the next stage to lock the action space of the previous stage.

[0047] When each parallel instance runs independently, the simulation platform constructs a vectorized parallel environment at the underlying level. Each parallel environment independently and randomly initializes the initial pose of the assembly and calls a unified inverse kinematics solver to drive the end effector. To improve data reuse, the system automatically records the state vector and the corresponding action instruction sequence, generating a structured assembly trajectory dataset based on the recorded data. Furthermore, the bit manipulation logic of the mask formula ensures strict decoupling of the action spaces of each instance during state transitions, thereby avoiding policy interference between parallel threads. Through the synergistic effect of state-driven and mask-based approaches, the system not only achieves standardized control of the assembly process but also significantly improves the throughput and sample quality of large-scale data acquisition.

[0048] This embodiment provides optimized and alternative solutions for the implementation of the symbolic distance function and the state scheduling architecture to enhance the system's adaptability under complex working conditions. Although the above embodiment uses analytical expressions to construct the symbolic distance field, explicit analytical expressions may struggle to cover local topological distortions when dealing with non-standard threads or worn geometry. Alternatively, the symbolic distance function construction module can use a neural network fitting function to construct the distance field. By inputting a large amount of point cloud sampling data into the neural network for regression training, the system can approximate the implicit distance distribution of any complex thread. Once the network converges to a preset error range, the fitting function can seamlessly replace the analytical expression, continue to be bound to the rigid body, and participate in the contact mechanics solution.

[0049] Furthermore, for the state machine control architecture, behavior trees can optionally be used instead of traditional finite state machines. When there are multi-objective conflicts or external disturbances in the assembly environment, the behavior tree can dynamically reorganize the control flow based on priority and subtree execution results. Even if execution fails in a certain branch, the behavior tree can still quickly backtrack to the previous stable node, thereby ensuring the robustness of the assembly process. If the thread geometry itself has variable pitch or non-cylindrical features, the thread geometry modeling module only needs to modify the parameterized definition of the helical curve, and the distance mapping unit can still perform Euclidean distance mapping according to the same logic. In summary, this embodiment further expands the applicability of the simulation method through a flexible and replaceable function construction method and scheduling architecture, while maintaining the integrity of the core implicit contact mechanism.

[0050] Although various embodiments of the invention have been described above, it should be understood that they are presented by way of example only and not as limitations. It will be apparent to those skilled in the art that various combinations, modifications, and alterations can be made without departing from the spirit and scope of the invention. Therefore, the breadth and scope of the invention disclosed herein should not be limited by the exemplary embodiments disclosed above, but should be defined solely by the appended claims and their equivalents.

Claims

1. A simulation method for test tube thread assembly based on robot embodied operation data construction, characterized in that, include: The thread centerline is modeled as a three-dimensional cylindrical helical curve, and the distance from the spatial point to the helix is ​​calculated based on the thread radius, pitch, and parameter variables. The thread geometry is determined using the radius and parameter set, and a signed distance function is constructed for both the nut and the bolt. Bind the symbolic distance function to the simulated rigid body; Calculate the contact overlap region and gradient of the symbolic distance function, and solve for the contact force and torque to perform helical relative motion and frictional self-locking; A multi-dimensional state machine is constructed to control the assembly process. The multi-dimensional state machine defines a state vector, which is... This includes the approach state, alignment state, contact state, screw-in state, and end state; The steps for constructing a multi-dimensional state machine to control the assembly process include: The contact state is determined based on the real-time collision distance. The determination of entering the screw-in state is based on the contact torque threshold. When the axial displacement reaches the preset stroke, the system is determined to enter the end state. Configure a masking mechanism that selectively masks or activates the input channels and action space of parallel instances through bit operations; The three-dimensional cylindrical helical curve is as follows: ； in, Where is the thread radius. For pitch, For parameter variables; The formula for calculating the distance from a spatial point to the spiral is: ； in, , ; The radius of the tube is [missing information]. Let be any point in three-dimensional space from which the distance needs to be calculated, with coordinates as... , For point The initial angle parameters of the spiral corresponding to the projection. For point The shortest distance to the center line of the thread helix For the point of departure The most recent spiral coil number sequence, This is a four-quadrant arctangent function; the input is the ordinate and abscissa of a point, and the output is the polar angle of the plane. The coordinates of the matching point on the spiral line that matches the distance point P; The thread geometry profile is a set of parameters , This represents the total axial length of the thread along the Z-axis. This represents the height of the thread.

2. The method according to claim 1, characterized in that, The mask formula is: 。 3. The method according to claim 1, characterized in that, The masking mechanism triggers a state transition when a preset geometric or mechanical threshold is met, and simultaneously releases the pose constraints of the next stage to lock the action space of the previous stage. The formula for the target pose is: ； The state transition formula is: ； in This is the difference between the current pose of the robotic arm and the target pose. This is the difference between the current contact force between the test tube cap and the test tube itself and the target contact force. The pose difference threshold, The contact force difference threshold, This is an element-wise multiplication operation. For the k-th independent subtask, the subtarget pose is... is the weighted weight coefficient corresponding to the k-th sub-objective.

4. The method according to claim 1, characterized in that, Also includes: A vectorized parallel environment is constructed in the simulation platform. Each parallel environment independently and randomly initializes the initial pose of the assembly and calls a unified inverse kinematics solver to drive the end effector. Automatically record state vectors and corresponding action command sequences, and generate a structured assembly trajectory dataset based on the recorded data.

5. The method according to claim 1, characterized in that, The symbolic distance function is constructed using an explicit analytical expression or a neural network fitting function.

6. A simulation system for test tube thread assembly based on robot embodied operation data construction, characterized in that, include: The thread geometry modeling module is used to model the thread centerline as a three-dimensional cylindrical helical curve, calculate the distance from a spatial point to the helix, and introduce a tubular radius and parameter set to uniquely determine the thread geometry profile. The symbolic distance function construction module is used to construct symbolic distance functions for nuts and bolts respectively; A rigid body binding module is used to bind the signed distance function to a simulated rigid body; The contact mechanics solution module is used to calculate the contact overlap region and gradient of the sign distance function, solve the contact force and torque, realize the relative motion of the helical joints and frictional self-locking, and the system is not configured with explicit helical joint constraints. A multi-dimensional state machine control assembly module is provided, which is equipped with a state vector register, which sequentially stores the approach state, alignment state, contact state, screw-in state and end state. The multi-dimensional state machine control module is used to control the assembly process using a multi-dimensional state machine. The multi-dimensional state machine defines a state vector, which is... This includes the approach state, alignment state, contact state, screw-in state, and end state; The steps for constructing a multi-dimensional state machine to control the assembly process include: The contact state is determined based on the real-time collision distance. The determination of entering the screw-in state is based on the contact torque threshold. When the axial displacement reaches the preset stroke, the system is determined to enter the end state. Configure the mask module to selectively mask or activate the input channels and action space of parallel instances through bit operations; The three-dimensional cylindrical helical curve is as follows: ； in, Where is the thread radius. For pitch, For parameter variables; The formula for calculating the distance from a spatial point to the spiral is: ； in, , ; The radius of the tube is [missing information]. Let be any point in three-dimensional space from which the distance needs to be calculated, with coordinates as... , For point The initial angle parameters of the spiral corresponding to the projection. For point The shortest distance to the center line of the thread helix For the point of departure The most recent spiral coil number sequence, This is a four-quadrant arctangent function; the input is the ordinate and abscissa of a point, and the output is the polar angle of the plane. The coordinates of the matching point on the spiral line that matches the distance point P; The thread geometry profile is a set of parameters , This represents the total axial length of the thread along the Z-axis. This represents the height of the thread.

7. The system according to claim 6, characterized in that, The multidimensional state machine control module includes: The distance determination unit is used to determine whether the contact state has been entered based on the real-time collision distance; Torque threshold unit, used to determine entry into the screw-in state based on the contact torque threshold; The stroke detection unit is used to determine the end state when the axial displacement reaches the preset stroke. A mask control module is used to selectively mask or activate the input channels and action space of parallel instances through bit operations. The mask configuration module is used to selectively mask or activate the input channels and action space of parallel instances through bit operations.

8. The system according to claim 6, characterized in that, The thread geometry modeling module includes a distance mapping unit, which is used to perform parametric calculations based on the Euclidean distance mapping between spatial point coordinates and the helix centerline.

Images