Laser cutting machine feeding control method and system, electronic equipment and readable storage medium

Abstract

This application relates to the field of industrial automation technology, and in particular to a method, system, electronic device, and readable storage medium for controlling the loading of a laser cutting machine. The method involves real-time acquisition of sensing data of the raw material to be loaded; transmitting the sensing data to a pre-trained multimodal fusion model for analysis and processing to generate gripping decision information including the coordinates of the gripping spatial points, gripping posture, and target gripping force; dynamically planning the motion path of the robotic arm and the force control parameters of the end effector based on the decision information to form loading control commands; and controlling the robotic arm and end effector based on the loading control commands to execute the gripping and loading actions of the raw material, while adjusting the force control parameters in a closed loop based on real-time force feedback data during the gripping process. This achieves automated operation of the loading process in a laser cutting machine, reduces manual intervention, improves production efficiency, reduces the risk of raw material damage, and meets the needs of multi-variety and flexible production.

Description

Technical Field

[0001] This application relates to the field of industrial automation technology, and in particular to a method, system, electronic device and readable storage medium for controlling the feeding of a laser cutting machine. Background Technology

[0002] Before processing raw materials such as sheets and pipes, laser cutting machines need to accurately pick up and place the raw materials at the processing position through a feeding system. As the manufacturing industry develops towards flexible production with multiple varieties and small batches, the types of raw materials to be fed are becoming increasingly diverse, including irregularly shaped parts with various geometric shapes (such as curved panels and irregular pipes), and the incoming materials are often placed in a random stacked state.

[0003] Common automated feeding technologies mainly rely on two-dimensional visual positioning or simple three-dimensional matching based on preset templates. However, for irregularly shaped raw materials with complex and varied geometries, preset templates are difficult to fully cover, leading to recognition failures or inaccurate positioning. Secondly, for raw materials with random stacking states (such as positional offset, angle tilt, and interlayer compression), single visual information cannot reliably perceive their spatial posture and stable gripping points, easily leading to slippage, collisions, or material deformation during the gripping process. Furthermore, the lack of real-time perception and feedback of physical interactions (such as contact forces) during the gripping process makes it impossible to adjust according to the actual situation at the moment of gripping.

[0004] Therefore, traditional feeding technology for laser cutting machines is difficult to effectively handle the reliable gripping of irregularly shaped raw materials with uncertain geometry, spatial position, and surface characteristics. This results in poor adaptability of automated feeding processes to such materials, low gripping success rates, and often requires manual intervention.

[0005] The above content is only used to help understand the technical solution of this application and does not represent an admission that the above content is prior art. Summary of the Invention

[0006] The main purpose of this application is to provide a laser cutting machine feeding control method, system, electronic device and readable storage medium, which aims to solve the technical problems of poor adaptability and low grasping success rate of traditional laser cutting machine feeding methods due to reliance on preset templates and single sensors for raw materials with irregular shapes and random stacking states.

[0007] To achieve the above objectives, this application proposes a laser cutting machine feeding control method, which includes:

[0008] Real-time acquisition of sensing data of raw materials to be loaded, including at least three-dimensional point cloud data, surface image data and force feedback data during the grasping process;

[0009] The perceived data is transmitted to a pre-trained multimodal fusion model for analysis and processing to generate grasping decision information including grasping space point coordinates, grasping posture and target grasping force;

[0010] Based on the decision information, the motion path of the robotic arm and the force control parameters of the end effector are dynamically planned to form a feeding control command;

[0011] Based on the feeding control command, the robotic arm and end effector are controlled to perform the grasping and feeding actions of raw materials, and the force control parameters are adjusted in a closed loop according to real-time force feedback data during the grasping process.

[0012] In one embodiment, the pre-trained multimodal fusion model is a neural network based on the Transformer architecture; the step of transmitting the perception data to the pre-trained multimodal fusion model for analysis and processing to generate grasping decision information including grasping space point coordinates, grasping posture, and pre-target grasping force includes:

[0013] The three-dimensional point cloud data, surface image data, and force feedback data are respectively feature-encoded;

[0014] Interactive and alignment operations are performed on the encoded features to generate unified multimodal fusion features;

[0015] The multimodal fusion features are passed to the regression and classification network head, and the grasping decision information is output.

[0016] In one embodiment, the step of transmitting the perceived data to a pre-trained multimodal fusion model for analysis and processing to generate grasping decision information including grasping space point coordinates, grasping posture, and target grasping force further includes:

[0017] Identify surface defects of the raw materials in the surface image data;

[0018] If surface defects exist in the raw material, the grasping posture is planned to avoid the defective area when generating the grasping decision information.

[0019] In one embodiment, the step of dynamically planning the motion path of the robotic arm and the force control parameters of the end effector based on the decision information to form a feeding control command includes:

[0020] The grasping space point coordinates and grasping posture in the grasping decision information are used as path endpoint constraints, and the three-dimensional point cloud data is transmitted to the motion planning algorithm to generate a collision-free motion trajectory from the current pose of the robotic arm to the path endpoint in three-dimensional space.

[0021] Based on the target grasping force and the physical property model of the raw materials, the initial grasping force threshold and force control stiffness parameters of the end effector are determined.

[0022] The collision-free motion trajectory, the initial gripping force threshold, and the force control stiffness parameters are encapsulated to form the feeding control command.

[0023] In one embodiment, the step of controlling the robotic arm and end effector based on the feeding control command to perform the grasping and feeding actions of raw materials, and adjusting the force control parameters in a closed loop based on real-time force feedback data during the grasping process includes:

[0024] Based on the feeding control command, the robotic arm is controlled to move along the planned motion trajectory, and when the end effector contacts the surface of the raw material, it switches to a force-position hybrid control mode.

[0025] In the force-position hybrid control mode, with the goal of maintaining the target grasping force, the position and / or output torque of the end effector are dynamically adjusted according to the real-time force feedback data;

[0026] When the real-time force feedback data is stable within the preset successful grasping force range and continues for more than the first set time, the grasping is determined to be successful, and the robotic arm is controlled to perform subsequent loading movement.

[0027] If a sudden change or slippage is detected in the real-time force feedback data during the grasping process, the grasping failure process is triggered.

[0028] In one embodiment, before the step of transmitting the perceived data to a pre-trained multimodal fusion model for analysis and processing to generate grasping decision information including grasping space point coordinates, grasping posture, and target grasping force, the method further includes:

[0029] In response to a command to change the end effector or a command to change the target raw material type, the system retrieves the multimodal fusion model parameters and force control parameter baselines that match the current hardware configuration and raw material type from the pre-stored parameter package.

[0030] In one embodiment, after the step of controlling the robotic arm and end effector based on the feeding control command to perform the grasping and feeding actions of raw materials, and adjusting the force control parameters in a closed loop based on real-time force feedback data during the grasping process, the method further includes:

[0031] During the operation of the robotic arm, the joint vibration spectrum and drive current data of the robotic arm are continuously collected;

[0032] The joint vibration spectrum and drive current data are transmitted to the health prediction model to determine the health status of the robotic arm, and a maintenance prompt is issued when the health status reaches the health warning threshold.

[0033] Furthermore, to achieve the above objectives, this application also proposes a laser cutting machine feeding control system, characterized in that the laser cutting machine feeding control system includes:

[0034] A multimodal sensing module is used to collect sensing data of the raw materials to be fed in real time. The sensing data includes at least three-dimensional point cloud data, surface image data, and force feedback data during the grasping process.

[0035] The edge computing module is used to transmit the perceived data to a pre-trained multimodal fusion model for analysis and processing, and generate grasping decision information including grasping space point coordinates, grasping posture and target grasping force;

[0036] The planning and control module is used to dynamically plan the motion path of the robotic arm and the force control parameters of the end effector based on the decision information, and to generate feeding control commands.

[0037] The loading execution module is used to control the robotic arm and end effector based on the loading control command to perform the grasping and loading actions of raw materials, and to make closed-loop adjustments to the force control parameters based on real-time force feedback data during the grasping process.

[0038] In addition, to achieve the above objectives, this application also proposes an electronic 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 laser cutting machine feeding control method described above.

[0039] In addition, to achieve the above objectives, this application also proposes a readable 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 laser cutting machine feeding control method described above.

[0040] One or more technical solutions proposed in this application have at least the following technical effects:

[0041] By acquiring comprehensive status information of the raw materials to be fed through multimodal perception, and combining it with a pre-trained multimodal fusion model to generate accurate grasping decisions, and then through dynamic path planning and force control parameter adjustment, the robotic arm can achieve compliant grasping and stable feeding of the raw materials. At the same time, with the help of a closed-loop adjustment mechanism of real-time force feedback, it can effectively compensate for uncertainties such as visual positioning errors and material deformation, improve the robustness and adaptability of the feeding process, and ultimately achieve automated operation of the feeding process of the laser cutting machine, reduce manual intervention, improve production efficiency, reduce the risk of raw material damage, and meet the needs of multi-variety and flexible production. Attached Figure Description

[0042] 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.

[0043] 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.

[0044] Figure 1 This is a flowchart illustrating the first embodiment of the laser cutting machine feeding control method of this application;

[0045] Figure 2 This is a flowchart illustrating the second embodiment of the laser cutting machine feeding control method of this application;

[0046] Figure 3 This is a flowchart illustrating the third embodiment of the laser cutting machine feeding control method of this application;

[0047] Figure 4 This is a schematic diagram of the module structure of the laser cutting machine feeding control system of this application;

[0048] Figure 5 This is a schematic diagram of the equipment structure of the hardware operating environment involved in the laser cutting machine feeding control method in the embodiments of this application.

[0049] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0050] 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.

[0051] 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.

[0052] It should be noted that the execution subject of the laser cutting machine feeding control method in various embodiments of this application can be a laser cutting machine feeding control system, or a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, mobile phone, etc., or an electronic device capable of realizing the above functions, etc. This embodiment does not specifically limit it in this way. The following uses the laser cutting machine feeding control system as the execution subject as an example to describe this embodiment and the following embodiments.

[0053] Based on this, this application proposes a laser cutting machine feeding control method according to the first embodiment. Please refer to [link / reference]. Figure 1 The laser cutting machine feeding control method includes steps S100 to S400:

[0054] Step S100: Real-time acquisition of sensing data of the raw materials to be loaded. The sensing data includes at least three-dimensional point cloud data, surface image data, and force feedback data during the gripping process.

[0055] It should be noted that sensing data of the raw materials to be fed can be collected through a multimodal sensing system. Specifically, the multimodal sensing system may include a structured light camera, a ToF camera, an RGB-D camera, a near-infrared sensor, and a six-dimensional force sensor integrated on the end effector. Among them, the three-dimensional point cloud data is collected by fusing the structured light camera and the ToF camera; the surface image data is collected by combining the RGB-D camera with the near-infrared sensor; and the force feedback data is collected by the six-dimensional force sensor.

[0056] Optionally, a binocular structured light 3D camera and a ToF camera work together to scan the raw material stack from different angles, and generate 3D point cloud data through point cloud registration and fusion algorithms. The 3D point cloud data contains the geometric shape, spatial position, and stacking layer information of the raw materials. An RGB-D camera simultaneously acquires color images and depth information of the raw materials to form surface image data. Meanwhile, an infrared laser projector is used as an auxiliary light source to ensure image quality in low ambient light conditions. A six-dimensional force sensor is installed on the end flange of the robotic arm to measure the force and torque of the end effector in three translational and three rotational directions in real time during grasping attempts and gripping, generating force feedback data.

[0057] The multimodal sensing system connects to an industrial edge computing gateway via an Ethernet or USB interface to achieve synchronous data acquisition and initial timestamp alignment.

[0058] Step S200: The perceived data is transmitted to a pre-trained multimodal fusion model for analysis and processing to generate grasping decision information including grasping space point coordinates, grasping posture and target grasping force.

[0059] It should be noted that industrial edge computing gateways can be used for local data processing and model inference, for real-time preprocessing of sensed data, and for local deployment of multimodal fusion models.

[0060] The industrial edge computing gateway first performs filtering, noise reduction, downsampling and compression on the collected 3D point cloud data, and performs enhancement and segmentation preprocessing on the surface image data. At the same time, it aligns the multi-source data in the time and space dimensions to ensure that the perceptual information input into the multimodal fusion model has consistency and integrity.

[0061] The multimodal fusion model can be updated using a small-sample incremental learning approach. Specifically, when laser cutting of a new type of raw material is required, a small amount of sample data of the new raw material and successfully captured labels are collected. The multimodal fusion model is then fine-tuned based on a meta-learning framework, and the fine-tuned multimodal fusion model parameters are deployed to industrial edge computing devices.

[0062] In this way, by adapting to the geometric features and physical properties of new raw materials, the time and computational resource consumption caused by fully retraining the multimodal fusion model can be avoided. Through the task adaptation mechanism in the meta-learning framework, the multimodal fusion model can extract key features from a small number of samples and construct a grasping strategy mapping for new raw materials. At the same time, the fine-tuned parameter package is stored in a pre-stored parameter library. When encountering similar raw materials again, the parameter package can be directly called, further improving response speed and flexible production capabilities. In addition, since incremental learning with small samples retains the multimodal fusion model's grasping decision-making ability for the original raw materials, it can ensure compatibility and stability in multi-variety mixed-line production scenarios, improving the applicability of the laser cutting machine's feeding process.

[0063] Alternatively, the multimodal fusion model can be trained through the following steps:

[0064] A training dataset is constructed, which contains perception data of various irregularly shaped raw materials under different stacking states, as well as corresponding standard grasping points and grasping force labels. With the goal of minimizing the error between the predicted grasping information and the grasping force labels, the neural network is trained in an end-to-end manner. During the training process, a contrastive learning loss function is used to optimize the alignment of different modal feature spaces.

[0065] In this way, the model can effectively mine the correlation between the geometric structure information of the 3D point cloud, the texture features of the surface image, and the physical interaction law of the force feedback data. Thus, when faced with unseen irregular raw materials or complex stacking scenarios, it can still quickly and accurately generate robust grasping decision information, providing a reliable basis for the subsequent precise grasping and loading actions of the robotic arm, and improving the adaptability of the laser cutting machine's loading control system to the needs of flexible production and the loading success rate.

[0066] In this embodiment, the AI ​​processing module of the industrial edge computing gateway is loaded with a pre-trained multimodal fusion model. The multimodal fusion model takes the collected 3D point cloud data, surface image data, and force feedback data as input. The multimodal fusion model first extracts features from the input multi-source data. Then, the multimodal fusion model comprehensively analyzes these features and finally outputs grasping decision information. This grasping decision information is a data structure that includes at least the coordinates of the grasping spatial point, the grasping posture, and the target grasping force. The coordinates of the grasping spatial point refer to a 3D coordinate (X, Y, Z) in the coordinate system of the point cloud of the raw material to be loaded, representing the recommended optimal grasping contact point; the grasping posture is a quaternion or rotation matrix, representing the posture that the end effector should have when it reaches the grasping point (such as the opening direction of the gripper); the target grasping force is a scalar value in Newtons (N), representing the initial normal grasping force that needs to be applied to achieve stable grasping.

[0067] Step S300: Based on the decision information, dynamically plan the motion path of the robotic arm and the force control parameters of the end effector to form a feeding control command.

[0068] It should be noted that the industrial edge computing gateway includes the MoveIt framework based on ROS (Robot Operating System). The MoveIt framework first uses the coordinates of the grasping space points and the grasping posture from the decision information as the endpoint constraints of the path. Then, it combines 3D point cloud data as obstacle detection information and calls algorithms such as RRT* (Rapidly-exploring Random Tree Star) to search for a collision-free motion trajectory from the current position to the target grasping position in the robot arm's configuration space. This trajectory includes a series of joint angles and end-effector pose points.

[0069] Simultaneously, based on the target gripping force and the pre-input physical properties of the raw materials (such as material density and estimated coefficient of friction), a set of force control parameters is set, including an initial gripping force threshold (e.g., 110% of the target gripping force as the upper limit) and a force control stiffness parameter (which determines the proportional relationship between position deviation and correction force). Finally, the planned motion trajectory sequence and the force control parameter set are packaged to form the final feeding control instruction.

[0070] Step S400: Based on the feeding control command, control the robotic arm and end effector to perform the grasping and feeding actions of raw materials, and adjust the force control parameters in a closed loop according to real-time force feedback data during the grasping process.

[0071] In one feasible embodiment, step S400 may include steps S410 to S440:

[0072] Step S410: Based on the feeding control command, control the robotic arm to move along the planned motion trajectory, and switch to force-position hybrid control mode when the end effector contacts the surface of the raw material;

[0073] Step S420: In the force-position hybrid control mode, with the goal of maintaining the target grasping force, the position and / or output torque of the end effector are dynamically adjusted according to the real-time force feedback data.

[0074] Step S430: When the real-time force feedback data is stable within the preset successful grasping force range and continues for more than the first set time, the grasping is determined to be successful, and the robotic arm is controlled to perform subsequent loading movement.

[0075] Step S440: If a sudden change or slippage feature is detected in the real-time force feedback data during the grasping process, the grasping failure process is triggered.

[0076] It should be noted that dynamically adjusting the position and / or output torque of the end effector can be achieved through impedance control or admittance control algorithms. By dynamically adjusting the position and / or output torque of the end effector, compensation can be made for visual positioning deviations, material deformation, or slippage. After triggering the gripping failure process, steps S100 to S300 above are re-executed to generate new loading control instructions.

[0077] In this embodiment, the robotic arm first approaches the target along a planned motion trajectory. When the contact switch or force sensor Z-axis force value on the end effector (such as a pneumatic parallel gripper) exceeds a small threshold (e.g., 4N), it is determined that contact has been made with the material surface, and the system switches from pure position control mode to force-position hybrid control mode. In force-position hybrid control mode, the six-dimensional force sensor continuously provides feedback on the actual gripping force. If the actual gripping force is less than the target gripping force, the gripper is controlled to close further or the end effector to advance further along the gripping direction (position adjustment). If the actual gripping force is close to or exceeds the gripping force threshold, the gripper is controlled to release further or retract. This closed-loop adjustment compensates for errors in visual positioning, unevenness of the material surface, or slippage tendencies.

[0078] When the gripping force fed back by the force sensor stabilizes within the range of "expected gripping force ±10%" and remains stable for more than a first set duration (e.g., 200 milliseconds), the gripping is considered successful. The robotic arm is then controlled to lift the workpiece and move it to the designated position on the laser cutting machine's worktable according to a preset path, completing the loading process. If the force sensor detects a sudden disappearance of force (dropping) or a violent fluctuation in lateral force (slippage), the gripping failure process is triggered. The robotic arm is then controlled to stop its current action, return to a safe position, and re-trigger the entire process starting from step S100 for a second gripping attempt.

[0079] For example, when switching to the force-position hybrid control mode, an admittance control algorithm is used as the specific implementation method. The force error detected by the force sensor (actual gripping force F_actual - target gripping force F_expected) is converted into a position correction amount ΔX through a virtual admittance model (mass-damping-spring system). ΔX is superimposed on the originally planned position command, thereby allowing the end-effector position to be dynamically adjusted according to the actual contact situation while maintaining a constant expected force. For example, when the actual force is smaller, a positive ΔX is generated, causing the end-effector to slightly "press" into the material until the force reaches equilibrium. Assume that the successful gripping force range is set to [0.9F_expected, 1.1F_expected]. The first set duration is adjustable according to the material characteristics; it can be set shorter for rough materials (e.g., 150ms) and longer for smooth and slippery surfaces (e.g., 300ms). Slip characteristics are determined by monitoring the variance of the lateral forces (Fx, Fy) fed back by the force sensor. If the variance of the lateral force exceeds the threshold for a short period of time during the gripping phase, it is determined that slippage may have occurred, triggering a failure process.

[0080] Thus, by switching control modes during contact, a smooth transition from non-contact position guidance to contact force control adjustment is achieved, ensuring the smoothness of the action when the end effector contacts the raw material surface. Through closed-loop adjustment of force control parameters driven by real-time force feedback, deviations caused by visual positioning errors, uneven raw material surfaces, or material deformation are effectively compensated for, maintaining a stable gripping force to prevent workpiece drop or damage. By setting the force stability threshold and duration conditions for successful gripping, the gripping status is accurately determined, avoiding workpiece drop during the feeding process due to misjudgment. At the same time, by quickly responding to abnormal characteristics of gripping failure and triggering a retry process, the robustness and automated continuous operation capability of the laser cutting machine's feeding process are improved, ensuring the efficiency and reliability of feeding multiple types of raw materials in flexible production scenarios.

[0081] In the technical solution provided in this embodiment, by simultaneously acquiring three-dimensional point cloud, surface image and force feedback data, multi-dimensional perception information about the geometry, surface texture and physical interaction characteristics of raw materials is constructed. This avoids the limitation of information loss when using a traditional single vision sensor when facing irregularly shaped materials with reflective, complex texture or severe occlusion. It provides a comprehensive and robust data foundation for subsequent decision-making.

[0082] By comprehensively analyzing multimodal perception data through a pre-trained multimodal fusion model, and based on learning the intrinsic correlation between different modal information (for example, associating geometric protrusions in point clouds with bright reflective areas in images to determine whether they are valid grasping points), it can still infer and generate suitable grasping points, postures, and target grasping forces when faced with irregularly shaped raw materials or non-standard stacking states that have not been pre-modeled, thereby improving the decision-making adaptability to uncertain objects.

[0083] By dynamically planning force control parameters while generating motion trajectories, the system considers subsequent force interaction requirements when generating motion paths and sets reasonable initial parameters for the force control closed loop (such as gripping force thresholds based on material properties). Through the "force-position coordination" design in the planning stage, the system avoids the connection conflicts or parameter mismatch problems that may be caused by the separation of motion planning and force control modules in traditional methods, thereby improving the coherence and reliability of the overall motion sequence of the robotic arm.

[0084] By introducing a closed-loop adjustment based on real-time force feedback during the grasping process, interference caused by visual positioning errors, local material deformation, or slight slippage can be compensated in real time. The online correction mechanism enables the grasping action to be adaptively adjusted according to the actual physical contact situation, thereby improving the grasping success rate and stability under non-ideal contact conditions.

[0085] In summary, the embodiments of this application acquire comprehensive state information of the raw materials to be fed through multimodal perception, generate accurate grasping decisions by combining a pre-trained multimodal fusion model, and then achieve compliant grasping and stable feeding of the raw materials by the robotic arm through dynamic path planning and force control parameter adjustment. At the same time, the closed-loop adjustment mechanism of real-time force feedback effectively compensates for uncertainties such as visual positioning errors and material deformation, improves the robustness and adaptability of the feeding process, and ultimately achieves automated operation of the feeding process of the laser cutting machine, reduces manual intervention, improves production efficiency, reduces the risk of raw material damage, and meets the needs of multi-variety and flexible production.

[0086] Reference Figure 2 Based on the first embodiment of this application, in the second embodiment of this application, the 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. Based on this, the pre-trained multimodal fusion model is a neural network based on the Transformer architecture; step S200 may include steps S211 to S213:

[0087] Step S211: Feature encoding is performed on the three-dimensional point cloud data, surface image data, and force feedback data respectively;

[0088] Step S212: Perform interactive and alignment operations on the encoded features to generate unified multimodal fusion features;

[0089] Step S213: Pass the multimodal fusion features to the regression and classification network head and output the grasping decision information.

[0090] It should be noted that interaction and alignment operations can be achieved through a cross-modal attention mechanism.

[0091] When performing feature encoding, for 3D point cloud data, a PointNet++ network is used as the encoder to convert the disordered point cloud into a fixed-dimensional global feature vector F_3d; for surface image data (RGB images), a lightweight CNN (Convolutional Neural Network) is used to extract image features, and then global average pooling is used to obtain the feature vector F_rgb; for force feedback data, the force pattern of the "successful capture moment" in historical data is used as a priori. During actual inference, the input of this path is temporarily a preset zero vector or mean vector, mainly used for feature alignment during model training, and the output feature vector is F_force.

[0092] In this embodiment, the three feature vectors F_3d, F_rgb, and F_force are concatenated and input into a multi-layer Transformer encoder. The Transformer's self-attention mechanism learns the association weights between the three modal features. For example, image texture features can help interpret whether a depression in the point cloud is a scratch or a shadow, thereby achieving deep fusion of multi-source information and outputting a unified multimodal fusion feature vector F_fused.

[0093] F_fused is passed to the regression network head and the classification network head. The regression network head and the classification network head are two parallel network heads. The regression network head is a fully connected layer that directly outputs the coordinates of the grasp point (x, y, z) and the grasp pose (quaternions q1, q2, q3, q4). The classification network head is a fully connected layer connected to Softmax, which outputs a discrete "target grasp force level" (e.g., high, medium, low), which is mapped to a specific force value (e.g., 200N, 100N, 50N).

[0094] In this way, targeted modal feature encoding ensures that core information such as the geometric structure of 3D point clouds, the texture details of surface images, and the physical interaction characteristics of force feedback are accurately extracted. By leveraging the cross-modal attention mechanism of the Transformer architecture, deep interaction and global alignment of multi-source features are achieved, effectively eliminating information barriers and semantic biases between different modalities. Finally, through the collaborative computation of regression and classification network heads, grasping decision information with spatial accuracy, posture rationality, and force control adaptability is output, so that the multimodal fusion model can still stably output robust decision results when facing irregular raw materials or complex stacking scenarios, providing reliable support for the precise grasping and compliant feeding of robotic arms.

[0095] Furthermore, step S200 may also include identifying surface defects of the raw material in the surface image data; if the surface defects of the raw material exist, when generating the grasping decision information, planning the grasping posture to avoid the defect area.

[0096] Specifically, the collected surface image data (RGB image) is input into a pre-trained semantic segmentation model. This model is trained to identify common defects on the raw material surface, such as "scratches," "rust," and "oil stains," and generates corresponding defect region masks on the image. When generating grasping decision information, the model checks whether the position of the recommended grasping point projected onto the RGB image falls within any defect mask area. If it does, a correction algorithm is activated. In the point cloud, with the original recommended point as the center, a suboptimal alternative point that avoids the defect projection area is searched within a spherical space of a certain radius (e.g., 20 mm) as the new grasping point. The grasping posture is also recalculated accordingly.

[0097] This avoids contact between the end effector and defective areas, preventing slippage caused by abnormal friction coefficients in defective areas (such as reduced friction coefficients due to oil contamination) or breakage during gripping due to weaker structures in scratched or corroded areas. This ensures the physical integrity of the raw material and prevents damage from affecting the processing accuracy and finished product quality of subsequent laser cutting. Furthermore, by optimizing the gripping strategy to actively avoid defective areas, the multimodal fusion model's adaptability to irregularly shaped and defective raw materials is further enhanced. This strengthens the robustness of the feeding process and its adaptability to flexible production scenarios, ensuring stable completion of the feeding task even with raw materials in less-than-ideal conditions.

[0098] Reference Figure 3 Based on the above embodiments of this application, in the third embodiment of this application, the same or similar content as the above embodiments can be referred to the above description, and will not be repeated hereafter. Based on this, step S3 may include steps S310 to S330:

[0099] Step S310: The coordinates of the grasping space points and the grasping posture in the grasping decision information are used as path endpoint constraints, and the three-dimensional point cloud data is transmitted to the motion planning algorithm to generate a collision-free motion trajectory from the current posture of the robotic arm to the path endpoint in three-dimensional space.

[0100] Step S320: Determine the initial gripping force threshold and force control stiffness parameters of the end effector based on the target gripping force and the physical property model of the raw materials.

[0101] Step S330: Encapsulate the collision-free motion trajectory, the initial gripping force threshold, and the force control stiffness parameters to form the feeding control command.

[0102] It should be noted that motion planning algorithms are algorithms that optimize trajectories based on sampling. Their optimization objectives include at least one of the following: shortest path, optimal time, smooth motion, and minimum joint torque.

[0103] In this embodiment, the RRTConnect algorithm from OMPL (Open Motion Planning Library) can be used as the motion planning algorithm. A preliminary collision-free motion trajectory is generated by randomly sampling in the joint space of the robotic arm and constructing two search trees (one from the starting point and one from the ending point). Subsequently, a trajectory optimization algorithm (such as the TrajOpt algorithm) is called to smooth the preliminary collision-free motion trajectory, with the optimization objective considering both the shortest trajectory time and smooth joint motion (minimizing acceleration jumps).

[0104] Once the raw material type is determined (e.g., 2mm 304 stainless steel plate), the corresponding density parameters, elastic modulus, etc., are retrieved from the pre-stored material database. Combined with the target gripping force F_expected, an empirical formula is used to calculate the gripping force threshold F_threshold (F_threshold = k * F_expected, where the coefficient k is typically 1.2~1.5) and the force-controlled stiffness parameter K (unit: N / m). The selection of the force-controlled stiffness parameter K is related to the material hardness; softer materials use a smaller K value to achieve a more compliant contact.

[0105] The loading control command is a structure that includes the optimized joint space trajectory point sequence, contact detection force threshold (e.g., 4N), target gripping force F_expected, force control stiffness K, and maximum allowable gripping force F_threshold.

[0106] In this way, the grasping decision information is transformed into structured motion instructions that the robotic arm can directly execute. This ensures that the robotic arm avoids obstacles such as raw material piles and equipment frames during its movement from the current pose to the target grasping pose through collision-free motion trajectory planning, guaranteeing motion safety and optimal path (such as shortest path and smooth motion). Furthermore, by combining the material's physical properties with the initial force control parameters determined by the target grasping force, reasonable benchmark thresholds and stiffness settings are provided for subsequent contact force feedback closed-loop adjustments, avoiding force impact or material damage at the moment of grasping due to initial parameter mismatch. Finally, through unified encapsulation of instructions, the coordinated linkage between trajectory motion and force control adjustment is achieved, ensuring that the robotic arm can accurately follow the preset path to reach the target position when performing the loading action, and can quickly enter a stable force control state after contacting the material. This provides instruction-level support for the compliance, stability, and efficiency of the entire loading process, achieving seamless connection from perception and decision-making to action execution, and improving the overall performance of the laser cutting machine's loading control system.

[0107] Based on the above embodiments of this application, in the fourth embodiment of this application, the same or similar content as the above embodiments can be referred to the above description, and will not be repeated hereafter. On this basis, before step S2, it may also include: responding to the instruction to change the end effector or the instruction to change the target raw material type, and calling the multimodal fusion model parameters and force control parameter baseline that match the current hardware configuration and raw material type from the pre-stored parameter package.

[0108] In this embodiment, the operator can select the "end effector model" (e.g., "wide-mouth planar gripper - model A") and the "target raw material type" (e.g., "aluminum alloy profile - section B") via the user interface. The system then generates a hardware configuration and material type combination key (e.g., "Gripper_A_Material_B"). This key is searched in a pre-stored parameter package database. The database stores a dedicated configuration file for each combination, containing multimodal fusion model parameters (i.e., a fine-tuned model weight file) optimized for the gripper's field of view and the material's reflectivity. It also includes a baseline of force control parameters (including initial suggested values ​​for F_expected, K, and F_threshold) set for the gripper's force sensor range and the material's hardness and coefficient of friction. The system automatically loads this configuration file, completing the parameter switching without reprogramming or lengthy manual calibration, thus shortening production preparation time.

[0109] In this way, by adapting to dynamic changes in hardware and raw materials, model decision-making biases or force control parameter inaccuracies caused by differences in end-effector structure (such as suction cup versus gripper type) or changes in the physical properties of raw materials (such as thickness, hardness, and surface friction coefficient) are avoided. The quick recall of pre-stored parameter packages eliminates the tedious process of retraining the model or manually adjusting parameters, shortening changeover time and meeting the needs of flexible production for multi-variety, high-frequency material changes. Simultaneously, it ensures that even after hardware or material changes, the multimodal fusion model can still accurately infer and acquire the grasping strategy based on the matched parameters, and the force control system can quickly enter a stable adjustment state based on baseline parameters, maintaining the compliance and reliability of the feeding process, further improving the automation level and production adaptability of the laser cutting machine's feeding system.

[0110] Based on the above embodiments of this application, in the fifth embodiment of this application, the same or similar content as the above embodiments can be referred to the above description, and will not be repeated hereafter. On this basis, after step S4, it further includes: continuously collecting the joint vibration spectrum and drive current data of the robotic arm during the operation of the robotic arm; transmitting the joint vibration spectrum and drive current data to the health prediction model to determine the health status of the robotic arm, and issuing a maintenance prompt when the health status reaches the health warning threshold.

[0111] In this embodiment, the drive current data of the servo motors of each joint can be continuously collected through the built-in IO module of the robotic arm controller. Simultaneously, vibration sensors (such as IEPE accelerometers) are installed on the robotic arm base or key parts of the upper arm to collect joint vibration spectrum data. Current and vibration data over a period of time (e.g., the last 8 hours) are input into a health prediction model (e.g., using a one-dimensional convolutional neural network) to extract features from the vibration spectrum, which are then fused with time-domain features such as the effective value of the current. This health prediction model outputs a "health index" or a specific fault type probability (e.g., "joint reducer wear probability" or "motor bearing failure probability"). When the fault probability value output by the model exceeds a preset health warning threshold (e.g., 60%), a clear maintenance prompt is issued to maintenance personnel via an audible and visual alarm, the workshop MES system, and / or a mobile app, suggesting an inspection during the next planned downtime period, thereby achieving an upgrade from scheduled maintenance to predictive maintenance.

[0112] Thus, this embodiment of the application identifies potential faults such as wear on the joint bearings of the robotic arm, aging of the drive motor, and loosening of the transmission mechanism in advance, avoiding interruptions in the feeding process due to sudden failures of the robotic arm and ensuring the continuous and stable operation of the laser cutting machine's feeding stage. Real-time monitoring and early warning of the robotic arm's health status enables preventative maintenance, replacing the traditional periodic maintenance model, reducing unnecessary downtime and lowering maintenance costs. Simultaneously, it ensures that the robotic arm performs gripping and feeding actions in a healthy state, avoiding decreased gripping accuracy or damage to raw materials caused by component performance degradation, further consolidating the reliability of the feeding process and the stability of production quality, providing equipment-level assurance for the efficient and continuous operation of the entire laser cutting production line.

[0113] This application also provides a laser cutting machine feeding control system. Please refer to... Figure 4 The laser cutting machine feeding control system includes:

[0114] The multimodal sensing module 10 is used to collect sensing data of the raw materials to be fed in real time. The sensing data includes at least three-dimensional point cloud data, surface image data and force feedback data during the grasping process.

[0115] Edge computing module 20 is used to transmit the perceived data to a pre-trained multimodal fusion model for analysis and processing, and generate grasping decision information including grasping space point coordinates, grasping posture and target grasping force;

[0116] The planning and control module 30 is used to dynamically plan the motion path of the robotic arm and the force control parameters of the end effector based on the decision information, and to form a feeding control command.

[0117] The feeding execution module 40 is used to control the robotic arm and the end effector based on the feeding control command to perform the grasping and feeding actions of raw materials, and to make closed-loop adjustments to the force control parameters based on real-time force feedback data during the grasping process.

[0118] In addition, the laser cutting machine feeding control system also includes a model parameter matching module, which is used to respond to the command to change the end effector or the command to change the target raw material type. It calls the multimodal fusion model parameters and force control parameter baseline that match the current hardware configuration and raw material type from the pre-stored parameter package.

[0119] In addition, a health detection module is used to continuously collect the joint vibration spectrum and drive current data of the robotic arm during its operation; transmit the joint vibration spectrum and drive current data to a health prediction model to determine the health status of the robotic arm, and issue a maintenance prompt when the health status reaches a health warning threshold.

[0120] The laser cutting machine feeding control system provided in this application adopts the laser cutting machine feeding control method in the above embodiments, which can solve the technical problems of poor adaptability and low grasping success rate of traditional laser cutting machine feeding methods due to reliance on preset templates and single sensors for raw materials with irregular shapes and random stacking states. Compared with the prior art, the beneficial effects of the laser cutting machine feeding control system provided in this application are the same as those of the laser cutting machine feeding control method provided in the above embodiments, and other technical features in the laser cutting machine feeding control system are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0121] This application provides an electronic 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, the instructions being executed by the at least one processor to enable the at least one processor to perform the laser cutting machine feeding control method in the above embodiments.

[0122] The following is for reference. Figure 5 The diagram illustrates a structural schematic of an electronic device suitable for implementing embodiments of this application. The electronic devices in these embodiments may include, but are not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 5The electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0123] like Figure 5 As shown, the electronic 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 electronic 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, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. The communication device 1009 allows the electronic device to communicate wirelessly or wiredly with other devices to exchange data. Although the diagrams show electronic devices with various systems, it should be understood that it is not required to implement or have all of the systems shown. More or fewer systems may be implemented alternatively.

[0124] 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.

[0125] The electronic device provided in this application adopts the laser cutting machine feeding control method in the above embodiments. Compared with the prior art, the beneficial effects of the electronic device provided in this application are the same as those of the laser cutting machine feeding control method provided in the above embodiments. Furthermore, the other technical features of the electronic device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0126] 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.

[0127] 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.

[0128] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the laser cutting machine feeding control method in the above embodiments.

[0129] 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 fibers, 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.

[0130] The aforementioned computer-readable storage medium may be included in an electronic device or may exist independently without being assembled into an electronic device.

[0131] The aforementioned computer-readable storage medium carries one or more programs. When these programs are executed by an electronic device, the electronic device causes the following: It collects real-time sensing data of the raw material to be loaded, the sensing data including at least three-dimensional point cloud data, surface image data, and force feedback data during the grasping process; it transmits the sensing data to a pre-trained multimodal fusion model for analysis and processing, generating grasping decision information including grasping space point coordinates, grasping posture, and target grasping force; based on the decision information, it dynamically plans the motion path of the robotic arm and the force control parameters of the end effector to form a loading control command; and based on the loading control command, it controls the robotic arm and the end effector to perform the grasping and loading actions of the raw material, and adjusts the force control parameters in a closed loop according to the real-time force feedback data during the grasping process.

[0132] 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).

[0133] 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.

[0134] 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.

[0135] The readable storage medium provided in this application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., a computer program) for executing the above-described laser cutting machine feeding control method. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as the beneficial effects of the laser cutting machine feeding control method provided in the above embodiments, and will not be repeated here.

[0136] This application provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the laser cutting machine feeding control method described above.

[0137] The computer program product provided in this application can solve the technical problems of poor adaptability and low grasping success rate of traditional laser cutting machine feeding methods, which rely on preset templates and single sensors and have poor adaptability to irregularly shaped and randomly stacked raw materials. 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 laser cutting machine feeding control method provided in the above embodiments, and will not be repeated here.

[0138] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent scope of this application.

Claims

1. A method for controlling the feeding of a laser cutting machine, characterized in that, The laser cutting machine feeding control method includes: Real-time acquisition of sensing data of raw materials to be loaded, including at least three-dimensional point cloud data, surface image data and force feedback data during the grasping process; The perceived data is transmitted to a pre-trained multimodal fusion model for analysis and processing to generate grasping decision information including grasping space point coordinates, grasping posture and target grasping force; Based on the decision information, the motion path of the robotic arm and the force control parameters of the end effector are dynamically planned to form a feeding control command; Based on the feeding control command, the robotic arm and end effector are controlled to perform the grasping and feeding actions of raw materials, and the force control parameters are adjusted in a closed loop according to real-time force feedback data during the grasping process.

2. The laser cutting machine feeding control method as described in claim 1, characterized in that, The pre-trained multimodal fusion model is a neural network based on the Transformer architecture; the step of transmitting the perception data to the pre-trained multimodal fusion model for analysis and processing to generate grasping decision information including grasping space point coordinates, grasping posture, and pre-target grasping force includes: The three-dimensional point cloud data, surface image data, and force feedback data are respectively feature-encoded; Interactive and alignment operations are performed on the encoded features to generate unified multimodal fusion features; The multimodal fusion features are passed to the regression and classification network head, and the grasping decision information is output.

3. The laser cutting machine feeding control method as described in claim 1, characterized in that, The step of transmitting the perceived data to a pre-trained multimodal fusion model for analysis and processing, and generating grasping decision information including grasping space point coordinates, grasping posture, and target grasping force, further includes: Identify surface defects of the raw materials in the surface image data; If surface defects exist in the raw material, the grasping posture is planned to avoid the defective area when generating the grasping decision information.

4. The laser cutting machine feeding control method as described in claim 1, characterized in that, The step of dynamically planning the motion path of the robotic arm and the force control parameters of the end effector based on the decision information to form a feeding control command includes: The grasping space point coordinates and grasping posture in the grasping decision information are used as path endpoint constraints, and the three-dimensional point cloud data is transmitted to the motion planning algorithm to generate a collision-free motion trajectory from the current pose of the robotic arm to the path endpoint in three-dimensional space. Based on the target grasping force and the physical property model of the raw materials, the initial grasping force threshold and force control stiffness parameters of the end effector are determined. The collision-free motion trajectory, the initial gripping force threshold, and the force control stiffness parameters are encapsulated to form the feeding control command.

5. The laser cutting machine feeding control method as described in claim 1, characterized in that, The step of controlling the robotic arm and end effector based on the feeding control command to perform the grasping and feeding actions of raw materials, and adjusting the force control parameters in a closed loop based on real-time force feedback data during the grasping process includes: Based on the feeding control command, the robotic arm is controlled to move along the planned motion trajectory, and when the end effector contacts the surface of the raw material, it switches to a force-position hybrid control mode. In the force-position hybrid control mode, with the goal of maintaining the target grasping force, the position and / or output torque of the end effector are dynamically adjusted according to the real-time force feedback data; When the real-time force feedback data is stable within the preset successful grasping force range and continues for more than the first set time, the grasping is determined to be successful, and the robotic arm is controlled to perform subsequent loading movement. If a sudden change or slippage is detected in the real-time force feedback data during the grasping process, the grasping failure process is triggered.

6. The laser cutting machine feeding control method as described in claim 1, characterized in that, Before the step of transmitting the perceived data to a pre-trained multimodal fusion model for analysis and processing to generate grasping decision information including grasping space point coordinates, grasping posture, and target grasping force, the following steps are also included: In response to a command to change the end effector or a command to change the target raw material type, the system retrieves the multimodal fusion model parameters and force control parameter baselines that match the current hardware configuration and raw material type from the pre-stored parameter package.

7. The laser cutting machine feeding control method as described in claim 1, characterized in that, After the step of controlling the robotic arm and end effector based on the feeding control command to perform the grasping and feeding actions of raw materials, and adjusting the force control parameters in a closed loop based on real-time force feedback data during the grasping process, the method further includes: During the operation of the robotic arm, the joint vibration spectrum and drive current data of the robotic arm are continuously collected; The joint vibration spectrum and drive current data are transmitted to the health prediction model to determine the health status of the robotic arm, and a maintenance prompt is issued when the health status reaches the health warning threshold.

8. A material feeding control system for a laser cutting machine, characterized in that, The laser cutting machine feeding control system includes: A multimodal sensing module is used to collect sensing data of the raw materials to be fed in real time. The sensing data includes at least three-dimensional point cloud data, surface image data, and force feedback data during the grasping process. The edge computing module is used to transmit the perceived data to a pre-trained multimodal fusion model for analysis and processing, and generate grasping decision information including grasping space point coordinates, grasping posture and target grasping force; The planning and control module is used to dynamically plan the motion path of the robotic arm and the force control parameters of the end effector based on the decision information, and to generate feeding control commands. The loading execution module is used to control the robotic arm and end effector based on the loading control command to perform the grasping and loading actions of raw materials, and to make closed-loop adjustments to the force control parameters based on real-time force feedback data during the grasping process.

9. An electronic device, 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 laser cutting machine feeding control method as described in any one of claims 1 to 7.

10. A readable storage medium, characterized in that, The readable storage medium is a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the laser cutting machine feeding control method as described in any one of claims 1 to 7.

Images