Grabbing task generation method, device and storage medium

Abstract

The application discloses a grasping task generation method and device and a storage medium, relates to the technical field of robot simulation grasping, and comprises the following steps: constructing a monotone scanning sequence for an object set under a preset same geometric condition based on a predetermined target physical parameter; performing simulation grasping on each sample in the monotone scanning sequence under a preset unified simulation grasping condition, and obtaining a simulation result data set of the simulation grasping; performing contact state judgment and boundary identification on the monotone scanning sequence based on the simulation result data set, generating an executable task set; and constructing a continuous difficulty task path for the executable task set, and obtaining a grasping task set. The application constructs a monotone scanning sequence, combines a unified simulation condition, contact state transition detection and boundary identification, and continuous difficulty task path construction, so that a grasping task set with stronger physical interpretability is generated around a parameter region of a grasping instability boundary.

Description

Technical Field

[0001] This application relates to the field of robot simulation grasping technology, and in particular to grasping task generation methods, equipment and storage media. Background Technology

[0002] Ideally, robot grasping task samples should cover the complete transition range from stable grasping to grasping failure. However, common grasping task generation methods often employ random perturbation or large-scale uniform sampling, resulting in a scattered sample distribution. This makes it difficult to effectively focus on the critical parameter region that is truly close to the grasping instability boundary. Furthermore, since common grasping task generation methods only output binary labels for grasping success or failure, or a single quality score, they lack a characterization of the contact physical mechanism (such as the state transition from static contact to sliding) that causes grasping to evolve from an executable state. Therefore, they cannot accurately identify the parameter executable boundary between "first occurrence of sliding" and "first grasping failure," and the generated grasping tasks are difficult to cover the complete transition range from stable grasping to grasping failure.

[0003] Therefore, how to generate a set of grasping tasks with stronger physical interpretability around the parameter region of the grasping instability boundary has become a technical problem that this application urgently needs to solve. Summary of the Invention

[0004] The main objective of this application is to provide a method, device, and storage medium for generating grasping tasks, aiming to solve the technical problem of how to generate a set of grasping tasks with stronger physical interpretability around the parameter region of the grasping instability boundary.

[0005] To achieve the above objectives, this application proposes a method for generating crawling tasks, the method comprising: A monotonic scanning sequence is constructed based on a set of objects with predetermined target physical parameters and the same geometric conditions. Under preset uniform simulation capture conditions, simulation capture is performed on each sample in the monotonic scanning sequence, and the simulation result dataset of the simulation capture is obtained. Based on the simulation result dataset, the contact state determination and boundary identification are performed on the monotonic scanning sequence to generate a set of executable tasks. A set of tasks with continuously increasing difficulty is constructed for the set of executable tasks to obtain a set of crawling tasks.

[0006] In one embodiment, the step of constructing a monotonic scan sequence for a set of objects under preset geometric conditions based on predetermined target physical parameters further includes: Select a reference geometric model and construct a set of candidate objects that have the same geometric shape as the reference geometric model; Determine whether each candidate sub-object in the candidate object set maintains the same geometric conditions as the baseline geometric model; If the candidate sub-object maintains the same geometric conditions as the baseline geometric model, then an object set is constructed based on the candidate sub-object.

[0007] In one embodiment, the step of constructing a monotonic scan sequence based on a set of objects with the same preset geometric conditions and predetermined target physical parameters includes: The target physical parameter is selected from multiple candidate physical parameters based on the parameter selection command sent by the user terminal. The scanning step size of the target physical parameters is determined according to preset constraints, and a monotonic scanning direction is set. Based on the scanning step size and the scanning direction, different target physical parameter values ​​are sequentially assigned to the candidate sub-objects in the object set to obtain a monotonic scanning sequence.

[0008] In one embodiment, the simulation result dataset includes contact timing data and grasping results. The step of performing contact state determination and boundary identification on the monotonic scanning sequence based on the simulation result dataset to generate an executable task set includes: Based on the contact timing data, the contact state of each sample in the monotonic scan sequence is determined to obtain the transition marker corresponding to the sample. Based on the transition markers and the capture results, boundary identification is performed on the monotonic scan sequence to obtain the first transition sample index and the first failure sample index. An executable task set is generated based on the first transition sample index and the first failure sample index.

[0009] In one embodiment, the step of determining the contact state of each sample in the monotonic scan sequence based on the contact timing data to obtain the transition marker corresponding to the sample includes: Obtain the normal contact force, tangential contact force, and relative tangential velocity of the sample at each simulation moment in the contact timing data; The maximum static friction force at the current simulation moment is calculated based on the normal contact force and the preset friction coefficient, and the preset force threshold is subtracted from the maximum static friction force to obtain the sliding judgment force threshold. Under the condition that the relative tangential velocity of the contact is not greater than a preset velocity threshold and the tangential contact force is less than the sliding determination force threshold, it is determined that the sample is in a static contact state at the current simulation moment; Under the condition that the relative tangential velocity of the contact is greater than a preset velocity threshold and the tangential contact force is not less than the sliding determination force threshold, it is determined that the sample is in a sliding state at the current simulation moment; Track whether the sample exhibits a state transition from the static contact state to the sliding state along the time axis of the contact timing data; If the state transition occurs, the transition of the sample is marked as a true value.

[0010] In one embodiment, the step of performing boundary identification on the monotonic scan sequence based on the transition marker and the capture result to obtain the first transition sample index and the first failure sample index includes: Obtain the transition marker and capture result corresponding to each sample in the monotonic scan sequence; The monotonic scan sequence is traversed along the monotonic scan direction, and the smallest index in which the transition is marked as a true value is recorded as the first transition sample index. The smallest index of the crawling result that is false is denoted as the first failed sample index.

[0011] In one embodiment, the step of generating a set of executable tasks based on the first transition sample index and the first failure sample index includes: If the index of the first transition sample is less than the index of the first failure sample, the index of the last successful sample is determined based on the index of the first failure sample. The parameter executable boundary is defined by the first transition sample index and the last successful sample index, and a set of executable tasks in the boundary neighborhood is generated based on the parameter executable boundary; The samples in the monotonic scan sequence whose index is less than the index of the first transition sample and whose capture result is true are denoted as the benchmark stable task set; The samples in the monotonic scanning sequence whose index is not less than the index of the first failure sample and whose capture result is false are denoted as the failure verification set. An executable task set is generated based on the boundary neighborhood executable task set, the benchmark stable task set, and the failure verification set.

[0012] In one embodiment, the step of constructing a series of task paths with varying difficulty for the set of executable tasks to obtain a set of crawling tasks includes: The set of executable tasks in the boundary neighborhood of the executable task set is sorted based on a preset difficulty score in order to construct a continuous difficulty task path; A set of grabbing tasks is generated based on the set of executable tasks that construct the continuous difficulty task path.

[0013] In addition, to achieve the above objectives, this application also proposes a crawling task generation device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the crawling task generation method as described above.

[0014] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the crawling task generation method described above.

[0015] One or more technical solutions proposed in this application have at least the following technical effects: First, a monotonic scanning sequence is constructed based on a pre-determined target physical parameters and a set of objects under the same geometric conditions. Simulated grasping is then performed on each sample in the monotonic scanning sequence under pre-defined unified simulation grasping conditions. By combining the construction of the monotonic scanning sequence with the execution of simulated grasping under the same conditions, interference from geometry and control strategies is eliminated, ensuring that changes in the grasping results are uniquely attributed to the monotonic changes in the target physical parameters. This provides a controllable variable basis for locating the instability boundary. Furthermore, by determining the contact state and identifying the boundary based on the simulation result dataset, the physical transition point from static contact to sliding can be extracted. The resulting "executable task set" naturally focuses on the parameter region closest to the instability boundary. Additionally, a continuously challenging task path is constructed for the executable task set, giving the tasks a clear progression of difficulty. This ensures that the generated grasping task set not only revolves around the critical region but also possesses a clear physical evolution logic. In summary, this application, by constructing a monotonic scanning sequence and combining it with unified simulation conditions, contact state transition detection and boundary identification, and the construction of continuously challenging task paths, generates a more physically interpretable grasping task set around the parameter region of the grasping instability boundary. Attached Figure Description

[0016] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating the first embodiment of the task generation method for this application. Figure 2 This is a flowchart illustrating the second embodiment of the task generation method of this application. Figure 3 This is a flowchart illustrating the fourth embodiment of the task generation method of this application. Figure 4 This is a schematic diagram of the module structure of the task generation device according to an embodiment of this application; Figure 5 This is a schematic diagram of the device structure of the hardware operating environment involved in the task generation method in this application embodiment.

[0019] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0020] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0021] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0022] The main solution of this application embodiment is as follows: construct a monotonic scanning sequence based on a set of objects with the same preset geometric conditions and predetermined target physical parameters; perform simulated grasping on each sample in the monotonic scanning sequence under preset unified simulation grasping conditions, and obtain a simulation result dataset of the simulated grasping; perform contact state determination and boundary recognition on the monotonic scanning sequence based on the simulation result dataset to generate an executable task set; construct a continuous difficulty task path for the executable task set to obtain a grasping task set.

[0023] In this embodiment, for ease of description, the following description will focus on the task generation system as the execution entity.

[0024] This application's embodiments take into account that: ideal robot grasping task samples need to cover the complete transition range from stable grasping to grasping failure. However, common grasping task generation methods often use random perturbation or large-scale uniform sampling, resulting in a scattered sample distribution, making it difficult to effectively focus on the critical parameter region that is truly close to the grasping instability boundary; furthermore, since common grasping task generation methods only output binary labels for grasping success or failure, or a single quality score, they lack a characterization of the contact physical mechanism (such as the state transition from static contact to sliding) of the grasping evolution from an executable state to an inexecutable state. Therefore, they cannot accurately identify the parameter executable boundary between "first occurrence of sliding" and "first grasping failure", and the generated grasping tasks are difficult to cover the complete transition range from stable grasping to grasping failure.

[0025] Therefore, this application provides a solution. First, a monotonic scanning sequence is constructed based on a pre-determined target physical parameters and a set of objects under the same geometric conditions. Then, simulation grasping is performed on each sample in the monotonic scanning sequence under pre-defined unified simulation grasping conditions. By combining the construction of the monotonic scanning sequence with the execution of simulation grasping under the same simulation grasping conditions, the interference of geometry and control strategies is eliminated, making the change in grasping results solely attributable to the monotonic change of the target physical parameters, thereby providing a controllable variable basis for locating the instability boundary. Furthermore, by performing contact state determination and boundary identification based on the simulation result dataset, the physical transition point from static contact to sliding can be extracted. The resulting "executable task set" naturally focuses on the parameter region closest to the instability boundary. In addition, a continuous difficulty task path is constructed for the executable task set, giving the tasks a clear difficulty progression relationship, so that the generated grasping task set not only revolves around the critical region but also has a clear physical evolution logic. In summary, this application generates a set of grasping tasks with stronger physical interpretability by constructing monotonic scanning sequences and combining them with unified simulation conditions, contact state transition detection and boundary recognition, and continuous difficulty task path construction, thereby focusing on the parameter region of the grasping instability boundary.

[0026] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device or a task generation system capable of performing the above functions. The following description uses a task generation system as an example to illustrate this embodiment and the subsequent embodiments.

[0027] Based on this, embodiments of this application provide a method for generating crawling tasks, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the task generation method of this application.

[0028] In this embodiment, the task generation method includes steps S10 to S40: Step S10: Construct a monotonic scanning sequence for a set of objects under the same preset geometric conditions based on predetermined target physical parameters; The target physical parameter refers to a physical parameter that is pre-selected from multiple candidate physical parameters that affect the stability of the grasp and is intended to undergo monotonic change. Candidate physical parameters may include the contact friction coefficient, object mass, contact stiffness, and damping parameters.

[0029] The set of objects under the same preset geometric conditions refers to a set of objects generated by a reference geometric model by keeping the geometric shape unchanged and only allowing the target physical parameters to take different values. Any two objects in this set are the same in terms of geometric shape, size, and topology or meet the preset same geometric determination conditions.

[0030] A monotonic scan sequence refers to an ordered sequence formed by arranging objects in a set of objects in a monotonically increasing or monotonically decreasing order of the target physical parameter values.

[0031] Understandably, by constructing a family of geometric objects and establishing a monotonic scanning sequence for them, the geometric factors can be completely fixed, so that the changes in the subsequent grasping simulation results can be attributed solely to the monotonic changes in the target physical parameters. This provides a pure variable factor control environment for accurately identifying the grasping instability boundary and avoids the interference of geometric disturbances on the identification of parameter boundaries.

[0032] Specifically, the system first reads a baseline geometric model, then copies the baseline geometric model to generate multiple candidate sub-objects with the exact same geometric shape as the baseline geometric model. The system receives parameter selection instructions sent by the user and determines the target physical parameters for this scan from the candidate physical parameters. The system automatically determines the scan step size according to preset constraints. The scan step size can be a constant step size, or a smaller step size can be used in certain regions of the parameter space (such as near the estimated boundary) to refine boundary identification.

[0033] Furthermore, the system sets a monotonic scanning direction. When the target physical parameter is the coefficient of friction, it preferably scans from the direction of high friction coefficient to low friction coefficient; when the target physical parameter is the mass of the object, it can choose a direction from light to heavy or from heavy to light. Then, according to this scanning direction and step size, the system sequentially assigns a target physical parameter value to each candidate sub-object in the object set, forming an ordered parameter value sequence, and maps this sequence to the objects in the object set one by one, thereby constructing a monotonic scanning sequence.

[0034] Step S20: Perform simulation capture on each sample in the monotonic scanning sequence under preset unified simulation capture conditions, and obtain the simulation result dataset of the simulation capture. The pre-defined uniform simulation grasping conditions refer to the set of simulation environment and control parameters that remain unchanged for all samples throughout the simulation process. These include, but are not limited to: the three-dimensional model and kinematic parameters of the robotic arm or gripper, the grasping pose (i.e., the spatial position and orientation of the gripper relative to the grasped object), the closing velocity curve of the gripper, the normal target force (i.e., the expected normal force applied by the gripper to the contact surface), the contact model (e.g., the Coulomb friction model or a more refined viscoelastic contact model), the simulation duration, the data sampling period, and the geometric definition of key contact points or key contact areas.

[0035] The simulation results dataset refers to a set of time-series data and result labels output after performing simulation for each sample, including at least: a sequence of normal contact force changes over time. Sequence of tangential contact force changes over time A sequence of contact relative tangential velocities changing over time. Contact area identifiers or contact point trajectory data, and the final capture results. (in =1 indicates that the fetch was successful. =0 indicates that the crawl failed.

[0036] Specifically, the system first loads a pre-defined unified simulation grasping condition configuration file, and sequentially reads the object model and target physical parameter values ​​corresponding to each sample according to a monotonic scanning sequence. For each sample, the system calls the simulation engine to create a simulation scene based on the loaded unified simulation grasping condition configuration file, and records the normal contact force, tangential contact force, relative tangential velocity, and contact point coordinates at each sampling moment in the simulation scene. After the simulation ends, the system generates a grasping result Y based on the pre-defined grasping success criteria (e.g., the object does not fall and the relative displacement is less than a threshold at the end of the simulation). i The time series data and result labels of all samples are organized into a structured simulation result dataset.

[0037] Step S30: Based on the simulation result dataset, perform contact state determination and boundary identification on the monotonic scanning sequence to generate an executable task set; Contact state determination refers to determining whether the contact point of each sample is in a "static contact state" (sticking) or a "slipping state" at each moment during the grasping process, based on the time-series contact data in the simulation result dataset and according to preset physical rules.

[0038] Boundary identification refers to determining the sample index (first transition sample index) and the sample index (first failure sample index) that first appear to transition from static contact to sliding state along the direction of the monotonic scanning sequence, and thereby locating the parameter executable boundary—that is, the parameter subsequence between the first transition sample and the last successful sample before the first failure.

[0039] An executable task set refers to a task set that includes at least all successfully captured samples within the boundary neighborhood, and may optionally include stable and successful tasks outside the boundary as well as failed tasks.

[0040] Specifically, the system first uses the time-series data of each sample to determine the static contact or sliding state at each moment using a preset speed threshold and a sliding judgment force threshold (maximum static friction force minus preset force threshold), and marks the samples that have transitioned from static contact to sliding. Then, it locates the index of the first transition sample and the index of the first failure sample along the scanning direction. Finally, when the index of the first transition is less than the index of the first failure, it extracts the successful samples between the two as the set of executable tasks in the boundary neighborhood, thus completing the generation of the set of executable tasks.

[0041] Step S40: Construct a continuous difficulty task path for the executable task set to obtain a crawling task set.

[0042] A continuous difficulty task path refers to a sequence of tasks with monotonically increasing difficulty formed by sorting the executable tasks in the boundary neighborhood of the executable task set according to how close their parameters are to the first failure boundary. The difficulty score of each task can be quantified based on its relative position in the boundary subsequence or its distance to the first failure sample.

[0043] The grasping task set refers to the final output, which is an ordered list of tasks containing task identifiers, parameter values, grasping results, transition markers, difficulty scores, and continuous difficulty path information. It can be used for progressive training, course learning, or boundary performance evaluation of robot grasping models.

[0044] Specifically, the system obtains the set of executable tasks in the boundary neighborhood and their index range, sorts them according to the scanning direction, calculates the difficulty score for each task (such as the relative distance between the index and the boundary), arranges them from easy to difficult according to the score to form a continuous difficulty task path, and finally packages and outputs the task identifier, parameter values, crawling results, transition markers and difficulty scores to obtain the crawling task set.

[0045] For example, in one specific implementation, a regular columnar body or other homogeneous objects are selected as a family of geometric objects. A two-finger gripper is used as the grasping actuator. The grasping pose, closing velocity, normal target force, contact model, and simulation duration are fixed. Only the friction coefficient μ between the object and the gripper is changed, and a monotonic scanning sequence from high to low is set. A complete grasping simulation is performed for each μ value, and the results are extracted. , , And the final result of the capture. If a contact state transition is detected for the first time at a certain μ value along the scanning sequence, and a capture failure occurs for the first time at a lower μ value, the executable boundary of the parameters can be identified by the first transition sample and the last successful sample before the first failure, and further a boundary neighborhood task set and a continuous difficulty task path can be generated.

[0046] In another specific implementation, while keeping the geometric shape, grasping pose, control strategy and contact model unchanged, the target physical parameters can be set to the object mass m, and the first transition sample, the first failure sample and their corresponding boundary subsequence can be identified by a monotonic scanning method from light to heavy or from heavy to light.

[0047] This embodiment provides a method for generating grasping tasks. First, a monotonic scanning sequence is constructed based on a pre-determined target physical parameters and a set of objects under the same geometric conditions. Then, simulated grasping is performed on each sample in the monotonic scanning sequence under pre-defined unified simulation grasping conditions. By combining the construction of the monotonic scanning sequence with the execution of simulated grasping under the same simulation grasping conditions, interference from geometry and control strategies is eliminated, ensuring that changes in the grasping results are uniquely attributed to the monotonic changes in the target physical parameters. This provides a controllable variable basis for locating the instability boundary. Furthermore, by performing contact state determination and boundary identification based on the simulation result dataset, the physical transition point from static contact to sliding can be extracted. The resulting "executable task set" naturally focuses on the parameter region closest to the instability boundary. In addition, a continuous difficulty task path is constructed for the executable task set, giving the tasks a clear difficulty progression relationship. This ensures that the generated grasping task set not only revolves around the critical region but also has a clear physical evolution logic. In summary, this application generates a set of grasping tasks with stronger physical interpretability by constructing monotonic scanning sequences and combining them with unified simulation conditions, contact state transition detection and boundary recognition, and continuous difficulty task path construction, thereby focusing on the parameter region of the grasping instability boundary.

[0048] Based on the first embodiment of this application, a second embodiment of this application is proposed. In the second embodiment of this application, content that is the same as or similar to that in the first embodiment described above can be referred to the above description and will not be repeated hereafter.

[0049] Based on this, please refer to Figure 2 , Figure 2 This is a flowchart illustrating the second embodiment of this application, as shown below. Figure 2 As shown, before step S10, the task generation method further includes steps S01 to S03: Step S01: Select a reference geometric model and construct a set of candidate objects that have the same geometric shape as the reference geometric model; The baseline geometric model refers to the original three-dimensional geometric shape that serves as the basis for generating the same family of geometric objects; the candidate object set refers to a group of initial objects generated by copying or instantiating the baseline geometric model as a template. Each candidate sub-object in this set has not yet been assigned the final target physical parameter value in the initial state and only has the same geometric shape attributes as the baseline geometric model.

[0050] Specifically, the system first reads a baseline geometric model file from a pre-defined model library and parses its geometric data (including vertex coordinates, facet connectivity, normal information, etc.). Based on the number of objects N specified by the user, the system calls the copy function of the geometry engine to generate N candidate sub-objects with the exact same topological structure and geometric metric values ​​as the baseline geometric model.

[0051] In one possible implementation, the system uses a shallow copy to directly copy the model reference, and then instantiates it when physical parameters are assigned later. In another possible implementation, the system directly performs a deep copy to generate N independent model objects, reserving physical attribute fields for each object. The system stores all generated candidate sub-objects in list form, forming a candidate object set.

[0052] Step S02: Determine whether each candidate sub-object in the candidate object set maintains the same geometric conditions as the baseline geometric model; Understandably, in order to ensure that all objects participating in the simulation grabbing are not significantly different from the benchmark model in terms of geometric shape, thereby eliminating the grabbing result deviation caused by geometric deformation or reconstruction error, further judgment is made on each candidate sub-object in the candidate object set to ensure that the only factor that changes during the parameter scanning process is the target physical parameters.

[0053] Specifically, in one possible implementation, the system extracts the mesh topology (e.g., vertex adjacency) of the candidate sub-object and the baseline geometric model, calculates the topological difference, and determines that the candidate passes if the difference is zero. In another possible implementation, the system performs dense point cloud sampling on both models, calculates the average Euclidean distance and the maximum distance between the sampled points, and determines that the candidate is geometrically consistent if the average distance is less than a preset threshold (e.g., 0.001 times the model feature size) and the maximum distance is less than twice the threshold. The system generates a Boolean-type determination result for each candidate sub-object, indicating whether it passes the same geometric condition verification.

[0054] Step S03: If the candidate sub-object maintains the same geometric conditions as the baseline geometric model, then construct an object set based on the candidate sub-object.

[0055] Specifically, all candidate sub-objects that pass the validation are selected and moved from the candidate object set to the official object set. For candidate sub-objects that fail the validation, the system marks them as invalid and removes them from the set.

[0056] Additionally, the system initializes a list of physical parameters for each object in the object collection. The values ​​of the target physical parameters are temporarily empty or set to default values, and will be assigned values ​​when the monotonic scan sequence is constructed.

[0057] In this embodiment, the system selects a baseline geometric model and filters out candidate sub-objects that are completely identical to its geometric shape, constructing a set of objects with fixed geometry that only need to be parameterized. This process eliminates the interference of geometric shape differences on the grasping results, ensuring that subsequent changes in grasping performance are solely attributable to the monotonic changes in the target's physical parameters, providing a clean and controllable experimental object basis for accurately locating the executable boundaries of the parameters.

[0058] Based on any of the above embodiments of this application, a third embodiment of this application is proposed. In this third embodiment, content that is the same as or similar to any of the above embodiments can be referred to the above description, and will not be repeated hereafter.

[0059] In this embodiment, step S10, which constructs a monotonic scanning sequence based on a set of objects with predetermined target physical parameters and the same geometric conditions, may include steps S11 to S13: Step S11: Select the target physical parameter from multiple candidate physical parameters based on the parameter selection instruction sent by the user terminal; Specifically, the system lists a preset list of candidate physical parameters and their possible value ranges. The user selects one of the parameters through interactive operation, and the system receives the instruction and parses out the name and type of the target physical parameter.

[0060] Step S12: Determine the scanning step size of the target physical parameters according to the preset constraints, and set the monotonic scanning direction; The preset constraints refer to a set of rules set in advance to regulate the scanning step size and scanning direction. These rules include at least the following: adjacent samples only change in the target physical parameters while the other physical parameters remain unchanged; the scanning step size must meet the continuity requirement (i.e., the step size is constant or changes smoothly according to a preset rule); and the scanning direction is fixed to either monotonically increasing or monotonically decreasing, and cannot be reversed midway.

[0061] The scan step size refers to the difference in target physical parameters between two adjacent samples in a monotonic scan sequence. It can be a constant step size or a variable step size with a smaller step size in certain regions of the parameter space (such as near the prediction boundary).

[0062] Monotonic scanning direction refers to the trend in which the values ​​of the target physical parameters increase monotonically (from easy to difficult or from difficult to easy) or decrease monotonically as the sample index increases.

[0063] Specifically, the system first reads the determined target physical parameters to obtain their physical value range. The system automatically calculates the scanning step size based on preset constraints: in one possible implementation, the system uses a constant step size, dividing the value range by the total number of samples minus one to obtain the step size value. In another possible implementation, the system employs an adaptive variable step size strategy, using larger step sizes at both ends of the parameter space to quickly scan stable regions, while automatically reducing the step size near empirically predicted critical regions to improve boundary recognition resolution. The system sets a monotonic scanning direction: if the target physical parameter is the coefficient of friction and the desired progression is from easy to difficult, the direction is set to decrease (from high friction to low friction); if the desired progression is from difficult to easy, the direction is set to increase. Users can also directly specify the direction via commands. The system stores the determined step size and direction as scanning parameters.

[0064] Step S13: Based on the scanning step size and the scanning direction, assign different target physical parameter values ​​to the candidate sub-objects in the object set in sequence to obtain a monotonic scanning sequence.

[0065] Specifically, the system obtains the object collection {O i} and scan step size And the scan direction dir, traversing each object according to the inherent order of the object collection. For the i-th object O i The system calculates the target's physical parameters based on the scanning direction (dir). If the scan direction dir is increasing, then = If the direction is decreasing, then = .in, The initial values ​​for the target physical parameters.

[0066] In this embodiment, the system selects the physical parameters of a single target according to user instructions, determines the scanning step size and monotonic direction that satisfy the continuity constraint, and sequentially assigns parameter values ​​to a geometrically consistent set of objects, constructing a scanning sequence with monotonically changing parameters. This sequence ensures that the grasping difficulty evolves monotonically along the sequence, providing an ordered and high-resolution sample basis for subsequent accurate positioning of the first transition and the first failure boundary, significantly improving the accuracy and efficiency of boundary recognition.

[0067] Based on any of the above embodiments of this application, a fourth embodiment of this application is proposed. In this fourth embodiment, content that is the same as or similar to any of the above embodiments can be referred to the above description, and will not be repeated hereafter.

[0068] Based on this, please refer to Figure 3 , Figure 3 This is a flowchart illustrating the fourth embodiment of this application, as shown below. Figure 3As shown, the simulation result dataset includes contact time series data and grasping results. Step S30, which involves determining the contact state and identifying the boundary of the monotonic scanning sequence based on the simulation result dataset to generate an executable task set, may include steps S31-S33: Step S31: Based on the contact timing data, determine the contact state of each sample in the monotonic scan sequence to obtain the transition marker corresponding to the sample; The transition flag is a Boolean variable used to identify whether at least one key contact point of the sample has undergone a state transition from static contact state to sliding state during the grasping process. A true value (such as 1) indicates that a transition has occurred, and a false value (such as 0) indicates that no transition has occurred.

[0069] Specifically, the system first extracts the normal contact force time-series data, tangential contact force time-series data, and contact relative tangential velocity time-series data for each sample from the simulation result dataset. For each sample, the system iterates through each simulation time t in chronological order and determines the contact state of the sample at the current simulation time t. Further, the system maintains the contact states of key contact points throughout the entire time window. If the contact state of a key contact point changes within the time window, the transition flag for that sample is set to true; if no transition is detected within the entire time window, it is marked as false. In one possible implementation, the system also allows the detection of multiple contact points, marking any contact point that experiences a transition as true.

[0070] Step S32: Based on the transition markers and the capture results, perform boundary identification on the monotonic scan sequence to obtain the first transition sample index and the first failure sample index; The first transition sample index refers to the position number of the first sample marked as true in the scan sequence when traversing along the monotonic scan direction.

[0071] The first failed sample index refers to the sequence number of the first sample that is fetched as a false value (failure) when traversing in the same direction.

[0072] Specifically, by traversing along the monotonic scanning direction and jointly analyzing the transition markers and grasping results, the system accurately locates two key boundary points: the sample that first shows signs of physical sliding and the sample that first fails to grasp.

[0073] Step S33: Generate an executable task set based on the first transition sample index and the first failure sample index.

[0074] The set of executable tasks includes the set of executable tasks in the boundary neighborhood, the set of benchmark stable tasks, and the set of failure verification tasks.

[0075] Specifically, the first jump index marks the critical starting point when crawling begins to enter an unstable state, and the first failure index marks the critical ending point when crawling completely loses stability. The region between these two is the most valuable parameter interval for research, namely the executable region of the boundary neighborhood.

[0076] In this embodiment, transition markers are obtained by determining the state of the contact time-series data at each time step. The first transition and the first failure sample index are accurately located by combining the capture results, thereby generating a set of executable tasks in the executable region of the focused boundary neighborhood.

[0077] Furthermore, based on the fourth embodiment of this application, a feasible implementation method is proposed. In this implementation method, step S31 may include steps S311 to S316: Step S311: Obtain the normal contact force, tangential contact force, and relative tangential velocity of the sample at each simulation moment in the contact timing data; Contact time series data refers to a series of contact physical quantities that change over time and are recorded according to a fixed sampling period during the simulation grasping process. Each sampling moment corresponds to a set of data points.

[0078] Specifically, for the sample currently being processed, the system reads its pre-stored arrays of normal contact force, tangential contact force, and relative tangential velocity. The length of each array is equal to the simulation duration divided by the sampling period.

[0079] Step S312: Calculate the maximum static friction force at the current simulation moment based on the normal contact force and the preset friction coefficient, and subtract the preset force threshold from the maximum static friction force to obtain the sliding judgment force threshold. The preset friction coefficient refers to the target physical parameter value assigned to the sample. It reflects the frictional characteristics between the contact surfaces and is dimensionless.

[0080] Maximum static friction refers to the minimum tangential force required for the contact surfaces to begin sliding relative to each other. Its calculation formula is: ,in, Indicates the current simulation time. t The normal contact force.

[0081] The formula for calculating the sliding force threshold is: ,in, A preset force threshold is used to construct a hysteresis interval to avoid misjudgment of the state caused by numerical jitter.

[0082] Step S313: Under the condition that the relative tangential velocity of the contact is not greater than a preset velocity threshold and the tangential contact force is less than the sliding determination force threshold, it is determined that the sample is in a static contact state at the current simulation moment. Under the condition that the relative tangential velocity at contact is not greater than a preset velocity threshold and the tangential contact force is less than the sliding determination force threshold, i.e. and Under the condition that the sample is at the current simulation time, determine the current simulation time. t It is in a sticking static contact state.

[0083] in, Indicates the preset speed threshold. Indicates the current simulation time. t The contact relative tangential velocity below, This represents the tangential contact force at the current simulation moment.

[0084] Step S314: Under the condition that the relative tangential velocity of the contact is greater than a preset velocity threshold and the tangential contact force is not less than the sliding determination force threshold, it is determined that the sample is in a sliding state at the current simulation moment. Under the condition that the relative tangential velocity of the contact is greater than a preset velocity threshold and the tangential contact force is not less than the sliding determination force threshold, i.e. and Under the condition that the sample is at the current simulation time, determine the current simulation time. t It is in a slipping state.

[0085] in, Indicates the preset speed threshold. Indicates the current simulation time. t The contact relative tangential velocity below, This represents the tangential contact force at the current simulation moment.

[0086] Step S315: Track whether the sample exhibits a state transition from the static contact state to the sliding state along the time axis of the contact timing data; The contact state determination results are read sequentially along the simulation time axis according to the sampling period, while maintaining the state variables of the previous time. For each current time, if the previous time was a static contact state and the current time is a sliding state, it is determined that a state transition from static contact to sliding has occurred.

[0087] Step S316: If the state transition occurs, the transition of the sample is marked as a true value.

[0088] If at least one state transition from static contact to sliding is detected during the entire grasping simulation of the sample, the transition flag corresponding to the sample is set to a true value (e.g., logic value 1 or Boolean value true); if no transition event is ever detected, the transition flag is set to a false value (e.g., 0 or false). This transition flag is stored in association with the sample identifier and serves as input for subsequent boundary identification.

[0089] Furthermore, based on the fourth embodiment of this application, a feasible implementation method is proposed. In this implementation method, step S32 may include steps S321 to S323: Step S321: Obtain the transition marker and capture result corresponding to each sample in the monotonic scan sequence; The system obtains the transition marker corresponding to each sample in the monotonic scan sequence, denoted as z[1,...,n], where n is the length of the monotonic scan sequence, and i=1,2,...,n corresponds to the index order of the samples in the sequence. Further, the system obtains the grasping result Y[1,...,n] corresponding to each sample in the monotonic scan sequence, where Y[i]=1 indicates success, and Y[i]=0 indicates failure. Here, n is the length of the monotonic scan sequence, and i=1,2,...,n corresponds to the index order of the samples in the sequence. The system stores these two arrays in index-aligned order to ensure that the transition marker corresponding to index i and the grasping result belong to the same sample.

[0090] Step S322: Traverse the monotonic scan sequence along the monotonic scan direction and record the smallest index of the transition marked as true as the first transition sample index. Specifically, the system iterates incrementally along the monotonic scanning direction of the monotonic scanning sequence starting from i=1, checking whether z[i] is true for each index i. If z[i] is true, the current index i is immediately set as the first transition sample index i_tr, and the traversal is terminated. The first transition sample index i_tr = min{i|z[i]=1}.

[0091] Understandably, if no true value is found after traversing all indices, i_tr retains its initial special value, and the system records that no first transition sample was found.

[0092] Step S323: Record the smallest index of the capture result that is false as the first failed sample index.

[0093] Specifically, the system iterates incrementally along the monotonic scanning direction of the monotonic scanning sequence starting from i=1, and checks whether Y[i] is equal to 0 (false value) for each index i. If Y[i]=0, the current index i is immediately recorded as the first failed sample index i_fail, and the traversal is terminated, where the first failed sample index i_fail=min{i|Y[i]=0}.

[0094] It can be understood that if no failed sample is found after traversing all indexes, i_fail remains at the initial special value, and the system records that the first failed sample is not found.

[0095] Further, based on the fourth embodiment of the present application, a feasible implementation is proposed. In this implementation, step S33 may include steps S331 to S335: In step S331, under the condition that the first transition sample index is smaller than the first failed sample index, the last successful sample index is determined based on the first failed sample index; It should be noted that the condition that the first transition sample index is smaller than the first failed sample index means that the first transition sample index and the first failed sample index exist, and the first transition sample index is smaller than the first failed sample index.

[0096] The system judges whether i_tr<i_fail is satisfied, and if so, proceeds to execute; if not (for example, i_tr≥i_fail or any index is not found), the system determines that the current scan sequence cannot form a valid boundary, terminates subsequent task generation and outputs a corresponding prompt.

[0097] When the first transition sample index and the first failed sample index exist, and the condition that the first transition sample index is smaller than the first failed sample index is satisfied, the system calculates the last successful sample index i_last through arithmetic operation: i_last=i_fail-1, that is, the index of the sample immediately before the first failed sample, and the grasping result of this sample is success.

[0098] In step S332, a parameter executable boundary is defined by the first transition sample index and the last successful sample index, and a boundary neighborhood executable task set is generated based on the parameter executable boundary; The parameter executable boundary refers to a continuous index interval [i_tr, i_last] defined by the first transition sample index i_tr and the last successful sample index i_last. Each sample in this interval corresponds to a target physical parameter value, and all these samples exhibit final success in the grasping simulation, but a state transition from static contact to sliding may have occurred.

[0099] The boundary neighborhood executable task set is denoted as T_boundary, which is a task set formed by all successfully grasped samples within the parameter executable boundary interval, and each task has a critical characteristic of "close to instability but not yet failed".

[0100] Specifically, the system obtains the index of the first transition sample i_tr and the index of the last successful sample i_last, as well as the capture result Y[i] for each sample. The system constructs an empty set of executable tasks T_boundary within the boundary neighborhood. The system traverses the index i from i_tr to i_last (inclusive). For each index i, it checks whether the corresponding capture result Y[i] is true (successful). Since i_last is defined before the first failure, all samples with i in the range [i_tr, i_last] should theoretically succeed, but for safety, the system still performs verification. If Y[i]=1, the system adds the relevant information of the sample (such as object identifier, target physical parameter value, transition flag, capture result, etc.) to T_boundary.

[0101] In one possible implementation, the system also records additional features such as the transition time and sliding duration of each boundary sample for subsequent difficulty ranking. The system outputs T_boundary, obtaining the set of executable tasks in the boundary neighborhood, Tboundary={O i |i_tr<=i<=i_last,Y[i]=1}.

[0102] Step S333: The samples in the monotonic scan sequence whose index is less than the index of the first transition sample and whose capture result is true are denoted as the benchmark stable task set; The system constructs an empty baseline stable task set T_stable. The system iterates through indices i from 1 to i_tr-1 (i.e., all indices less than i_tr). For each index i, if Y[i] = 1 (success), the relevant information for that sample is added to T_stable, where T_stable = {O_i}. i |i <i_tr,Y[i]=1}。

[0103] Step S334: The samples in the monotonic scanning sequence whose index is not less than the index of the first failure sample and whose capture result is false are denoted as the failure verification set; The system constructs an empty failure verification set T_fail. The system iterates through indices i from i_fail to n (the end of the sequence). For each index i, if Y[i] = 0 (failure), the relevant information for that sample is added to T_fail. Since i_fail is defined as the first failure index, all subsequent failure samples are also included. Where T_fail = {O i |i>=i_fail,Y[i]=0}.

[0104] Step S335: Generate an executable task set based on the boundary neighborhood executable task set, the benchmark stable task set, and the failure verification set.

[0105] The system acquires the set of executable tasks in the boundary neighborhood, the set of benchmark stable tasks, and the set of failure verification tasks. These sets are merged into a unified data structure, typically stored as a dictionary or a list of objects. Each task contains the following fields: sample index, object identifier, target physical parameter value, capture result (success / failure), transition flag (true / false), and set type ("stable", "boundary", or "fail"). The system can add an additional "difficulty score" field (initially set to be calculated) for tasks within the T_boundary. Finally, the system outputs the set of executable tasks.

[0106] Based on any of the above embodiments of this application, a fifth embodiment of this application is proposed. In this fifth embodiment, content that is the same as or similar to any of the above embodiments can be referred to the above description, and will not be repeated hereafter.

[0107] In this embodiment, step S40, which constructs a continuous difficulty task path for the executable task set to obtain the crawling task set, may include steps S41-S42: Step S41: Sort the set of executable tasks in the boundary neighborhood of the executable task set based on the preset difficulty definition score to construct a continuous difficulty task path; The preset difficulty definition score refers to a numerical indicator used to quantify the relative difficulty of each task in the set of executable tasks in the boundary neighborhood. This score can be calculated based on the relative position of the sample in the parameter executable boundary subsequence, for example, using the formula d_i=(i-i_tr) / max(1,i_last-i_tr), where i is the sample index, i_tr is the index of the first jump sample, and i_last is the index of the last successful sample. The score ranges from 0 to 1. The lower the score, the easier the task is (closer to the first jump point), and the higher the score, the more difficult the task is (closer to the first failure point).

[0108] Specifically, the system first obtains the set of executable tasks T_boundary in the boundary neighborhood, as well as i_tr and i_last. The system iterates through each task in T_boundary, obtaining its sample index i (the position of this index in the monotonic scan sequence). The system calculates the difficulty score for each task according to a preset difficulty definition formula. All tasks in T_boundary are sorted in ascending order of difficulty score (from easiest to hardest), forming an ordered task list, which is the continuous difficulty task path.

[0109] Step S42: Generate a set of grabbing tasks based on the set of executable tasks that construct the continuous difficulty task path.

[0110] Specifically, the system collects the following information: baseline geometric model identifier and object set information, target physical parameter names and monotonic scan sequences (parameter values ​​for each sample), capture results for each sample, transition markers for each sample, the index of the first transition sample i_tr and the index of the first failure sample i_fail, the index of the last successful sample i_last, three task sets T_stable, T_boundary, and T_fail, and the paths for tasks of varying difficulty. The system organizes the above information according to a preset data structure to generate a capture task set.

[0111] In this embodiment, by calculating and sorting difficulty scores for tasks in the boundary neighborhood, a continuous difficulty task path from easy to difficult is constructed, and then all task information is integrated to output a structured crawling task set. This process organizes discrete critical parameter samples into a smooth and progressive learning sequence, significantly improving the usability and efficiency of the task set in course learning and boundary performance evaluation.

[0112] This application also provides a task generation device for grabbing tasks; please refer to [reference needed]. Figure 4 The task generation device includes: Scanning module 10 is used to construct a monotonic scanning sequence for a set of objects under preset geometric conditions based on predetermined target physical parameters; The simulation module 20 is used to perform simulation capture on each sample in the monotonic scanning sequence under preset uniform simulation capture conditions, and to obtain the simulation result dataset of the simulation capture. The determination module 30 is used to determine the contact state and identify the boundary of the monotonic scanning sequence based on the simulation result dataset, and generate a set of executable tasks. The generation module 40 is used to construct a continuous difficulty task path for the executable task set to obtain a capture task set.

[0113] The crawling task generation apparatus provided in this application, employing the crawling task generation method in the above embodiments, can solve the technical problem of crawling task generation. Compared with the prior art, the beneficial effects of the crawling task generation apparatus provided in this application are the same as those of the crawling task generation method provided in the above embodiments, and other technical features in the crawling task generation apparatus are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0114] This application provides a crawling task generation device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the crawling task generation method in the first embodiment described above.

[0115] The following is for reference. Figure 5 The diagram illustrates a structural schematic suitable for implementing the crawling task generation device of the embodiments of this application. The crawling task generation device in the embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 5 The grabbing task generation device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0116] like Figure 5 As shown, the grasping task generation device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory 1002 or a program loaded from a storage device 1003 into a random access memory 1004. The random access memory 1004 also stores various programs and data required for the operation of the grasping task generation device. The processing unit 1001, the read-only memory 1002, and the random access memory 1004 are interconnected via a bus 1005. An input / output interface 1006 is also connected to the bus. Typically, the following systems can be connected to the input / output interface 1006: input devices 1007 including, for example, a touchscreen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. The communication device 1009 allows the grabbing task generation device to communicate wirelessly or wiredly with other devices to exchange data. Although grabbing task generation devices with various systems are shown in the figures, it should be understood that implementation or possession of all the systems shown is not required. More or fewer systems may be implemented alternatively.

[0117] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from read-only memory 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0118] The crawling task generation device provided in this application, employing the crawling task generation method in the above embodiments, can solve the technical problem of crawling task generation. Compared with the prior art, the beneficial effects of the crawling task generation device provided in this application are the same as those of the crawling task generation method provided in the above embodiments, and other technical features in this crawling task generation device are the same as those disclosed in the method of the previous embodiment, and will not be repeated here.

[0119] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0120] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0121] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, which are used to execute the crawling task generation method in the above embodiments.

[0122] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0123] The aforementioned computer-readable storage medium may be included in the capture task generation device; or it may exist independently and not assembled into the capture task generation device.

[0124] The aforementioned computer-readable storage medium carries one or more programs. When the one or more programs are executed by the crawling task generation device, the crawling task generation device: constructs a monotonic scanning sequence based on a pre-determined target physical parameters and a set of objects under preset geometric conditions; performs simulated crawling on each sample in the monotonic scanning sequence under preset unified simulation crawling conditions, and obtains a simulation result dataset of the simulated crawling; performs contact state determination and boundary recognition on the monotonic scanning sequence based on the simulation result dataset, and generates an executable task set; and constructs a continuously challenging task path for the executable task set to obtain a crawling task set.

[0125] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0126] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0127] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0128] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described crawling task generation method, thereby solving the technical problem of crawling task generation. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the crawling task generation method provided in the above embodiments, and will not be repeated here.

[0129] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described task generation method.

[0130] The computer program product provided in this application can solve the technical problem of crawling task generation. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as the beneficial effects of the crawling task generation method provided in the above embodiments, and will not be repeated here.

[0131] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for generating a crawling task, characterized in that, The method for generating crawling tasks includes: The steps of constructing a monotonic scanning sequence for a set of objects under the same preset geometric conditions based on predetermined target physical parameters include: selecting a target physical parameter from multiple candidate physical parameters based on a parameter selection instruction sent by the user terminal; determining the scanning step size of the target physical parameter according to preset constraints and setting a monotonic scanning direction; and assigning different target physical parameter values ​​to candidate sub-objects in the set of objects in sequence according to the scanning step size and the scanning direction to obtain a monotonic scanning sequence. Under preset uniform simulation capture conditions, simulation capture is performed on each sample in the monotonic scanning sequence, and the simulation result dataset of the simulation capture is obtained; the simulation result dataset includes contact time series data and capture results; Based on the simulation result dataset, contact state determination and boundary identification are performed on the monotonic scanning sequence to generate an executable task set; the steps include: determining the contact state of each sample in the monotonic scanning sequence based on the contact time series data to obtain a transition marker corresponding to the sample; performing boundary identification on the monotonic scanning sequence based on the transition marker and the grasping result to obtain the first transition sample index and the first failure sample index; and generating an executable task set based on the first transition sample index and the first failure sample index. Construct a series of task paths with varying difficulty for the set of executable tasks to obtain a set of crawling tasks; The step of determining the contact state of each sample in the monotonic scan sequence based on the contact time series data to obtain the transition marker corresponding to the sample includes: Obtain the normal contact force, tangential contact force, and relative tangential velocity of the sample at each simulation moment in the contact timing data; The maximum static friction force at the current simulation moment is calculated based on the normal contact force and the preset friction coefficient, and the preset force threshold is subtracted from the maximum static friction force to obtain the sliding judgment force threshold. Under the condition that the relative tangential velocity of the contact is not greater than a preset velocity threshold and the tangential contact force is less than the sliding determination force threshold, it is determined that the sample is in a static contact state at the current simulation moment; Under the condition that the relative tangential velocity of the contact is greater than a preset velocity threshold and the tangential contact force is not less than the sliding determination force threshold, it is determined that the sample is in a sliding state at the current simulation moment; Track whether the sample exhibits a state transition from the static contact state to the sliding state along the time axis of the contact timing data; If the state transition occurs, the transition of the sample is marked as a true value; If the state transition does not occur, the transition of the sample is marked as a false value. The steps of generating a set of executable tasks based on the first transition sample index and the first failure sample index include: If the index of the first transition sample is less than the index of the first failure sample, the index of the last successful sample is determined based on the index of the first failure sample. The parameter executable boundary is defined by the first transition sample index and the last successful sample index, and a set of executable tasks in the boundary neighborhood is generated based on the parameter executable boundary; The samples in the monotonic scan sequence whose index is less than the index of the first transition sample and whose capture result is true are denoted as the benchmark stable task set; The samples in the monotonic scanning sequence whose index is not less than the index of the first failure sample and whose capture result is false are denoted as the failure verification set. An executable task set is generated based on the boundary neighborhood executable task set, the benchmark stable task set, and the failure verification set.

2. The task generation method as described in claim 1, characterized in that, The step of constructing a monotonic scanning sequence based on a set of objects with predetermined target physical parameters and the same geometric conditions also includes: Select a reference geometric model and construct a set of candidate objects that have the same geometric shape as the reference geometric model; Determine whether each candidate sub-object in the candidate object set maintains the same geometric conditions as the baseline geometric model; If the candidate sub-object maintains the same geometric conditions as the baseline geometric model, then an object set is constructed based on the candidate sub-object.

3. The task generation method as described in claim 1, characterized in that, The step of performing boundary identification on the monotonic scan sequence based on the transition markers and the capture results to obtain the first transition sample index and the first failure sample index includes: Obtain the transition marker and capture result corresponding to each sample in the monotonic scan sequence; The monotonic scan sequence is traversed along the monotonic scan direction, and the smallest index in which the transition is marked as a true value is recorded as the first transition sample index. The smallest index of the crawling result that is false is denoted as the first failed sample index.

4. The task generation method as described in claim 1, characterized in that, The step of constructing a series of task paths with varying difficulty for the set of executable tasks to obtain a set of crawling tasks includes: The set of executable tasks in the boundary neighborhood of the executable task set is sorted based on a preset difficulty score in order to construct a continuous difficulty task path; A set of grabbing tasks is generated based on the set of executable tasks that construct the continuous difficulty task path.

5. A task generation device for capturing data, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the grabbing task generation method as described in any one of claims 1 to 4.

6. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the crawling task generation method as described in any one of claims 1 to 4.

Images