Multi-robot path planning method and system for magnetic bottle cap assembly line

Abstract

The application provides a multi-robot path dynamic programming method and system for a magnetic bottle cap assembly line, and belongs to the technical field of automatic packaging. The method can divide two types of coupled magnetic force regions, repulsion of same poles and attraction of different poles, by the magnetic induction intensity of the target magnetic bottle cap combined with the physical profile of the robot body; at the same time, the interference type between adjacent robots is dynamically determined by combining the real-time motion trajectory coordinates of the robot, and the corresponding force vector matrix and gravitational potential field gradient are constructed for repulsion interference and attraction interference respectively to generate an obstacle avoidance path with differentiated matching magnetic field characteristics. The method brings the invisible magnetic force range into the constraint condition of the exclusive working space of the robot in advance, thereby avoiding the risk of entering the magnetic force region by mistake and improving the accuracy and running safety of path planning during high-speed collaborative operation of multiple robots.

Description

Technical Field

[0001] This application relates to the field of automated packaging technology, specifically to a multi-robot path dynamic planning method and system for magnetic bottle cap assembly lines. Background Technology

[0002] Currently, the collaborative operation of multiple robotic arms in magnetic bottle cap assembly lines is a core element in improving the production efficiency of these lines. Multiple robotic arms operate at high speeds within a limited physical space, and the rationality of their path planning directly determines the overall production capacity and safety of the line. However, current path planning methods typically define safety boundaries solely based on the physical outline of the robotic arms, treating them as regular geometric shapes for obstacle avoidance calculations. This conventional spatial division ignores the inherent magnetic properties of special components, leading to unexpected mutual repulsion or attraction when the robotic arms handle magnetic parts, thus affecting the accuracy of path planning.

[0003] The information described in the background section is only for enhancing the understanding of the background of this application, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0004] In view of this, this application provides a multi-robot path dynamic planning method and system for magnetic bottle cap assembly lines, which can improve the accuracy of path planning.

[0005] In a first aspect, embodiments of this application provide a multi-manipulator path dynamic planning method for a magnetic bottle cap assembly line. The method includes: acquiring the magnetic induction intensity of a target magnetic bottle cap and the physical contour data of each manipulator body; determining the magnetic pole vector of the target magnetic bottle cap based on the magnetic induction intensity, and determining the magnetic force range of each manipulator based on the magnetic pole vector and the physical contour data; determining the coupled magnetic force region for collaborative operation of each manipulator based on the magnetic force range, wherein the coupled magnetic force region includes a region of repulsion between like-named magnetic poles and a region of attraction between unlike-named magnetic poles; acquiring the motion trajectory coordinate sequence of the current manipulator; if any coordinate point in the trajectory coordinate sequence falls into the region of repulsion between like-named magnetic poles of an adjacent manipulator, then determining... There is repulsive interference between adjacent robotic arms, and the adjacent robotic arms are treated as repulsive obstacles. The coordinates of the repulsive obstacles are obtained, and the repulsive force vector matrix and the first gravitational potential field gradient are determined based on the coordinates of the repulsive obstacles. A first target obstacle avoidance path is determined based on the repulsive force vector matrix and the first gravitational potential field gradient. If the coordinate points in the trajectory coordinate sequence fall into the opposite magnetic pole attraction region of the adjacent robotic arm, it is determined that there is attractive interference between the adjacent robotic arms, and the adjacent robotic arms are treated as attractive obstacles. The coordinates of the attractive obstacles are obtained, and the attractive force vector matrix and the second gravitational potential field gradient are determined based on the coordinates of the attractive obstacles. A second target obstacle avoidance path is determined based on the attractive force vector matrix and the second gravitational potential field gradient.

[0006] Secondly, embodiments of this application provide a multi-robot path dynamic planning system for a magnetic bottle cap assembly line. This system includes: a first acquisition module, a first determination module, a second determination module, a second acquisition module, a third determination module, a first obstacle avoidance module, a fourth determination module, and a second obstacle avoidance module. The first acquisition module is used to acquire the magnetic induction intensity of the target magnetic bottle cap and the physical contour data of each robot body. The first determination module is used to determine the magnetic pole vector of the target magnetic bottle cap based on the magnetic induction intensity, and to determine the magnetic force range of each robot based on the magnetic pole vector and the physical contour data. The second determination module is used to determine the coupled magnetic force region for collaborative operation of each robot based on the magnetic force range, the coupled magnetic force region including a region of repulsion between like magnetic poles and a region of attraction between unlike magnetic poles. The second acquisition module is used to acquire the motion trajectory coordinate sequence of the current robot. The third determination module is used to determine that there is repulsive interference between adjacent robots if any coordinate point in the trajectory coordinate sequence falls into the region of repulsion between like magnetic poles of an adjacent robot. The system includes a first obstacle avoidance module, which acquires the coordinates of the repulsive obstacle, determines the repulsive force vector matrix and the first gravitational potential field gradient based on the coordinates of the repulsive obstacle, and determines the first target obstacle avoidance path based on the repulsive force vector matrix and the first gravitational potential field gradient; a fourth determination module, which determines that if the coordinates in the trajectory coordinate sequence fall into the opposite magnetic pole attraction region of the adjacent robot, there is attraction interference between the adjacent robot, and the adjacent robot is treated as an attraction obstacle; and a second obstacle avoidance module, which acquires the coordinates of the attraction obstacle, determines the attraction vector matrix and the second gravitational potential field gradient based on the coordinates of the attraction obstacle, and determines the second target obstacle avoidance path based on the attraction vector matrix and the second gravitational potential field gradient.

[0007] This application provides a method and system for dynamic path planning of multiple robotic arms in a magnetic bottle cap assembly line. This method quantifies the magnetic induction intensity and magnetic pole vector of the target magnetic bottle cap, and, combined with the physical contour of the robotic arm, accurately determines the asymmetrical magnetic force range during operation. It further divides the magnetic force into two types of coupled magnetic force regions: repulsion between like poles and attraction between unlike poles. Simultaneously, it dynamically determines the interference type between adjacent robotic arms based on the real-time motion trajectory coordinates of the robotic arms, constructing corresponding force vector matrices and gravitational potential field gradients for repulsive and attractive interference respectively, thereby generating differentiated obstacle avoidance paths that match the characteristics of the magnetic field. This method pre-defines the invisible magnetic force range into the constraints of the robotic arm's exclusive workspace, thus avoiding the risk of accidental entry into magnetic areas. This improves the accuracy and operational safety of path planning during high-speed collaborative operation of multiple robotic arms, effectively ensuring the capacity release and equipment operational stability of the magnetic bottle cap assembly line. Attached Figure Description

[0008] To more clearly illustrate the technical solutions in the embodiments of this application or the conventional technology, the drawings used in the description of the embodiments or the conventional technology will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0009] Figure 1 This is a flowchart illustrating a multi-robot path dynamic planning method for a magnetic bottle cap assembly line provided in an exemplary embodiment of this application.

[0010] Figure 2 This is a flowchart illustrating a multi-robot path dynamic planning method for a magnetic bottle cap assembly line provided in another exemplary embodiment of this application.

[0011] Figure 3 This is a flowchart illustrating a multi-robot path dynamic planning method for a magnetic bottle cap assembly line, provided in another exemplary embodiment of this application.

[0012] Figure 4 This is a flowchart illustrating a multi-robot path dynamic planning method for a magnetic bottle cap assembly line, provided in another exemplary embodiment of this application.

[0013] Figure 5 This is a flowchart illustrating a multi-robot path dynamic planning method for a magnetic bottle cap assembly line, provided in another exemplary embodiment of this application.

[0014] Figure 6 This is a flowchart illustrating a multi-robot path dynamic planning method for a magnetic bottle cap assembly line, provided in another exemplary embodiment of this application.

[0015] Figure 7 This is a flowchart illustrating a multi-robot path dynamic planning method for a magnetic bottle cap assembly line, provided in another exemplary embodiment of this application.

[0016] Figure 8 This is a flowchart illustrating a multi-robot path dynamic planning method for a magnetic bottle cap assembly line, provided in another exemplary embodiment of this application. Detailed Implementation

[0017] To enable those skilled in the art to better understand the technical solution, the present application will be described in detail below with reference to the embodiments. The description in this section is only exemplary and explanatory, and should not be used to limit the scope of protection of the present application in any way.

[0018] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that this application will be more comprehensive and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a full understanding of embodiments of this application.

[0019] The terms “a,” “one,” and “the” are used to indicate the existence of one or more elements / components / etc.; the terms “including” and “having” are used to indicate an open-ended inclusion and that other elements / components / etc. may exist in addition to those listed. The terms “first” and “second” are used only as markers and are not a limitation on the number of objects.

[0020] Currently, the collaborative operation of multiple robotic arms in magnetic bottle cap assembly lines is a core element in improving the production efficiency of these lines. Multiple robotic arms operate at high speeds within a limited physical space, and the rationality of their path planning directly determines the overall production capacity and safety of the line. However, current path planning methods typically define safety boundaries solely based on the physical outline of the robotic arms, treating them as regular geometric shapes for obstacle avoidance calculations. This conventional spatial division ignores the inherent magnetic properties of special components, leading to unexpected mutual repulsion or attraction when the robotic arms handle magnetic parts, thus affecting the accuracy of path planning.

[0021] Specifically, during the actual assembly of magnetic bottle caps, the polarity of the magnets inside the cap strictly restricts the operating trajectory of the robotic arm. The robotic arm must approach and place the cap along a specific magnetic field line. This specific approach route means that the robotic arm not only occupies its own physical volume but also generates an invisible magnetic field range in the magnetic attraction direction. If the safe distance is judged solely by the physical outline, other robotic arms can easily stray into this invisible magnetic field area. For example, when robotic arm one is preparing to vertically attract the cap from directly above, its magnetic field range has already extended outwards. If robotic arm two then passes by this area from the side according to the usual safe distance, although their metal shells do not touch, robotic arm two will be instantly attracted by the magnetic force of robotic arm one's working area, causing a serious deviation in the grasping trajectory or even a collision between the devices.

[0022] Therefore, how to incorporate the vector dimension of magnetic attraction direction into spatial calculations, and how to pre-define the asymmetric magnetic force range into the robot's exclusive workspace to avoid area intrusion when planning obstacle avoidance paths, has become the technical problem that this application needs to solve.

[0023] This application provides a multi-robot path dynamic planning method for magnetic bottle cap assembly lines, such as... Figure 1 The method shown is a dynamic programming approach for multiple robotic arms on a magnetic bottle cap assembly line. This method may include the following steps: Step S110: Obtain the magnetic induction intensity of the target magnetic bottle cap and the physical contour data of each robotic arm body; Step S120: Determine the magnetic pole vector of the target magnetic bottle cap based on the magnetic induction intensity, and determine the magnetic force range of each robot arm based on the magnetic pole vector and physical contour data; Step S130: Determine the coupled magnetic force region for the coordinated operation of each robot arm based on the range of magnetic force action. The coupled magnetic force region includes the region of repulsion between like magnetic poles and the region of attraction between unlike magnetic poles. Step S140: Obtain the current motion trajectory coordinate sequence of the robotic arm; Step S150: If any coordinate point in the trajectory coordinate sequence falls into the same magnetic pole repulsion region of the adjacent robot arm, it is determined that there is repulsion interference between the adjacent robot arms, and the adjacent robot arm is regarded as a repulsion obstacle. Step S160: Obtain the coordinates of the repulsive obstacle, determine the repulsive force vector matrix and the gradient of the first gravitational potential field based on the coordinates of the repulsive obstacle, and determine the first target obstacle avoidance path based on the repulsive force vector matrix and the gradient of the first gravitational potential field. Step S170: If the coordinate point in the trajectory coordinate sequence falls into the opposite magnetic pole attraction area of ​​the adjacent robot arm, it is determined that there is attraction interference between the adjacent robot arms, and the adjacent robot arm is regarded as an attraction obstacle. Step S180: Obtain the coordinates of the attracting obstacle, determine the attraction vector matrix and the second gravitational potential field gradient based on the coordinates of the attracting obstacle, and determine the second target obstacle avoidance path based on the attraction vector matrix and the second gravitational potential field gradient.

[0024] According to the multi-robot path dynamic planning method for magnetic bottle cap assembly lines provided in this application, this method can accurately determine the asymmetrical magnetic force range during operation by quantifying the magnetic induction intensity and magnetic pole vector of the target magnetic bottle cap and combining it with the physical contour of the robot body. It further divides the magnetic force into two types of coupled magnetic force regions: repulsion between like poles and attraction between unlike poles. Simultaneously, it dynamically determines the interference type between adjacent robots by combining the real-time motion trajectory coordinates of the robots, and constructs corresponding force vector matrices and gravitational potential field gradients for repulsive and attractive interference respectively, thereby generating differentiated obstacle avoidance paths that match the characteristics of the magnetic field. This method pre-includes the invisible magnetic force range in the constraints of the robot's exclusive workspace, thus avoiding the risk of accidental entry into magnetic areas, improving the accuracy of path planning and operational safety during high-speed collaborative operation of multiple robots, and effectively ensuring the capacity release and equipment operation stability of the magnetic bottle cap assembly line.

[0025] The following is a detailed description of each step of the multi-robot path dynamic planning method for magnetic bottle cap assembly lines provided in this application: In one embodiment of this application, step S110 involves acquiring the magnetic induction intensity of the target magnetic bottle cap and the physical contour data of each robotic arm body. Specifically, when the target magnetic bottle cap arrives at the inspection station via the conveyor line, the Hall sensor array and vision inspection unit above the station activate detection, collecting the magnetic induction intensity values ​​of the bottle cap magnet surface point by point, while the vision inspection unit is responsible for collecting the spatial pose and shape contour of the bottle cap. The arithmetic mean of the values ​​at multiple sampling points is calculated to obtain the average magnetic induction intensity of the bottle cap. At the same time, it is verified whether all sampling point values ​​are within the preset qualified range, and unqualified bottle caps with uneven magnetization or magnet misalignment are rejected. Finally, the average magnetic induction intensity and magnet center coordinate data of qualified bottle caps are bound and stored with the unique identification of the bottle cap. The factory physical contour parameters of each robotic arm participating in the assembly operation are retrieved from the equipment parameter library, including the length and diameter of each link, and the external dimensions and installation offset of the end magnetic clamp. By combining the real-time joint angle data of each robot arm, the spatial position of each component of the robot arm in the global base coordinate system under the current posture is calculated based on the forward kinematics formula, and a dynamic axis-aligned bounding box under the current posture is generated to obtain real-time and effective physical contour data.

[0026] For example, taking a perfume magnetic bottle cap assembly line as an example, this production line is equipped with three four-axis industrial robots, each responsible for one of the three processes: cap loading, magnetic bonding, and finished product unloading. The production cycle is 60 pieces per minute. A certain type of round magnetic bottle cap arrives at the inspection station along the conveyor line. A Hall sensor array is arranged in a 9-point layout to complete sampling. The magnetic induction intensity of the 9 sampling points is measured to be 102 millitas, 105 millitas, 98 millitas, 101 millitas, 107 millitas, 99 millitas, 103 millitas, 100 millitas, and 104 millitas, respectively. The average magnetic induction intensity is calculated to be 102 millitas. Next, the factory outline parameters of the three robotic arms are retrieved, and combined with the joint angles of the current standby posture, the spatial bounding boxes of each robotic arm are generated through forward kinematics calculations: the current outline bounding box range of the No. 1 loading robotic arm is 200 mm to 800 mm on the X-axis, 300 mm to 600 mm on the Y-axis, and 0 mm to 400 mm on the Z-axis; the bounding box range of the No. 2 assembly robotic arm is 1000 mm to 1600 mm on the X-axis, 300 mm to 600 mm on the Y-axis, and 0 mm to 400 mm on the Z-axis; the bounding box range of the No. 3 unloading robotic arm is 1800 mm to 2400 mm on the X-axis, 300 mm to 600 mm on the Y-axis, and 0 mm to 400 mm on the Z-axis.

[0027] In one embodiment of this application, step S120, determining the magnetic pole vector of the target magnetic bottle cap based on the magnetic induction intensity, further includes the following steps: Figure 2 As shown, the specific content is as follows: Step S210: If the magnetic induction intensity of the target magnetic bottle cap exceeds the first preset magnetic induction intensity threshold, then based on the magnetic induction intensity of the target magnetic bottle cap, a support vector machine is used to determine the magnetic pole surface distribution matrix. Step S220: Determine the set of magnetic field line direction angles based on the magnetic pole surface distribution matrix; Step S230: If the magnetic field line direction angle in the magnetic field line direction angle set does not exceed the preset direction angle threshold, then remove the current magnetic field line direction angle to update the magnetic field line direction angle set; Step S240: Determine the magnetic pole vector of the target magnetic bottle cap based on the updated set of magnetic field line direction angles.

[0028] Specifically, if the magnetic induction intensity of the target magnetic bottle cap does not exceed the first preset magnetic induction intensity threshold, it indicates that the bottle cap is insufficiently magnetized or the magnetic field signal signal-to-noise ratio is insufficient, and the sample is directly deemed invalid and discarded. If the magnetic induction intensity of the target magnetic bottle cap exceeds the first preset magnetic induction intensity threshold, it indicates that the magnetic field signal meets the recognition requirements, and the magnetic induction intensity data of all sampling points are input into the pre-trained support vector machine model. The support vector machine model classifies and identifies the magnetic polarity of each sampling point, and outputs the normalized magnetic induction intensity value at the corresponding position. Arranged according to the spatial row and column order of the sampling array, a magnetic pole surface distribution matrix corresponding to the sampling layout is generated. Each element of the magnetic pole surface distribution matrix contains a polarity identifier and an intensity value, representing the magnetic pole distribution state on the surface of the target magnetic bottle cap. The setting of the first preset magnetic induction intensity threshold needs to match the lower limit of the magnetic cap's magnetization qualification. Bottle caps below the first preset magnetic induction intensity threshold are inherently unqualified and do not require further magnetic pole identification.

[0029] Furthermore, the generated magnetic pole distribution matrix is ​​read, and the spatial magnetic field vector of each sampling point is calculated using the magnetic induction intensity gradient, combined with the spatial coordinates corresponding to the matrix elements. The spatial magnetic field vector is projected onto the tangential plane of the bottle cap end face, and the inclination angle of the tangential component relative to the normal of the end face is calculated. This inclination angle is the magnetic field line direction angle of the corresponding sampling point. The magnetic field line direction angle is the inclination angle of the spatial magnetic field line relative to the normal of the bottle cap end face; a larger value indicates a higher degree of divergence of the magnetic field lines and a more significant magnetic pole position characteristic. All valid sampling points in the magnetic pole distribution matrix are traversed, and the corresponding magnetic field line direction angle is calculated for each. All calculated magnetic field line direction angles are summarized in order of sampling point position to form a set of magnetic field line direction angles, where each value corresponds to a unique spatial coordinate of the sampling point.

[0030] Furthermore, the magnetic field line direction angle values ​​in the set are iterated through one by one, and each current magnetic field line direction angle is compared with a preset direction angle threshold. If the current magnetic field line direction angle does not exceed the preset direction angle threshold, it means that the corresponding sampling point is located in the uniform magnetic field region at the center of the magnet, or is a weak feature point generated by environmental noise, and does not have the feature value for magnetic pole positioning. In this case, the current magnetic field line direction angle is removed from the set. If the current magnetic field line direction angle exceeds the preset direction angle threshold, it means that the corresponding sampling point is located in the divergent region at the edge of the magnet, and the magnetic pole position feature is significant. In this case, the current magnetic field line direction angle needs to be retained. After all iterations are completed, an updated set of magnetic field line direction angles containing only valid feature points is obtained. Among them, the setting of the preset direction angle threshold needs to consider: for a properly magnetized magnetic bottle cap, the magnetic field lines in the edge region of the magnet have a high degree of divergence, and the corresponding magnetic field line direction angles are generally higher. The magnetic field in the central region of the magnet is uniform, and the magnetic field lines are approximately perpendicular to the end face, and the corresponding magnetic field line direction angles are generally lower.

[0031] Furthermore, the updated set of magnetic field line direction angles is read, and the three-dimensional spatial coordinates of the sampling points corresponding to each valid direction angle are matched. The magnetic induction intensity of each sampling point is used as a weighting coefficient, and a geometric fitting algorithm is used to inversely solve for the center position of the magnet and the axial orientation of the main magnetic pole. The solved magnetic pole center coordinates are used as the vector starting point, and the axial orientation of the main magnetic pole is used as the vector direction to synthesize the complete magnetic pole vector of the target magnetic bottle cap. The magnetic pole vector is vector data representing the spatial position and orientation of the main magnetic pole of the target magnetic bottle cap, containing the three-dimensional coordinate information of the magnetic pole center and the axial direction information of the magnetic pole.

[0032] For example, taking a circular, axially magnetized perfume bottle cap as an example, the cap contains a neodymium iron boron circular magnet with the main magnetic pole on the end face being the N pole. A 3×3 Hall sampling array is used to collect magnetic induction intensity at 9 points, and the average magnetic induction intensity of the target magnetic bottle cap is calculated to be 102 millitalas. Comparing the average magnetic induction intensity of the target magnetic bottle cap (102 millitalas) with a first preset magnetic induction intensity threshold of 80 millitalas, it is confirmed that the magnetic induction intensity of the target magnetic bottle cap exceeds the first preset magnetic induction intensity threshold, and the magnetic field signal is valid. The 9 sets of sampled data are input into a pre-trained support vector machine model, which outputs a 3×3 magnetic pole distribution matrix. The polarity of all sampling points is identified as N pole, and the magnetic induction intensity at each position corresponds one-to-one with the sampled value. The intensity is highest at the center and gradually decreases towards the periphery. Based on the intensity gradient of the magnetic pole distribution matrix, the magnetic field line direction angle of each sampling point is calculated. The magnetic field line at the center sampling point is approximately perpendicular to the end face, and the magnetic field line direction angle is approximately 3 degrees. The magnetic field lines at the eight surrounding sampling points diverge outwards, with direction angles of 12 degrees, 10 degrees, 11 degrees, 9 degrees, 13 degrees, 10 degrees, 12 degrees, and 11 degrees respectively. These nine direction angles together form a set of magnetic field line direction angles. Next, each value in the set of magnetic field line direction angles is compared to a preset direction angle threshold of 8 degrees. The magnetic field line direction angle of the central sampling point, at 3 degrees, does not exceed the preset threshold and belongs to the central uniform magnetic field region, thus having no locational value; therefore, this value is removed from the set of magnetic field line direction angles. The remaining eight magnetic field line direction angles all exceed the preset threshold and are considered valid edge feature points; all are retained, resulting in an updated set of magnetic field line direction angles containing eight valid direction angle values. Based on the updated set of magnetic field line direction angles, the spatial coordinates of 8 effective points are matched, and the magnetic induction intensity is weighted and fitted. The three-dimensional coordinates of the magnetic pole center are obtained by inverse solution (1200.2 mm, 450.5 mm, 30 mm). The direction of the main magnetic pole is upward along the vertical end face. Finally, the magnetic pole vector of the target magnetic bottle cap is synthesized.

[0033] In one embodiment of this application, step S120, which determines the magnetic force range of each manipulator based on magnetic pole vectors and physical contour data, further includes the following steps: Figure 3 As shown, the specific content is as follows: Step S310: Based on the magnetic pole vector and physical contour data, the directional bounding box algorithm is used to construct the asymmetric workspace model of the robot, and the coordinates of each boundary point of the asymmetric extension region are determined according to the asymmetric workspace model. Step S320: Obtain the preset magnetic field spatial attenuation model, and input the coordinates of each boundary point into the preset magnetic field spatial attenuation model to output the magnetic induction intensity of each boundary point coordinate; Step S330: If the magnetic induction intensity of the current boundary point coordinates exceeds the second preset magnetic induction intensity threshold, then mark the current boundary point as an effective adsorption point, and perform three-dimensional surface fitting on all effective adsorption points to determine the magnetic force range of each robot arm.

[0034] Specifically, a bounding box algorithm can be used. The geometric center of the robot's physical contour is taken as the origin of the bounding box, and the axial direction of the magnetic pole vector is taken as a principal axis direction, generating a basic bounding box aligned with the magnetic pole direction. Based on this, a preset basic safety margin for the non-magnetic pole direction is superimposed on the six faces in the non-magnetic pole direction, and a preset initial asymmetric extension amount in the magnetic pole direction is superimposed on the end face pointing to the magnetic pole vector. This ultimately constructs an asymmetric workspace model that protrudes along the magnetic pole direction and is symmetrical in the other directions. After completing the workspace model construction, the 3D coordinates of all vertices of the asymmetric workspace model are extracted, along with all boundary vertices and uniformly distributed in-plane sampling points corresponding to the asymmetric extension region, forming a set of coordinates for each boundary point of the asymmetric extension region. The asymmetric workspace model is an asymmetric spatial bounding model based on the physical contour of the robot's physical contour, with an additional magnetic field area extended along the magnetic pole vector direction and only a basic safety margin retained in the non-magnetic pole directions. This model accurately reflects the directional working space range of the robot.

[0035] Furthermore, a preset magnetic field spatial attenuation model is retrieved from the parameter library to confirm that the reference magnetic induction intensity and magnetic source origin coordinates of the magnetic field spatial attenuation model are completely matched with the current magnetic gripper parameters of the robot arm. Next, the coordinates of all boundary points output in step S310 are input one by one into the preset magnetic field spatial attenuation model, with the center of the end surface of the magnetic pole vector as the magnetic source reference point, to calculate the magnetic induction intensity value at the corresponding position.

[0036] The mathematical formula for the magnetic field spatial attenuation model is as follows:

[0037] In the formula, represents the magnetic flux density at the target point in space, expressed in Tesla. d is the nominal surface magnetic flux density at the magnetic source reference point, i.e., the rated magnetic flux density at the center of the magnetic pole end face of the magnetic clamp / magnetic bottle cap; d is the axial perpendicular distance from the target point to the magnetic pole end face, i.e., the distance along the direction of the magnetic pole vector, in millimeters. The equivalent reference distance is used to correct the attenuation deviation caused by the non-point source of the magnet in the near field, and the unit is millimeters; n is the attenuation exponent, which characterizes the attenuation rate of the magnetic field with increasing distance and is determined by the shape of the magnet and the magnetization method. This is the radial offset correction coefficient, ranging from 0 to 1. It is used to correct the magnetic field attenuation after the target point deviates from the center axis of the magnetic pole. The larger the radial offset distance, the smaller the coefficient.

[0038] Furthermore, if the magnetic induction intensity of the current boundary point exceeds the second preset magnetic induction intensity threshold, it indicates that the location is within the effective magnetic force range, and the current boundary point is marked as an effective adsorption point. If the magnetic induction intensity of the current boundary point does not exceed the second preset magnetic induction intensity threshold, it indicates that the magnetic force at this location is weak and there is no substantial risk of interference, and the boundary point is discarded. After completing the screening of all boundary points, all effective adsorption points are used as control points, and a triangular mesh surface fitting algorithm is used to perform three-dimensional closed surface fitting. The fitting accuracy is controlled within a preset deviation of 1 mm, and finally a closed three-dimensional surface boundary is obtained. The spatial region enclosed by this surface is the current magnetic force range of the robot. The magnetic force range is the spatial region enclosed by the closed three-dimensional surface formed by fitting all effective adsorption points. It corresponds to the actual effective magnetic field space generated by the robot's magnetic adsorption operation. Adjacent robot arms entering this region will experience significant magnetic interference. The setting of the second preset magnetic induction intensity threshold needs to consider the following: when the spatial magnetic induction intensity reaches a certain level, it will generate an observable magnetic attraction force on the metal end effector of adjacent robotic arms, which is sufficient to cause interference problems such as trajectory deviation and grasping posture deviation; when it is below a certain level, the regional magnetic force is weak and will not cause substantial interference to the operation of the robotic arms. Therefore, the second preset magnetic induction intensity threshold needs to be calibrated through actual working condition testing to accurately distinguish between the effective magnetic interference area and the weak magnetic ineffective area, so as to match the actual working scenario of the magnetic bottle cap assembly line.

[0039] For example, taking the No. 1 assembly robot on a magnetic bottle cap assembly line as an example, the physical outline of the magnetic gripper at the end of the robot is a cuboid with a length of 200 mm, a width of 150 mm, and a height of 100 mm. The magnetic pole vector direction is vertically downward, and the reference magnetic induction intensity on the gripper surface is 1.5 Tesla. The preset basic safety margin in the non-magnetic pole direction is 10 mm, the initial asymmetric extension in the magnetic pole direction is 50 mm, and the second preset magnetic induction intensity threshold is 0.2 Tesla. Based on the robot's physical outline and the vertically downward magnetic pole vector, an asymmetric workspace model with a length of 220 mm, a width of 170 mm, and a height of 160 mm is generated, where the asymmetric extension region in the magnetic pole direction is the lower 50 mm cuboid section. The coordinates of all vertices of the model and the uniformly sampled points within the extension region are extracted, resulting in 32 boundary point coordinates. The preset magnetic field space attenuation model is retrieved, and with the center of the lower end face of the magnetic gripper as the magnetic source reference point, the coordinates of the 32 boundary points are input into the model one by one to calculate the magnetic induction intensity. For example, the calculated magnetic induction intensity at the boundary point 20 mm from the magnetic source reference point is approximately 0.8 Tesla, at 30 mm it is approximately 0.36 Tesla, at 40 mm it is approximately 0.2 Tesla, and at 50 mm it is approximately 0.128 Tesla. The magnetic induction intensity of each boundary point is compared to the second preset magnetic induction intensity threshold of 0.2 Tesla. The boundary point at 50 mm from the magnetic source reference point, with a magnetic induction intensity of 0.128 Tesla, does not exceed the second preset threshold and is therefore discarded. The remaining 26 boundary points, all within 40 mm of the magnetic source reference point, have magnetic induction intensities exceeding the second preset threshold and are marked as valid adsorption points. Using 26 effective adsorption points as control points, a three-dimensional surface was fitted to obtain a closed three-dimensional surface with a spherical decay at the bottom. The space enclosed by this surface is the magnetic field range of the No. 1 manipulator. This range extends about 40 mm along the magnetic pole direction and is asymmetrically distributed, thus restoring the actual magnetic field boundary.

[0040] In one embodiment of this application, step S130 involves determining the coupling magnetic force region for the collaborative operation of each robotic arm based on the magnetic force range. The coupling magnetic force region includes a region of repulsion between like magnetic poles and a region of attraction between unlike magnetic poles. The step also includes the following steps: Figure 4 As shown, the specific content is as follows: Step S410: Determine the magnetic field attenuation gradient and the interference surface of the magnetic field lines based on the magnetic field range, and determine the dynamic magnetic flux density value based on the magnetic field attenuation gradient; Step S420: Determine the spatial magnetic field superposition vector based on the dynamic magnetic flux density value and the interference surface of the magnetic field lines, and determine the magnetic attraction edge distortion profile based on the spatial magnetic field superposition vector; Step S430: Mesh the region within the magnetic edge distortion contour to generate multiple magnetic edge distortion mesh points; Step S440: If the magnetic induction intensity of the magnetic attraction edge distortion grid point exceeds the third preset magnetic induction intensity threshold, then the area where the adjacent magnetic attraction edge distortion grid points with the same magnetic polarity are located is taken as the same magnetic pole repulsion area, and the area where the adjacent magnetic attraction edge distortion grid points with opposite magnetic polarity are located is taken as the opposite magnetic pole attraction area. Step S450: Perform a union operation on the repulsion region of the same magnetic pole and the attraction region of the opposite magnetic pole, and use the result of the union operation as the coupling magnetic force region for the collaborative operation of each robot arm.

[0041] Specifically, complete spatial magnetic flux density distribution data of the magnetic field range corresponding to each robotic arm is obtained. A three-dimensional grid magnetic field distribution model is constructed based on the Biot-Savart law. The first-order partial derivatives of the magnetic flux density function are solved along the X, Y, and Z coordinate axes to synthesize the magnetic field attenuation gradient vector at each point in three-dimensional space. The magnitude of the magnetic field attenuation gradient vector corresponds to the change in magnetic flux density per unit distance at that location. Subsequently, an isosurface extraction algorithm is used to search for surfaces with zero magnetic potential energy gradient within the overlapping area of ​​the magnetic fields of multiple robotic arms. This surface is identified as the magnetic field line intersection interference surface, which is the interface where the magnetic fields of two robotic arms interact most strongly. After completing the gradient and interference surface extraction, the real-time motion velocity of each robotic arm is obtained. Combined with a preset dynamic velocity compensation coefficient, the static magnetic flux density is dynamically corrected, and the dynamic magnetic flux density value corresponding to each spatial point is calculated.

[0042] Furthermore, using the intersection of magnetic field lines as the calculation reference plane, the dynamic magnetic flux density vectors of each manipulator at the reference plane are superimposed according to the vector synthesis rule to obtain the spatial magnetic field superposition vector corresponding to each point on the reference plane. The spatial magnetic field superposition vector simultaneously contains information on the strength and deflection direction of the composite magnetic field. Since the magnetic fields of multiple manipulators are not coaxially symmetrically distributed, the boundary of the superimposed composite magnetic field no longer maintains the regular symmetrical shape of the individual manipulator's magnetic field. By using the moving cube isosurface extraction algorithm, the closed boundary surface corresponding to the threshold is extracted from the composite three-dimensional magnetic field distribution. This boundary surface is the magnetic attraction edge distortion profile. The magnetic attraction edge distortion profile shows a significant deformation compared to the regular boundary of the single manipulator's magnetic field range, restoring the actual effective magnetic field boundary after multi-magnetic field coupling.

[0043] Furthermore, the three-dimensional spatial region enclosed by the magnetic edge distortion contour is uniformly meshed according to a preset meshing step size, dividing the continuous spatial region into several cubic mesh units. Then, the center point of each cubic mesh is extracted as the calculation reference point, which is the magnetic edge distortion mesh point.

[0044] Furthermore, all magnetically attracted edge distortion grid points are traversed, and the magnetic flux density of each point is compared with a third preset magnetic flux density threshold. If the magnetic flux density of a magnetically attracted edge distortion grid point does not exceed the third preset threshold, it indicates that the magnetic coupling effect at that location is weak, and the grid point is directly discarded. If the magnetic flux density of a magnetically attracted edge distortion grid point exceeds the third preset threshold, it indicates that the location belongs to an effective coupling interference region, and the grid point is retained and marked as a valid grid point. After the filtering is completed, a connectivity analysis is performed on all valid grid points based on the eight-neighbor connectivity rule: valid grid points with the same magnetic polarity and spatially adjacent are aggregated into the same connected region, and the spatial range corresponding to this connected region is the same magnetic pole repulsion region; valid grid points with opposite magnetic polarity and spatially adjacent are aggregated into the same connected region, and the spatial range corresponding to this connected region is the opposite magnetic pole attraction region. The setting of the third preset magnetic induction intensity threshold needs to be calibrated based on physical test data of dual-magnet coupling superposition. For example, when the spatial superposition magnetic induction intensity reaches 0.3 Tesla, it will generate an observable magnetic force on the metal end effector of the adjacent robot arm. Whether it is like-name repulsion or unlike-name attraction, it is enough to cause the trajectory deviation to exceed the allowable range of assembly tolerance. In areas below the third preset magnetic induction intensity threshold, the magnetic coupling effect is weak and will not cause substantial interference to the operation of the robot arm. The third preset magnetic induction intensity threshold also needs to be verified through multiple sets of magnetic coupling tests with different spacing and polarity to accurately distinguish between the effective coupling interference area and the weak magnetic ineffective area, and match the actual operation accuracy requirements of the magnetic bottle cap assembly line.

[0045] Furthermore, a three-dimensional spatial union operation is performed on the regions of repulsion between like-named magnetic poles and attraction between unlike-named magnetic poles. This merges the spatial boundaries of the two types of regions, removes overlapping intervals, and generates a complete closed three-dimensional spatial region. This merged region completely covers all spatial ranges that may cause magnetic interference to the operation of the robotic arms. Both the repulsive effect of like-named magnetic poles and the attractive effect of unlike-named magnetic poles are included within the unified boundary. This region is the coupling magnetic region for the collaborative operation of the robotic arms.

[0046] For example, taking assembly robot No. 1 and loading robot No. 2 on a perfume magnetic bottle cap assembly line as an example, the two robots operate adjacently, and their magnetic force application areas partially overlap during operation. The magnetic clamp at the end of robot No. 1 has an N pole and a static surface magnetic induction intensity of 1.5 Tesla, while the magnetic clamp at the end of robot No. 2 has an S pole and a static surface magnetic induction intensity of 1.2 Tesla. The preset meshing step size is 2 mm, and the third preset magnetic induction intensity threshold is 0.3 Tesla. A three-dimensional magnetic field distribution model is constructed based on the magnetic force application areas of the two robots, and the maximum magnitude of the magnetic field attenuation gradient in the overlapping area is calculated to be 0.05 Tesla per millimeter. The intersection interference surface of the magnetic field lines of the two robots is located using an isosurface extraction algorithm. This interference surface is located in the middle of the two robots and is an irregular curved surface. Then, the magnetic field vectors of the two robots are synthesized on the intersection interference surface to obtain the spatial magnetic field superposition vector at each point. The direction of the synthesized magnetic field is deflected by about 15 degrees compared to the direction of the individual magnetic field. Using 0.3 Tesla as the isosurface reference, the closed boundary is extracted from the synthetic magnetic field to obtain the magnetic attraction edge distortion profile. This profile is no longer a regular sphere, but rather an ellipsoidal distortion shape with a major-to-minor axis ratio of 1.2:1, restoring the boundary deformation after the superposition of two magnetic fields. A 2-millimeter step size mesh is performed on the three-dimensional space enclosed by the magnetic attraction edge distortion profile, generating approximately 12,000 cubic mesh elements. The center point of each mesh is extracted to obtain the corresponding number of magnetic attraction edge distortion mesh points. All magnetic attraction edge distortion mesh points are traversed, and each is compared with a third preset magnetic induction intensity threshold of 0.3 Tesla. Edge mesh points with magnetic induction intensities below 0.3 Tesla are discarded, retaining approximately 7,800 effective mesh points. Subsequently, the eight-neighbor connected domain algorithm is executed: the effective grid points on the side closer to robot arm 1 exhibit an overall N-pole property, and after adjacent connections, they form a magnetic repulsion region with the same polarity. Within this region, like poles repel each other, and entering robot arms will experience a reverse thrust. Conversely, the effective grid points on the side closer to robot arm 2 exhibit an overall S-pole property, with the magnetic field polarity opposite to that of robot arm 1. After adjacent connections, they form a magnetic attraction region with unlike polarities. Within this region, unlike poles attract each other, and entering robot arms will experience a positive pull. Finally, a three-dimensional Boolean union operation is performed on the magnetic repulsion region with the same polarity and the magnetic attraction region with unlike polarities, merging the boundaries of the two regions and removing the overlapping interaction intervals in between, ultimately resulting in a completely closed ellipsoidal three-dimensional spatial region. This region is the coupling magnetic force region for the two robot arms to work collaboratively, completely covering all spatial ranges that would generate magnetic interference.

[0047] In one embodiment of this application, step S140 involves obtaining the motion trajectory coordinate sequence of the current robot arm. Specifically, the initial planned continuous trajectory data for the corresponding task can be retrieved from the motion controller of the current robot arm. This trajectory is a smooth, continuous path generated by cubic spline interpolation, covering the complete motion process from the current position to the target workstation. Simultaneously, the real-time pose data of the current robot arm's end effector is read to confirm the end effector coordinates at the current moment. This coordinate is used as the starting point for capturing the motion trajectory coordinate sequence, ensuring that the trajectory sequence is synchronized with the real-time operating state of the robot arm.

[0048] For example, the initial planned continuous trajectory for this task is retrieved from the servo motion controller of assembly robot No. 1. Simultaneously, the current real-time pose of the robot's end effector is read, confirming the current end effector coordinates as (200.2, 300.5, 150.1) mm, which is used as the starting point of the sequence. The continuous trajectory is sampled at equal intervals at a sampling frequency of 50 Hz, with one coordinate point extracted every 20 milliseconds, resulting in a total of 200 discrete coordinate points covering the entire motion process over the next 4 seconds. For example, the first point corresponds to the starting moment, with coordinates of (200.2, 300.5, 150.1) mm; the 100th point corresponds to the middle of the motion at 2 seconds, with coordinates of (800.3, 450.2, 100.5) mm; and the 200th point corresponds to the assembly endpoint at 4 seconds, with coordinates of (1200.1, 450.0, 30.0) mm.

[0049] In one embodiment of this application, in step S150, if any coordinate point in the trajectory coordinate sequence falls into the same-name magnetic pole repulsion region of the adjacent robot, it is determined that there is repulsive interference between the adjacent robots, and the adjacent robot is designated as a repulsive obstacle. Specifically, the three-dimensional boundary data of the motion trajectory coordinate sequence of the current robot and the same-name magnetic pole repulsion region of the adjacent robot can be read. Using the global operation coordinate system as a unified reference, each coordinate point in the motion trajectory coordinate sequence is traversed in chronological order, and it is determined whether the current coordinate point falls within the spatial boundary range of the same-name magnetic pole repulsion region. If it falls within, it is determined that there is repulsive interference between the adjacent robot and the current robot, and the adjacent robot that generates the same-name magnetic pole repulsion region is marked as a repulsive obstacle.

[0050] In one embodiment of this application, step S160, which involves obtaining the coordinates of the repulsive obstacle and determining the repulsive force vector matrix and the first gravitational potential field gradient based on the coordinates of the repulsive obstacle, further includes the following steps: Figure 5 As shown, the specific content is as follows: Step S510: Obtain the first Euclidean distance between the coordinates of each working node of the current robot and the repulsive obstacle, and determine the repulsive potential field and the first gravitational potential field based on the first Euclidean distance; Step S520: Determine the repulsive force vector matrix based on the repulsive potential field, and determine the gradient of the first gravitational potential field based on the first gravitational potential field.

[0051] Specifically, the three-dimensional coordinates of all working nodes in the multi-dimensional state space of the current robotic arm are read, along with the reference coordinates of the repulsive obstacle. All working nodes are traversed point by point, and the spatial straight-line distance between each working node and the reference point of the repulsive obstacle is calculated using the three-dimensional Euclidean distance formula. This distance is the first Euclidean distance of the corresponding node. After the distance calculation, a repulsive potential field is constructed: The first Euclidean distance of each working node is checked against the effective repulsive distance threshold. If the first Euclidean distance is greater than the effective repulsive distance threshold, the node is outside the repulsive range, and magnetic repulsion is negligible; the repulsive potential field value of this node is assigned to 0. If the first Euclidean distance is less than or equal to the effective repulsive distance threshold, the repulsive potential field value of this node is calculated based on the repulsive potential field gain coefficient, following a law inversely proportional to the square of the distance; the closer the distance, the higher the potential field value. The repulsive potential field is a repulsive scalar potential field constructed with the repulsive obstacle as the center. The potential field value monotonically increases as the distance between the node and the repulsive source decreases, representing the magnitude of the repulsive potential energy at each point in space. Its negative gradient direction corresponds to the direction of the repulsive force. The formula for calculating the repulsive potential field is as follows:

[0052] In the formula, Let be the repulsive potential field value corresponding to the i-th working node. This is the first Euclidean distance. This is the repulsive potential field gain coefficient, used to calibrate the overall strength of the repulsive potential field, and directly determines the amplitude of the maximum repulsive force.

[0053] Furthermore, a first gravitational potential field is constructed: using the assembly endpoint coordinates of the current task as the gravitational target point, the Euclidean distance from each working node to the gravitational target point is calculated point by point. Based on the gravitational potential field gain coefficient, the first gravitational potential field value of that node is calculated according to a law proportional to the distance; the farther the distance from the target point, the higher the potential field value. The first gravitational potential field is an attractive scalar potential field constructed with the current manipulator's target point as the center. The potential field value monotonically increases with the increase of the distance between the node and the target point, and is used to guide the manipulator to move towards the target position. Its negative gradient direction corresponds to the gravitational direction. The calculation formula for the first gravitational potential field is as follows:

[0054] In the formula, Let be the value of the first gravitational potential field corresponding to the i-th working node; This is the gravitational potential field gain coefficient; The Euclidean distance from each working node to the gravitational target point.

[0055] Furthermore, gradient calculations are performed on the three-dimensional scalar distribution of the repulsive potential field. The first-order partial derivatives of the potential field function are calculated along the X, Y, and Z coordinate axes, and the negative gradient direction is taken as the direction of the repulsive force. The magnitude of the negative gradient corresponds to the magnitude of the repulsive force. The direction of the repulsive force always points away from the repulsive obstacle along the line connecting the two reference points, ensuring the robot is pushed away from the repulsive region of the same magnetic pole. The repulsive force vectors corresponding to all working nodes are arranged in spatial order to construct a repulsive force vector matrix. Each element in the matrix contains complete three-dimensional directional components and the magnitude of the force. Simultaneously, gradient calculations are performed on the three-dimensional scalar distribution of the first gravitational potential field. Similarly, the first-order partial derivatives of the potential field function are calculated along the three coordinate axes, and the negative gradient direction is taken as the gravitational direction, obtaining the first gravitational potential field gradient vector corresponding to each working node. The direction of the first gravitational potential field gradient always points towards the assembly target point, and its magnitude changes linearly with the distance from the node to the target point, continuously guiding the robot to converge towards the target position.

[0056] For example, taking the No. 1 assembly robot on a magnetic bottle cap assembly line as an example, the No. 2 loading robot has been marked as a repulsion obstacle. The reference coordinates of the geometric center of the magnetic clamp at the end of the No. 2 robot are (800, 450, 100) mm, and the effective distance threshold for the repulsion effect is 150 mm. The coordinates of the assembly target point of the No. 1 robot are (1200, 450, 30) mm. The coordinates of 120 working nodes of the No. 1 robot are extracted, and the first Euclidean distance to the reference point of the repulsion obstacle is calculated point by point. Three typical nodes are selected for calculation: the coordinates of node A are (720, 450, 100) mm, and the calculated first Euclidean distance is 80 mm, which is less than the effective threshold of 150 mm. Substituting this into the calculation, the repulsion potential field value of this node is 125 joules. Node B has coordinates of (900, 450, 100) mm and a first Euclidean distance of 100 mm, corresponding to a repulsive potential field value of 80 joules. Node C has coordinates of (1000, 450, 100) mm and a first Euclidean distance of 200 mm, exceeding the effective threshold, resulting in a repulsive potential field value of 0. The first gravitational potential field values ​​of the three nodes are calculated simultaneously: Node A is 480 mm from the target point, corresponding to a first gravitational potential field value of 0.072 joules; Node B is 300 mm from the target point, corresponding to a first gravitational potential field value of 0.045 joules; and Node C is 200 mm from the target point, corresponding to a first gravitational potential field value of 0.03 joules. Solving for the negative gradient of the repulsive potential field yields a repulsive force of 31.25 N at node A, directed along the positive X-axis away from the repulsive obstacle. The repulsive force at node B is 16 N, also directed along the positive X-axis. The repulsive force at node C is 0. The system arranges the repulsive force vectors of 120 nodes in sequence, forming a 120-row, 3-column repulsive force vector matrix. At the same time, it solves for the negative gradient of the first gravitational potential field, and obtains that the magnitude of the first gravitational potential field gradient of all nodes is 0.15 N / m, and the direction is along the positive X-axis pointing towards the assembly target point.

[0057] In one embodiment of this application, step S160, which determines the first target obstacle avoidance path based on the repulsive force vector matrix and the gradient of the first gravitational potential field, further includes the following steps: Figure 6 As shown, the specific content is as follows: Step S610: Based on the repulsive force vector matrix and the gradient of the first gravitational potential field, the first heuristic cost algebraic sum of the current position of the manipulator is determined using the A* algorithm; Step S620: If the sum of the first heuristic cost algebras exceeds the preset heuristic cost algebras threshold, then the annealing simulation algorithm is used to determine the set of repulsion escape points; Step S630: Based on the set of repulsive force escape points, a smooth repulsive force escape curve is determined using a spline interpolation algorithm, and multiple repulsive force obstacle avoidance nodes are determined according to the repulsive force escape curve. Step S640: If the distance between the repulsion obstacle avoidance node and the repulsion obstacle exceeds the first preset safe distance threshold, then the current repulsion obstacle avoidance node is taken as the first target obstacle avoidance point, and the first target obstacle avoidance path is determined based on the first target obstacle avoidance point.

[0058] Specifically, the repulsive force vector matrix and the gradient data of the first gravitational potential field of all nodes can be read. The current spatial position of the manipulator's end effector is used as the starting node for path search, and the assembly target point is used as the ending node. Then, the A* path search algorithm is used, with the obstacle avoidance cost corresponding to the repulsive potential field as the actual cost term and the target approach cost corresponding to the first gravitational potential field as the heuristic guidance term. The two costs are weighted and summed to obtain the first heuristic cost algebraic sum of the current manipulator position. The repulsive obstacle avoidance cost increases non-linearly as the node approaches the repulsive obstacle, while the target approach cost decreases linearly as the node approaches the assembly target point.

[0059] Furthermore, if the sum of the first heuristic cost algebras exceeds a preset threshold, it indicates that the overall obstacle avoidance cost at the current position of the robotic arm is too high. Under the guidance of a conventional artificial potential field, the obstacle avoidance path will continue to meander within the high repulsion region, unable to escape the repulsion region of the same magnetic pole and advance towards the target point, thus falling into a local inferior solution dilemma. In this case, the annealing simulation algorithm is immediately initiated to perform global optimization. Using the current position of the robotic arm as the search origin, candidate coordinate points are randomly generated within a semi-circular candidate region facing the assembly target point. Using the sum of the first heuristic cost algebras as the fitness function, the algorithm iteratively searches according to the Metropolis criterion of the annealing simulation algorithm: in the initial stage, the algorithm accepts inferior solutions with higher costs and higher probabilities at higher temperatures to expand the global search range and avoid premature convergence to a local suboptimal solution; as the iteration progresses, the temperature gradually decreases, and the probability of accepting inferior solutions decreases synchronously, eventually converging to the globally optimal escape point sequence. All optimal candidate points satisfying the spatial and cost constraints are summarized in the order of motion to form a set of repulsion escape points. If the sum of the first heuristic cost algebra does not exceed the preset heuristic cost algebra and threshold, the current path cost is determined to be within a reasonable range, and a normal obstacle avoidance path is directly generated along the potential force direction. The setting of the preset heuristic cost algebra and threshold requires calibration through numerous simulations under varying degrees of interference to accurately identify all high-cost scenarios requiring active escape, while avoiding misjudging normal obstacle avoidance transition zones.

[0060] Furthermore, a cubic spline interpolation algorithm can be used, with natural splines as boundary conditions, to construct cubic polynomial curve segments between every two adjacent escape control points. This constrains the function values, first derivative values, and second derivative values ​​at all connection points to be continuous, generating a repulsive force escape curve with continuous curvature and a smooth transition. The maximum curvature of the repulsive force escape curve throughout its entire length conforms to the curvature constraints of the robot's joint dynamics, with no abrupt inflection points or directional jumps, allowing for smooth execution by the robot. After curve generation, the repulsive force escape curve is discretely sampled at equal intervals according to a preset sampling step size, extracting multiple discrete coordinate points, which are the repulsive force obstacle avoidance nodes.

[0061] Furthermore, all repulsion obstacle avoidance nodes can be traversed, and the Euclidean distance between each node and the reference point of the repulsion obstacle can be calculated point by point. The calculated result is then compared with a first preset safe distance threshold. If the distance between the current repulsion obstacle avoidance node and the repulsion obstacle exceeds the first preset safe distance threshold, it indicates that the current repulsion obstacle avoidance node is within a safe range, the magnetic induction intensity at the corresponding location is lower than the effective interference threshold, and it will not re-enter the same magnetic pole repulsion region. The current repulsion obstacle avoidance node is then marked as the first target obstacle avoidance point. If the calculated result does not exceed the first preset safe distance threshold, the corresponding node is removed to prevent the path from approaching the magnetic interference region again. After completing the safety verification and screening of all nodes, all valid first target obstacle avoidance points are connected sequentially according to the movement order, while simultaneously connecting the entry point of the original trajectory with the subsequent normal operation path to generate a complete and continuous first target obstacle avoidance path. The setting of the first preset safety distance threshold needs to be determined based on the superimposed effects of three error sources: repeated positioning error of the robot arm, magnetic field boundary fluctuation, and motion inertial offset, so as to ensure that the obstacle avoidance path and the boundary of the same magnetic pole repulsion area maintain sufficient safety margin and avoid the trajectory after obstacle avoidance from re-entering the magnetic interference area.

[0062] For example, taking the No. 1 assembly robot on a perfume magnetic bottle cap assembly line as an example, the No. 2 loading robot has been marked as a repulsive obstacle. After the No. 1 robot's original planned path intrudes into the repulsive region of the same magnetic pole, it gets stuck in a high-cost detour at coordinates (850, 450, 100) mm. The preset heuristic cost algebraic sum threshold is 80, the annealing simulation algorithm has an initial temperature of 100, a cooling coefficient of 0.95, a maximum iteration of 200, a first preset safe distance threshold of 20 mm, and the escape curve uses natural spline boundaries with a node sampling step size of 10 mm. Based on the repulsive force vector matrix and the gradient of the first gravitational potential field, the A* algorithm is called to calculate the first heuristic cost algebraic sum at the current position. The calculated current cost value is 92, which exceeds the preset heuristic cost algebraic sum threshold of 80, confirming that the current path is in a high-cost local inferior solution region, and conventional potential field guidance cannot efficiently escape. Then, the annealing simulation algorithm is initiated. Candidate points are randomly generated within a semi-circular region with a radius of 80 mm on the side facing the target point, using the current position as the origin. The fitness function is iteratively optimized using the sum of the first heuristic cost algebra. After 200 cooling iterations, the algorithm converges to obtain 5 escape control points with coordinates of (860, 470, 100) mm, (880, 490, 100) mm, (920, 480, 100) mm, (960, 460, 100) mm, and (1000, 450, 100) mm, respectively. Based on these 5 escape control points, a smooth repulsion escape curve is generated using a cubic natural spline interpolation algorithm. Subsequently, equidistant sampling is performed on the repulsion escape curve with a step size of 10 mm, resulting in 15 discrete repulsion obstacle avoidance nodes. The Euclidean distances between 15 repulsion obstacle avoidance nodes and the reference point (800, 450, 100 mm) of the repulsion obstacle were calculated point by point. All node distances were verified to be greater than a first preset safety distance threshold of 20 mm, with a minimum distance of 62 mm, meeting safety requirements. All nodes were marked as first target obstacle avoidance points. These first target obstacle avoidance points were then connected sequentially, with the front end connecting to the original trajectory entry point and the rear end connecting to the normal path leading to the assembly target point, ultimately generating a complete first target obstacle avoidance path. This, in turn, generated the final control commands to drive the adjacent robotic arms.

[0063] In one embodiment of this application, in step S170, if the coordinate point in the trajectory coordinate sequence falls into the attraction region of the opposite magnetic pole of the adjacent robot arm, it is determined that there is attraction interference between the adjacent robot arms, and the adjacent robot arm is regarded as an attraction obstacle. Specifically, the setting of the third preset magnetic induction intensity threshold needs to be calibrated based on the experimental data of the coupling and superposition of two opposite magnets. When the spatial superposition magnetic induction intensity reaches the third preset magnetic induction intensity threshold, the attraction force generated by the opposite magnetic poles is sufficient to cause the trajectory offset of the robot arm end to exceed the allowable range of assembly tolerance, and the attraction force increases rapidly with positive feedback as the distance decreases, posing a risk of component adsorption and adhesion. Therefore, it is regarded as an attraction obstacle. When it is below the third preset magnetic induction intensity threshold, the attraction force in the corresponding area is weak and will not cause substantial interference to the operation of the robot arm.

[0064] In one embodiment of this application, step S180, which involves obtaining the coordinates of the attracting obstacle and determining the attraction vector matrix and the second gravitational potential field gradient based on the coordinates of the attracting obstacle, further includes the following steps: Figure 7 As shown, the specific content is as follows: Step S710: Obtain the second Euclidean distance between each working node of the current robot and the coordinates of the attracted obstacle, and determine the attraction potential field and the second gravitational potential field based on the second Euclidean distance; Step S720: Determine the attraction vector matrix based on the attraction potential field, and determine the gradient of the second gravitational potential field based on the second gravitational potential field.

[0065] Specifically, the three-dimensional coordinates of all working nodes in the current multi-dimensional state space of the robot are obtained, along with the reference coordinates of the attraction obstacle, ensuring that all coordinates are within a unified global working coordinate system. All working nodes are traversed point by point, and the spatial straight-line distance between each working node and the reference point of the attraction obstacle is calculated using the three-dimensional Euclidean distance formula. The calculated spatial straight-line distance is the second Euclidean distance of the corresponding working node. After completing the distance calculation, it is determined whether the second Euclidean distance of each working node is less than the effective attraction distance threshold. If the second Euclidean distance is greater than the effective attraction distance threshold, it indicates that the corresponding working node is outside the attraction range of the opposite magnetic poles, and the attraction force can be ignored; the attraction potential field value of the corresponding working node is assigned to 0. If the second Euclidean distance is less than or equal to the effective attraction distance threshold, the closer the distance, the lower the potential field value, and the stronger the corresponding attraction force. The formula for calculating the attraction potential field is:

[0066] In the formula, For attraction potential field, The gain coefficient of the attraction potential field. Let be the i-th second Euclidean distance.

[0067] The formula for calculating the second gravitational potential field is as follows:

[0068] In the formula, This is the second gravitational potential field. This is the gain coefficient of the second gravitational potential field.

[0069] Furthermore, gradient calculations can be performed on the three-dimensional scalar distribution of the attractive potential field. The first-order partial derivatives of the potential field function can be calculated along the X, Y, and Z coordinate axes, respectively. The negative gradient direction is taken as the direction of the attractive force, and the magnitude of the negative gradient corresponds to the magnitude of the attractive force. Therefore, the gradient of the second gravitational potential field... The calculation formula is:

[0070] For example, taking the No. 1 assembly robot on a magnetic bottle cap assembly line as an example, the No. 2 loading robot has been marked as an attraction obstacle. The reference coordinates of the geometric center of the magnetic clamp at the end of the No. 2 robot are (800, 450, 100) mm; the assembly target point coordinates of the No. 1 robot are (1200, 450, 30) mm; the preset effective attraction distance threshold is 180 mm, the attraction potential field gain coefficient is 1.2, the second attraction potential field gain coefficient is 0.2, the working node spatial step size is 20 mm, and there are a total of 120 working nodes. The coordinates of the 120 working nodes of the No. 1 robot are extracted, and the second Euclidean distance to the reference point of the attraction obstacle is calculated point by point. Three typical nodes are selected for calculation: the coordinates of node A are (720, 450, 100) mm, and the calculated second Euclidean distance is 80 mm, which is less than the effective threshold of 180 mm. Substituting this into the calculation, the attraction potential field value of the corresponding working node is -15 joules. Node B has coordinates of (900, 450, 100) mm and a second Euclidean distance of 100 mm, corresponding to an attractive potential field value of -12 joules. Node C has coordinates of (1000, 450, 100) mm and a second Euclidean distance of 200 mm, exceeding the effective threshold, resulting in an attractive potential field value of 0. The second gravitational potential field values ​​of the three nodes are calculated simultaneously: Node A is approximately 485 mm from the target point, corresponding to a second gravitational potential field value of approximately 0.0235 joules; Node B is approximately 308 mm from the target point, corresponding to a second gravitational potential field value of approximately 0.0095 joules; and Node C is approximately 202 mm from the target point, corresponding to a second gravitational potential field value of approximately 0.0041 joules. Solving for the negative gradient of the attractive potential field yields an attractive force of 187.5 N for node A, directed along the negative X-axis towards the obstacle; an attractive force of 120 N for node B, also directed along the negative X-axis. The attractive force of node C is 0. The system arranges the attraction vectors of 120 working nodes sequentially, forming a 120-row, 3-column attraction vector matrix. Simultaneously, it solves for the negative gradient of the second gravitational potential field, obtaining that the magnitude of the second gravitational potential field gradient of all working nodes varies linearly with distance. The gradient magnitude at node A is approximately 0.097 N, and its direction is along the positive X-axis pointing towards the assembly target point.

[0071] In one embodiment of this application, step S180, which determines the second target obstacle avoidance path based on the attraction vector matrix and the gradient of the second gravitational potential field, further includes the following steps: Figure 8 As shown, the specific content is as follows: Step S810: Based on the attraction vector matrix and the gradient of the second gravitational potential field, the second heuristic cost algebraic sum of the current position of the manipulator is determined using the A* algorithm; Step S820: If the sum of the second heuristic cost algebras exceeds the preset heuristic cost algebras threshold, then the annealing simulation algorithm is used to determine the set of attraction escape points; Step S830: Based on the set of attraction escape points, a smooth attraction escape curve is determined using a spline interpolation algorithm, and multiple attraction obstacle avoidance nodes are determined according to the attraction escape curve; Step S840: If the distance between the attraction obstacle avoidance node and the attraction obstacle exceeds the second preset safe distance threshold, then the current attraction obstacle avoidance node is taken as the second target obstacle avoidance point, and the second target obstacle avoidance path is determined based on the second target obstacle avoidance point.

[0072] Specifically, the A* path search algorithm is employed, using the adsorption risk cost corresponding to the attraction potential field as the actual cost term and the target approach cost corresponding to the second gravitational potential field as the heuristic guidance term. The weighted sum of these two costs yields the second heuristic cost algebraic sum for the current position of the manipulator. The adsorption risk cost increases rapidly and non-linearly as the node approaches the attraction obstacle, while the target approach cost decreases linearly as the node approaches the assembly target point. If the second heuristic cost algebraic sum exceeds a preset threshold, it indicates that the adsorption risk cost at the current position of the manipulator is too high. Under conventional potential field guidance, the path will continuously deviate towards the attraction obstacle, failing to efficiently escape the opposite magnetic pole attraction region and advance towards the target point, posing a high risk of adsorption and adhesion. Therefore, with the current position of the manipulator as the search origin, candidate coordinate points are randomly generated within a fan-shaped candidate region facing away from the attraction obstacle and towards the assembly target point. The second heuristic cost algebraic sum is used as the fitness function, and the search is iteratively performed according to the Metropolis criterion of the annealing simulation algorithm. In the initial stage of the algorithm, inferior solutions with higher costs are accepted with higher probability at higher temperatures to expand the global search range and avoid premature convergence to local suboptimal solutions near the attraction source. As the iteration progresses, the temperature gradually decreases, and the probability of accepting inferior solutions decreases synchronously, eventually converging to the globally optimal escape point sequence. All optimal candidate points that satisfy spatial constraints, safety constraints, and cost constraints are summarized in the order of motion to form an attractive force escape point set. If the algebraic sum of the second heuristic cost does not exceed the preset threshold for the algebraic sum of the heuristic cost, the current path cost is determined to be within a reasonable range, and a conventional anti-deviation path is directly generated along the gradient direction of the second gravitational potential field.

[0073] Furthermore, all attraction-based obstacle avoidance nodes are traversed, and the Euclidean distance between each node and the reference point of the attraction obstacle is calculated. The calculated result is compared with a second preset safety distance threshold. If the distance between the current attraction-based obstacle avoidance node and the attraction obstacle exceeds the second preset safety distance threshold, it indicates that the current attraction-based obstacle avoidance node is within a safe range, the magnetic induction intensity at the corresponding location is lower than the effective attraction threshold, and it will not be significantly affected by the traction interference of opposite magnetic poles. At the same time, the current attraction-based obstacle avoidance node is marked as the second target obstacle avoidance point. If the distance does not exceed the second preset safety distance threshold, the corresponding node is removed to avoid the risk of path residue adsorption. All effective second target obstacle avoidance points are connected sequentially according to the movement sequence, and the entry point of the original trajectory is connected to the subsequent normal operation path to generate a complete and continuous second target obstacle avoidance path. The setting of the second preset safety distance threshold needs to consider the superimposed effects of three types of errors: repeated positioning error of the robot arm, motion inertial offset, and magnetic field boundary fluctuation, to avoid the path being pulled by attraction after obstacle avoidance and re-entering the interference area. At the same time, this threshold should not excessively lengthen the obstacle avoidance path.

[0074] This application also provides a multi-robot path dynamic planning system for a magnetic bottle cap assembly line. The system may include a first acquisition module, a first determination module, a second determination module, a second acquisition module, a third determination module, a first obstacle avoidance module, a fourth determination module, and a second obstacle avoidance module. Specifically, the first acquisition module acquires the magnetic induction intensity of the target magnetic bottle cap and the physical contour data of each robot body; the first determination module determines the magnetic pole vector of the target magnetic bottle cap based on the magnetic induction intensity, and determines the magnetic force range of each robot based on the magnetic pole vector and the physical contour data; the second determination module determines the coupled magnetic force region for collaborative operation of each robot based on the magnetic force range, the coupled magnetic force region including a region of repulsion between like poles and a region of attraction between unlike poles; the second acquisition module acquires the motion trajectory coordinate sequence of the current robot; and the third determination module determines that if any coordinate point in the trajectory coordinate sequence falls into the region of repulsion between like poles of an adjacent robot, a repulsion exists between the adjacent robots. The system is designed to: 1) identify the interference between adjacent robotic arms and treat them as repulsive obstacles; 2) identify the interference between adjacent robotic arms and treat them as attractive obstacles; 3) identify the interference between adjacent robotic arms and treat them as attractive obstacles; 4) identify the interference between adjacent robotic arms and treat them as attractive obstacles; 5) identify the interference between adjacent robotic arms and treat them as attractive obstacles; 6) identify the interference between adjacent robotic arms and treat them as attractive obstacles; 7) identify the interference between adjacent robotic arms and treat them as attractive obstacles; 8) identify the interference between adjacent robotic arms and treat them as attractive obstacles; 9) identify the interference between adjacent robotic arms and treat them as attractive obstacles; 10) identify the interference between adjacent robotic arms and treat them as repulsive obstacles; 11) identify the interference between adjacent robotic arms and treat them as repulsive obstacles; 12) identify the interference between adjacent robotic arms and treat them as attractive obstacles; 13) identify the interference between adjacent robotic arms and treat them as repulsive obstacles; 14) identify the interference between adjacent robotic arms and treat them as repulsive obstacles; 15) identify the interference between adjacent robotic arms and treat them as repulsive obstacles; 16) identify the interference between adjacent robotic arms and treat them as repulsive obstacles; 17) identify the interference between adjacent robotic arms and treat them as repulsive obstacles; 18) identify the interference between adjacent robotic arms and treat them as repulsive obstacles; 19) identify the interference between adjacent robotic arms and treat them as repulsive obstacles; 10) identify the interference between adjacent robotic arms and treat them as repulsive obstacles; 10) identify the interference between adjacent robotic arms and treat them as repulsive obstacles; 19 ...

[0075] It should be noted that the embodiments of the multi-robot path dynamic planning system for magnetic bottle cap assembly lines provided in this application can be used to execute the processing flow of the embodiments of the multi-robot path dynamic planning method for magnetic bottle cap assembly lines in the above embodiments. Its functions will not be repeated here, but can be referred to the detailed description of the above method embodiments.

[0076] This application also provides an electronic device, which includes one or more processors and memory resources represented by memory for storing instructions executable by the processors, such as application programs. The application programs stored in the memory may include one or more modules, each corresponding to a set of instructions. Furthermore, the processor is configured to execute instructions to perform the aforementioned method for dynamic path planning of multiple robotic arms for magnetic bottle cap assembly lines.

[0077] The electronic device may also include a power supply component configured to perform power management of the electronic device, a wired or wireless network interface configured to connect the electronic device to a network, and an input / output (I / O) interface. The electronic device can be operated based on operating devices stored in memory, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, or similar.

[0078] In one embodiment, a computer device, which may be a server, is also provided. The computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is connected to the system bus via the I / O interfaces. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes non-volatile storage media and internal memory. The non-volatile storage media stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database of the computer device stores data. The I / O interfaces of the computer device are used for exchanging information between the processor and external devices. The communication interface of the computer device is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements a dynamic programming method for multiple robotic arm paths in a magnetic bottle cap assembly line.

[0079] In one embodiment, a computer device is provided, which may be a terminal. The computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The input / output interface of the computer device is used for exchanging information between the processor and external devices. The communication interface of the computer device is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a multi-robot path dynamic planning method for a magnetic bottle cap assembly line. The display unit of the computer device is used to form a visually visible image and may be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0080] This application also provides a non-transitory computer-readable storage medium, which, when the instructions in the storage medium are executed by the processor of the electronic device, enables the electronic device to perform a method for dynamic path planning of multiple robotic arms for a magnetic bottle cap assembly line.

[0081] This application may take the form of a computer program product implemented on one or more storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing program code. Computer-readable storage media include permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information may be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to: phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0082] It should be noted that although the steps of the multi-robot path dynamic programming method for magnetic bottle cap assembly lines in this application are described in a specific order in the accompanying drawings, this does not require or imply that these steps must be performed in that specific order, or that all the steps shown must be performed to achieve the desired result. Additional or alternative steps, such as omitting certain steps, combining multiple steps into one step, and / or decomposing one step into multiple steps, should all be considered part of this application.

[0083] It should be understood that this application is not limited to the detailed structure and arrangement of the modules in the multi-robot path dynamic planning system for magnetic bottle cap assembly lines proposed in this specification. This application can have other embodiments and can be implemented and executed in various ways. The foregoing variations and modifications fall within the scope of this application. It should be understood that the invention and definition of this application extend to all alternative combinations of two or more individual features mentioned or apparent in the text and / or drawings. All these different combinations constitute multiple alternative aspects of this application. The embodiments described in this specification illustrate the best known mode for implementing this application and will enable those skilled in the art to utilize this application.

Claims

1. A dynamic path planning method for multiple robotic arms in a magnetic bottle cap assembly line, characterized in that, include: The magnetic induction intensity of the target magnetic bottle cap and the physical contour data of each robotic arm body are obtained respectively. The magnetic pole vector of the target magnetic bottle cap is determined based on the magnetic induction intensity, and the magnetic force range of each of the robotic arms is determined based on the magnetic pole vector and the physical contour data. The coupling magnetic field region for the coordinated operation of each robotic arm is determined based on the magnetic field range. The coupling magnetic field region includes a region of repulsion between like magnetic poles and a region of attraction between unlike magnetic poles. Obtain the current motion trajectory coordinate sequence of the robotic arm; If any coordinate point in the trajectory coordinate sequence falls into the same magnetic pole repulsion region of the adjacent robot arm, it is determined that there is repulsion interference between the adjacent robot arms, and the adjacent robot arm is regarded as a repulsion obstacle. Obtain the coordinates of the repulsive obstacle, determine the repulsive force vector matrix and the first gravitational potential field gradient based on the coordinates of the repulsive obstacle, and determine the first target obstacle avoidance path based on the repulsive force vector matrix and the first gravitational potential field gradient. If the coordinate point in the trajectory coordinate sequence falls into the opposite magnetic pole attraction region of the adjacent robot arm, then it is determined that there is attraction interference between the adjacent robot arms, and the adjacent robot arm is regarded as an attraction obstacle. The coordinates of the attracting obstacle are obtained, the attraction vector matrix and the second gravitational potential field gradient are determined based on the coordinates of the attracting obstacle, and the second target obstacle avoidance path is determined based on the attraction vector matrix and the second gravitational potential field gradient.

2. The multi-robot path dynamic planning method for magnetic bottle cap assembly lines according to claim 1, characterized in that, Determining the magnetic pole vector of the target magnetic bottle cap based on the magnetic induction intensity includes: If the magnetic induction intensity of the target magnetic bottle cap exceeds the first preset magnetic induction intensity threshold, then a support vector machine is used to determine the magnetic pole surface distribution matrix based on the magnetic induction intensity of the target magnetic bottle cap. The set of magnetic field line direction angles is determined based on the magnetic pole surface distribution matrix; If the magnetic field line direction angle in the set of magnetic field line direction angles does not exceed the preset direction angle threshold, the current magnetic field line direction angle is removed to update the set of magnetic field line direction angles. The magnetic pole vector of the target magnetic bottle cap is determined based on the updated set of magnetic field line direction angles.

3. The multi-robot path dynamic planning method for magnetic bottle cap assembly lines according to claim 1, characterized in that, The step of determining the magnetic force range of each of the robotic arms based on the magnetic pole vector and the physical contour data includes: Based on the magnetic pole vector and the physical contour data, an asymmetric workspace model of the robot is constructed using the orientation bounding box algorithm, and the coordinates of each boundary point of the asymmetric extension region are determined according to the asymmetric workspace model. A preset magnetic field spatial attenuation model is obtained, and the coordinates of each boundary point are input into the preset magnetic field spatial attenuation model to output the magnetic induction intensity of each boundary point coordinate. If the magnetic induction intensity of the current boundary point exceeds the second preset magnetic induction intensity threshold, the current boundary point is marked as an effective adsorption point, and three-dimensional surface fitting is performed on all the effective adsorption points to determine the magnetic force range of each manipulator.

4. The multi-robot path dynamic planning method for magnetic bottle cap assembly lines according to claim 1, characterized in that, The step of determining the coupling magnetic field region for the collaborative operation of each robotic arm based on the magnetic field range includes: The magnetic field attenuation gradient and the interference surface of the magnetic field lines are determined based on the magnetic field attenuation range, and the dynamic magnetic flux density value is determined based on the magnetic field attenuation gradient. The spatial magnetic field superposition vector is determined based on the dynamic magnetic flux density value and the magnetic field line intersection interference surface, and the magnetic attraction edge distortion profile is determined based on the spatial magnetic field superposition vector; The region within the magnetic edge distortion contour is meshed to generate multiple magnetic edge distortion mesh points; If the magnetic induction intensity of the magnetic attraction edge distortion grid point exceeds the third preset magnetic induction intensity threshold, then the area where the adjacent magnetic attraction edge distortion grid points with the same magnetic polarity are located is taken as the same magnetic pole repulsion area, and the area where the adjacent magnetic attraction edge distortion grid points with opposite magnetic polarity are located is taken as the opposite magnetic pole attraction area. Perform a union operation on the repulsion region of the same magnetic pole and the attraction region of the opposite magnetic pole, and use the result of the union operation as the coupling magnetic force region for the collaborative operation of each robot arm.

5. The multi-robot path dynamic planning method for magnetic bottle cap assembly lines according to claim 1, characterized in that, Determining the repulsive force vector matrix and the first gravitational potential field gradient based on the coordinates of the repulsive obstacle includes: Obtain the first Euclidean distance between the coordinates of each working node of the current robot arm and the repulsive obstacle, and determine the repulsive potential field and the first gravitational potential field based on the first Euclidean distance; The repulsive force vector matrix is ​​determined based on the repulsive potential field, and the first gravitational potential field gradient is determined based on the first gravitational potential field.

6. The multi-robot path dynamic planning method for magnetic bottle cap assembly lines according to claim 5, characterized in that, The step of determining the first target obstacle avoidance path based on the repulsive force vector matrix and the gradient of the first gravitational potential field includes: Based on the repulsive force vector matrix and the gradient of the first gravitational potential field, the first heuristic cost algebraic sum of the current position of the manipulator is determined using the A* algorithm. If the sum of the first heuristic cost algebras exceeds a preset heuristic cost algebras threshold, then the annealing simulation algorithm is used to determine the set of repulsion escape points. Based on the set of repulsive force escape points, a smooth repulsive force escape curve is determined using a spline interpolation algorithm, and multiple repulsive force obstacle avoidance nodes are determined based on the repulsive force escape curve. If the distance between the repulsive obstacle avoidance node and the repulsive obstacle exceeds a first preset safe distance threshold, then the current repulsive obstacle avoidance node is taken as the first target obstacle avoidance point, and the first target obstacle avoidance path is determined based on the first target obstacle avoidance point.

7. The multi-robot path dynamic planning method for magnetic bottle cap assembly lines according to claim 1, characterized in that, Determining the attraction vector matrix and the second gravitational potential field gradient based on the coordinates of the attraction obstacle includes: Obtain the second Euclidean distance between the coordinates of each working node of the current robot arm and the attracted obstacle, and determine the attractive potential field and the second gravitational potential field based on the second Euclidean distance; The attraction vector matrix is ​​determined based on the attraction potential field, and the gradient of the second gravitational potential field is determined based on the second gravitational potential field.

8. The multi-robot path dynamic planning method for magnetic bottle cap assembly lines according to claim 1, characterized in that, The step of determining the second target obstacle avoidance path based on the attraction vector matrix and the gradient of the second gravitational potential field includes: Based on the attraction vector matrix and the gradient of the second gravitational potential field, the second heuristic cost algebraic sum of the current position of the manipulator is determined using the A* algorithm. If the second heuristic cost algebraic sum exceeds the preset heuristic cost algebraic sum threshold, then the annealing simulation algorithm is used to determine the set of attraction escape points; Based on the set of attraction escape points, a smooth attraction escape curve is determined using a spline interpolation algorithm, and multiple attraction obstacle avoidance nodes are determined based on the attraction escape curve. If the distance between the attraction obstacle avoidance node and the attraction obstacle exceeds the second preset safe distance threshold, the current attraction obstacle avoidance node is taken as the second target obstacle avoidance point, and the second target obstacle avoidance path is determined based on the second target obstacle avoidance point.

9. A multi-robot path dynamic planning system for a magnetic bottle cap assembly line, characterized in that, include: The first acquisition module is used to acquire the magnetic induction intensity of the target magnetic bottle cap and the physical contour data of each robotic arm body respectively. The first determining module is used to determine the magnetic pole vector of the target magnetic bottle cap based on the magnetic induction intensity, and to determine the magnetic force range of each of the robotic arms based on the magnetic pole vector and the physical contour data. The second determining module is used to determine the coupled magnetic force region for the collaborative operation of each robot arm according to the magnetic force range, wherein the coupled magnetic force region includes a region of repulsion between like magnetic poles and a region of attraction between unlike magnetic poles. The second acquisition module is used to acquire the current motion trajectory coordinate sequence of the robotic arm; The third determining module is used to determine that there is a repulsive interference between adjacent robotic arms if any coordinate point in the trajectory coordinate sequence falls into the same magnetic pole repulsion region of the adjacent robotic arm and the current robotic arm, and to treat the adjacent robotic arm as a repulsive obstacle. The first obstacle avoidance module is used to obtain the coordinates of the repulsive obstacle, determine the repulsive force vector matrix and the first gravitational potential field gradient based on the coordinates of the repulsive obstacle, and determine the first target obstacle avoidance path based on the repulsive force vector matrix and the first gravitational potential field gradient. The fourth determining module is used to determine that there is attraction interference between adjacent robotic arms if the coordinate points in the trajectory coordinate sequence fall into the opposite magnetic pole attraction area of ​​the adjacent robotic arm and the current robotic arm, and to treat the adjacent robotic arm as an attraction obstacle. The second obstacle avoidance module is used to obtain the coordinates of the attracted obstacle, determine the attraction vector matrix and the second gravitational potential field gradient based on the coordinates of the attracted obstacle, and determine the second target obstacle avoidance path based on the attraction vector matrix and the second gravitational potential field gradient.

Images