A multi-target parallel shunting method and system for a logistic balance wheel sorting machine

CN122605720APending Publication Date: 2026-08-21JIANGYIN TIEYUAN INTELLIGENT TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202610530603.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-04-21
Publication Date
2026-08-21

AI Technical Summary

Technical Problem

这就导致轻软包裹常因受力过猛而发生翻滚、抛飞或破损;而重型或重心偏置的包裹则容易因受力不均而在分流时原地打转或卡顿

Benefits of technology

[0017]与现有技术相比,本发明所达到的有益效果是:通过多维感知精准提取包裹的真实物理接触印迹与三维动态重心,并创新性地引入网格自适应降级分配、防撕裂边界干涉动态仲裁以及防偏航力矩平衡解算机制,不仅从根本上消除了偏心件原地打转、轻软包裹受力翻滚以及密集并发状态下相邻包裹发生机械干涉与撕裂的技术隐患,实现了海量异形包裹的高密度、安全并行分流,同时还有效规避了全功率盲目输出带来的冗余功耗,极大地提升了物流分拣线的整体处理极限与设备使用寿命。

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122605720A_ABST
    Figure CN122605720A_ABST
Patent Text Reader

Abstract

The application discloses a logistics balance wheel sorting machine multi-target parallel shunting method and system, relates to the technical field of logistics warehouse automation and intelligent sorting control, and comprises a multi-dimensional sensing module, a unified control module and a matrix type independent execution module.The multi-dimensional sensing module is configured to acquire three-dimensional depth point cloud data and dynamic mass distribution characteristics of a package in real time before the package enters a shunting area.The unified control module is in network connection with the multi-dimensional sensing module and is configured to extract a real physical contact trace of the package and a conveying belt by threshold truncation according to the three-dimensional depth point cloud data, and calculate three-dimensional dynamic barycentric projection coordinates of the package in combination with the dynamic mass distribution characteristics, so that the barycentric projection coordinates are used as poles.The application solves the problems of yaw spinning, tearing of light and soft packages and edge stress tumbling during sorting of special eccentric parts.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of logistics warehousing automation and intelligent sorting control technology, specifically to a multi-objective parallel diversion method and system for a logistics swing wheel sorting machine. Background Technology

[0002] With the rapid development of e-commerce and express logistics, the parcel handling volume of distribution centers and warehousing nodes has increased dramatically. As a core piece of equipment in automated logistics sorting lines, the swing wheel sorter is widely used due to its relatively gentle sorting action and applicability to a wide variety of parcels. Its basic working principle involves controlling multiple sets of swing wheels at the bottom of the conveyor belt to turn as needed, using friction to change the direction of the parcel's movement and guide it into the corresponding slots on both sides.

[0003] However, the limitations of traditional balance wheel sorting systems are becoming increasingly apparent when faced with massive and dense complex parcel flows: First, existing systems mostly adopt a single-target serial control mode, which relies on passive triggering by sensors. To prevent adjacent packages from colliding in the sorting area or accidentally entering the same compartment, a large physical safety distance must be forcibly reserved between packages. This mode seriously wastes the physical space of the sorting machine, directly limiting the overall throughput of the sorting line and making it difficult to cope with the pressure of warehouse overflow during peak logistics periods.

[0004] Secondly, when attempting to exceed traditional safety distances for parallel sorting of high-density parcel flows, the shortcomings of traditional systems are fully exposed. Due to the lack of overall consideration of the actual physical contact surface and center of gravity shift of the parcels, the system typically outputs a fixed swing wheel deflection angle and maximum mechanical force based on a rough two-dimensional circumscribed rectangle. This results in lightweight, soft parcels often tumbling, being thrown, or breaking due to excessive force; while heavy or off-center-of-gravity parcels are prone to spinning in place or getting stuck during sorting due to uneven force. In addition, the moment a preceding parcel changes direction, it can easily cause physical interference to the parcels immediately following it, leading to deviations in the trajectory of subsequent parcels and consequently causing large-scale congestion and missorting.

[0005] Furthermore, the long-term, indiscriminate full-load output of redundant shifting force in traditional systems not only increases useless power consumption but also leads to excessive aging and severe wear of the actuators and transmission components. Therefore, there is an urgent need for a multi-objective parallel diversion method and system for logistics swing wheel sorting machines to break through the bottleneck of traditional physical spacing and achieve refined, adaptive, high-density parallel diversion based on the underlying dynamic properties of packages. Summary of the Invention

[0006] The purpose of this invention is to provide a multi-objective parallel diversion method and system for a logistics swing wheel sorting machine, so as to solve the problems mentioned in the background art.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a multi-objective parallel diversion method and system for a logistics swing wheel sorting machine, including a multi-dimensional perception module, a unified control module and a matrix-style independent execution module; The multi-dimensional sensing module is configured to acquire the three-dimensional depth point cloud data and dynamic quality distribution characteristics of the package in real time before the package enters the diversion area. The overall control module, network-connected to the multi-dimensional perception module, is configured to extract the actual physical contact imprints between the package and the conveyor belt based on the three-dimensional depth point cloud data through threshold truncation, and calculate the three-dimensional dynamic center of gravity projection coordinates of the package in combination with the dynamic mass distribution characteristics. Then, using the center of gravity projection coordinates as the pole, the area covered by the actual physical contact imprints is mapped and downgraded into an effective set of grids including core driving grids, edge auxiliary grids, and flexible following grids. The overall control module is also used to establish a torque balance equation based on the center of gravity coordinates and calculate and solve the differential thrust matrix allocated to each grid. The matrix-style independent execution module, connected to the overall control module, is used to independently and non-uniformly drive different effective grid sets at the bottom of multiple concurrent packages simultaneously, based on the differential thrust matrix.

[0008] According to the above technical solution, the multi-dimensional sensing module includes a three-dimensional depth camera array and a matrix-type dynamic weighing unit; the three-dimensional depth camera array is configured as a top-down scanning conveyor belt, and the output includes spatial coordinates. The high-density point cloud tensor; the matrix-type dynamic weighing unit is embedded below the conveyor belt in the preparation area to collect dynamic pressure distribution heat map of the package as it passes by, and to perform spatiotemporal registration with the high-density point cloud tensor to output multidimensional feature data containing mass distribution weights.

[0009] According to the above technical solution, the overall control module includes a physical imprint and center of gravity calculation unit, a grid degradation allocation unit, a boundary interference dynamic arbitration unit, and a torque balance differential speed calculation unit; The physical imprint and center of gravity calculation unit is used to filter out suspended geometry in the three-dimensional point cloud, extract only the point cloud that fits the surface of the conveyor belt to generate real physical contact imprints, and calculate the three-dimensional dynamic center of gravity projection coordinates. The grid degradation allocation unit is used to divide the balance wheel covered by the real physical contact imprint into the core driving grid, the edge auxiliary grid and the flexible following grid according to the radial distance gradient from the center of gravity, based on the physical coordinates of the balance wheel. The boundary interference dynamic arbitration unit is used to calculate the momentum contribution weight of the pendulum wheel in the overlapping area to each package when the actual physical contact imprints of two adjacent packages are extremely close, causing mesh overlap, and to perform arbitration on control ownership or non-powered isolation. The torque balance differential calculation unit is used to generate the differential thrust matrix that makes the resultant force vector pass through the center of gravity, based on the target deflection angle and the center of gravity projection coordinates.

[0010] According to the above technical solution, the matrix-type independent execution module includes several rubber balance wheel micro-units arranged in a two-dimensional matrix array. Each rubber balance wheel micro-unit is equipped with an independent servo direct drive motor and an electromagnetic deflection clutch. The electromagnetic deflection clutch supports independent control of angle and speed, and has a free wheel energy release mode to cut off power in response to the non-powered isolation zone command issued by the boundary interference dynamic arbitration unit.

[0011] According to the above technical solution, the overall control module further includes a material friction inversion and prediction unit, which is used to match a preset material library according to the surface texture features of the three-dimensional depth point cloud data, and invert and predict the non-uniform sliding friction coefficient matrix of the bottom of the package in combination with the dynamic mass distribution features, and input the matrix as a basic constraint condition into the torque balance equation.

[0012] A multi-objective parallel sorting method for a logistics swing wheel sorting machine using the system described above, the method comprising the following steps: Step S1: The multi-dimensional sensing module collects real-time data on the first [item] entering the diversion preparation area. The high-density point cloud tensor and dynamic pressure distribution data of each package were analyzed, and the bottom friction coefficient was obtained based on surface material matching. and current speed of movement ; Step S2: Set the bottom surface contact height threshold for the physical imprint and center of gravity calculation unit. In the high-density point cloud tensor Slice the point set and project it to generate a real physical contact imprint. Simultaneously, combining the dynamic pressure distribution data, the three-dimensional dynamic center of gravity of the package is calculated and projected onto the balance wheel plane to obtain the center of gravity projection coordinates. ; Step S3: The mesh degradation allocation unit projects coordinates based on the centroid. As the extreme point, calculate the true physical contact imprint. radial distance from each balance wheel to the pole The mesh is downgraded and divided into core-driven meshes, edge-assisted meshes, and flexible follower meshes, which together constitute an effective mesh set. ; Step S4: When the effective grid sets of adjacent packages produce an intersection region, the boundary interference dynamic arbitration unit calculates the momentum contribution weight of the pendulum wheel to each package within the intersection region, performs control arbitration, or sets it as a non-powered isolation zone. Step S5: The torque balance differential speed calculation unit calculates the torque balance differential speed based on the center of gravity projection coordinates. coefficient of friction with the bottom A moment balance equation is constructed with the constraint of eliminating yaw spin, and the equation is solved to generate a solution for the effective mesh set. The differential thrust matrix; Step S6: The matrix-style independent execution module executes the final drive commands for multiple packages in parallel at the same time according to the differential thrust matrix.

[0013] According to the above technical solution, in step S2, the actual physical contact imprint is calculated. With the projection coordinates of the center of gravity The specific method is as follows: By traversing the high-density point cloud tensor, filtering out The data characterizing the overhanging or unattached portions of the package allows for the effective delineation of the actual physical contact imprint. Strictly smaller than the two-dimensional bounding rectangle of the package; Calculate the projected coordinates of the centroid using the formula ; ; in, The total number of data points within the point cloud tensor. For the first The spatial coordinates of the points; The mass weight density mapped to this coordinate; For contact step function, when hour Otherwise .

[0014] According to the above technical solution, the method for calculating the momentum contribution and performing arbitration in step S4 further includes: For any balance wheel within the intersection region Get the package With adjacent packages Expected target shunt motion vector and ; Calculate the balance wheel separately For the package Momentum contribution weight and the package Momentum contribution weight ,in This represents the vector of the maximum static friction force of the balance wheel. This is the distance attenuation coefficient. and Balance wheel Distance from package With packages The physical distance of the center of gravity projection coordinates; Setting an arbitration tolerance threshold And execute the comparison logic: when At that time, the balance wheel Control is assigned to the effective grid set with the higher momentum contribution weight; when If the system determines that a struggle over the intersection could easily cause the package to tear in the opposite direction, the system will forcibly strip both parties of their control and move the balance wheel. Set it to a non-powered freewheel state, so that it acts as a physical buffer zone between the two packages in the intersection area.

[0015] According to the above technical solution, in step S5, based on the centroid projection coordinates... The specific methods for constructing torque balance equations include: Obtain the macroscopic basic translational force vector required to divert packages to the target compartment. And the target correction torque required to correct attitude yaw ; Projected coordinates by the center of gravity Using the origin as the starting point, calculate the effective mesh set. Inner The positional lever vector of the balance wheel ; Establish a system of matrix equilibrium equations: ; ; in, The allocation to the first [unit] obtained by the overall control module The differential thrust vector of each balance wheel; By solving the above set of matrix equilibrium equations, a differential thrust matrix is ​​generated to construct a non-uniform force field, ensuring that the resultant force vector output for the package accurately passes through the dynamic center of gravity and generates the desired torque.

[0016] According to the above technical solution, the specific method for dividing the mesh downgrade into a core-driven mesh, an edge-assisted mesh, and a flexible-following mesh in step S3 is as follows: Set the first radial threshold With the second radial threshold ,and ; When radial distance At that time, it is determined that the balance wheel is located in the core pressure zone of the center of gravity, and is assigned as the core drive grid, allowing the output of full-power main drive lateral thrust; when When it is assigned to the edge auxiliary grid, its output thrust upper limit decreases linearly with the proportion of distance from the center of gravity, and is only used to output attitude correction torque; when When the balance wheel is located at the edge of the wrap-around suspension or light pressure, it is assigned to a flexible following grid, forcing its steering angle to follow the core drive grid, but the motor output torque drops to a following state that approaches zero.

[0017] Compared with existing technologies, the beneficial effects achieved by this invention are as follows: by accurately extracting the real physical contact imprints and three-dimensional dynamic center of gravity of packages through multi-dimensional perception, and innovatively introducing grid adaptive degradation allocation, anti-tear boundary interference dynamic arbitration, and anti-yaw moment balance calculation mechanism, it not only fundamentally eliminates the technical risks of eccentric parts spinning in place, soft packages rolling under force, and mechanical interference and tearing of adjacent packages under dense concurrent conditions, but also realizes high-density, safe, and parallel diversion of massive irregularly shaped packages. At the same time, it effectively avoids the redundant power consumption caused by blind full-power output, and greatly improves the overall processing limit of logistics sorting lines and the service life of equipment. Attached Figure Description

[0018] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of the system module composition of the present invention; Figure 2 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example 1 Please see Figure 1 This invention provides a multi-objective parallel diversion system for a logistics swing wheel sorting machine. The system includes a multi-dimensional perception module, a unified control module, and a matrix-style independent execution module.

[0021] The multi-dimensional sensing module is configured to acquire the three-dimensional depth point cloud data and dynamic quality distribution characteristics of the package in real time before the package enters the diversion area. In actual industrial settings, reflective tape or black woven bag materials used for packaging can easily cause optical sensors to malfunction. To ensure undistorted acquisition of 3D depth point cloud data, the multi-dimensional sensing module includes a 3D depth camera array and a matrix-style dynamic weighing unit. The 3D depth camera array is configured as a top-down scanning conveyor belt, outputting spatial coordinates. The array utilizes high-density point cloud tensors. It preferably employs a multi-view gantry arrangement based on the fusion of structured light and Time-of-Flight (ToF) technology, with the camera trigger frequency strictly synchronized with the conveyor belt encoder to eliminate... The image shows the trailing shadow under high-speed motion conditions. Simultaneously, the matrix-type dynamic weighing unit is embedded below the conveyor belt in the preparation area to collect a dynamic pressure distribution heatmap of the package as it passes through, and performs spatiotemporal registration with the high-density point cloud tensor to output multidimensional feature data containing mass distribution weights. Through rigid coordinate system transformation, the pressure grid data at the bottom is mapped one by one to the point cloud projection surface above, achieving decoupling and binding of physical mass and geometric shape.

[0022] The overall control module, which is network-connected to the multi-dimensional perception module, is configured to extract the real physical contact imprints between the package and the conveyor belt based on the three-dimensional depth point cloud data by threshold truncation, and calculate the three-dimensional dynamic center of gravity projection coordinates of the package in combination with the dynamic mass distribution characteristics. Then, using the center of gravity projection coordinates as the pole, the area covered by the real physical contact imprints is mapped and downgraded into an effective set of meshes including core driving meshes, edge auxiliary meshes, and flexible following meshes.

[0023] This process is the core difference between this system and the traditional coarse control of the circumscribed rectangle. To achieve the above functions, in this embodiment of the invention, the overall control module includes a physical imprint and centroid calculation unit, a mesh degradation allocation unit, a boundary interference dynamic arbitration unit, and a torque balance differential speed calculation unit.

[0024] The physical imprint and centroid calculation unit is used to filter out suspended geometry in the 3D point cloud, extracting only the point cloud that conforms to the conveyor belt surface to generate a real physical contact imprint, and calculating the 3D dynamic centroid projection coordinates. When performing the task of filtering out suspended geometry, the system sets a bottom surface contact height threshold (preferably above the highest physical vertex of the balance wheel). This process filters out data above a certain height. This operation effectively eliminates interference from the sides of the wrapping or irregular protrusions on the contact area calculation, restoring the true dynamic stress cross section.

[0025] The grid degradation allocation unit is used to divide the balance wheel covered by the real physical contact imprint into the core driving grid, the edge auxiliary grid, and the flexible following grid according to the radial distance from the center of gravity, based on the balance wheel's physical coordinates. This gradient division avoids excessive thrust from the balance wheel far from the center of gravity, which could cause a lever-like tumble.

[0026] The boundary interference dynamic arbitration unit is used to calculate the momentum contribution weight of the pendulum wheel in the overlapping area to each package when the actual physical contact imprints of two adjacent packages are extremely close, causing mesh overlap, and to perform arbitration on control ownership or non-powered isolation.

[0027] The torque balance differential calculation unit is used to generate the differential thrust matrix that makes the resultant force vector pass through the center of gravity, based on the target deflection angle and the center of gravity projection coordinates.

[0028] For example, in this embodiment of the invention, the overall control module is further used to establish a torque balance equation based on the center of gravity coordinates and calculate the differential thrust matrix allocated to each grid. Before solving the torque balance equation, it is necessary to overcome the friction blind box problem caused by different packaging materials. Therefore, the overall control module further includes a material friction inversion and estimation unit, used to match a preset material library according to the surface texture features of the three-dimensional depth point cloud data, and combine the dynamic mass distribution features to invert and estimate the non-uniform sliding friction coefficient matrix of the bottom of the package, and input the matrix as a basic constraint condition into the torque balance equation. By extracting the variance of the local normal vector of the point cloud to evaluate the wrinkle degree and matching the material library, the system can assign different friction upper limit constraints to cardboard boxes, burlap sacks or coated parts to ensure that the differential thrust solved by the equation can be actually executed by the rubber balance wheel at the physical level, without slipping in place.

[0029] The matrix-style independent execution module, connected to the overall control module, is used to independently and non-uniformly drive different effective grid sets at the bottom of multiple concurrent packages simultaneously, based on the differential thrust matrix.

[0030] To achieve millisecond-level independent non-uniform drive and physical tear-resistant arbitration, the matrix-type independent execution module includes several rubber balance wheel micro-units arranged in a two-dimensional matrix array. Each rubber balance wheel micro-unit is equipped with an independent servo direct-drive motor and an electromagnetic deflection clutch. The electromagnetic deflection clutch not only supports independent control of angle and speed but also has a free-wheel energy release mode to cut off power in response to the powerless isolation strip command issued by the boundary interference dynamic arbitration unit. When two packages are vying for control of the same area and are evenly matched, cutting off power creates a free wheel, which can smoothly glide over the area using its own inertia, fundamentally preventing tearing accidents at the bottom of the package caused by the motor forcibly pulling.

[0031] Example 2 Please see Figure 2 Based on the multi-objective parallel diversion system of the logistics wheel sorting machine provided in Embodiment 1 above, this embodiment further provides a multi-objective parallel diversion method for the logistics wheel sorting machine of this system. Addressing the problems of package damage and missorting caused by fixed safety distances and blind output under full load in traditional methods, the method starts from the three-dimensional morphology and dynamic characteristics of the package's bottom layer, specifically including the following steps: Step S1: The multi-dimensional sensing module collects real-time data on the first [item] entering the diversion preparation area. The high-density point cloud tensor and dynamic pressure distribution data of each package were analyzed, and the bottom friction coefficient was obtained based on surface material matching. and current speed of movement ; For example, in high-speed operations in industrial settings (e.g., conveyor belt speeds reaching...), The initial motion state of the package has a decisive influence on its momentum distribution after entering the diversion zone. Traditional vision systems can only acquire two-dimensional contours and cannot perceive materials, causing the control system to treat rough cardboard boxes and extremely smooth waterproof woven bags the same, which can easily cause the woven bags to slip on the balance wheel.

[0032] Therefore, in this embodiment, after obtaining the high-density point cloud tensor in step S1, the system extracts the surface texture features of the bottom of the package using a point cloud processing algorithm (for example, by calculating the variance of the local normal vectors of the point cloud to characterize roughness). Subsequently, this roughness feature is substituted into a preset material library for matching, thereby providing the material for the first... Each package is matched with a specific bottom friction coefficient. (For example, ordinary corrugated cardboard boxes) Preferred value is Between, and the smooth woven bag Preferred value is (between). The coefficient of friction. With current speed These form the core boundary conditions for constructing the dynamic torque balance equations.

[0033] Step S2: Set the bottom surface contact height threshold for the physical imprint and center of gravity calculation unit. In the high-density point cloud tensor Slice the point set and project it to generate a real physical contact imprint. Simultaneously, combining the dynamic pressure distribution data, the three-dimensional dynamic center of gravity of the package is calculated and projected onto the balance wheel plane to obtain the center of gravity projection coordinates. ; Further, in step S2, the actual physical contact imprint is calculated. With the projection coordinates of the center of gravity The specific method is as follows: filter out high-density point cloud tensors by traversing them. The data characterizing the overhanging or unattached portions of the package allows for the effective delineation of the actual physical contact imprint. It is strictly smaller than the two-dimensional bounding rectangle of the package.

[0034] When handling soft-pack or irregularly shaped items, their two-dimensional circumscribed rectangles often contain a large amount of overhanging areas that do not physically contact the conveyor belt. Including the swing wheels in these air-filled areas in the drive system would directly cause the actuator motor to idle and could even malfunction subsequent packaging. Therefore, this step introduces a bottom surface contact height threshold. In specific implementation conditions, It is not absolute zero, but rather preferably set above the apex of the balance wheel. Between. The physical significance of this setting is that it can just tolerate the slight elastic sinking of the rubber balance wheel and the mechanical vibration of the belt when a heavy load passes by, while resolutely filtering out side overhang structures with a height greater than this threshold. This ensures the generated imprint... It reflects the actual stress cross section between the package and the equipment.

[0035] Calculate the projected coordinates of the centroid using the formula ; ; in, The total number of data points within the point cloud tensor. For the first The spatial coordinates of the points; The mass weight density mapped to this coordinate; For contact step function, when hour Otherwise .

[0036] For heterogeneous packages (such as cardboard boxes with heavy iron components at one end and bubble wrap at the other), there is a significant deviation between their geometric center and physical center of gravity. Traditional geometric center calculations lead to severely uneven stress distribution during sorting. The aforementioned discrete integral formula cleverly integrates visual shape and mass distribution. Among these, the contact step function... It acts as a spatial filter, forcing any point that is not in contact with the ground (i.e. The contribution of the point to the center of gravity is zero; while (Derived from the pressure-thermal map collected by the matrix dynamic weighing unit) this assigns the actual physical weight weight to the contact points. The pole coordinates calculated using this formula... It encapsulates the actual dynamic spin axis, laying a solid data foundation for completely eliminating yaw and spin.

[0037] Step S3: The mesh degradation allocation unit projects coordinates based on the centroid. As the extreme point, calculate the true physical contact imprint. radial distance from each balance wheel to the pole The mesh is downgraded and divided into core-driven meshes, edge-assisted meshes, and flexible follower meshes, which together constitute an effective mesh set. ; Furthermore, in step S3, the specific method for dividing the mesh downgrade into a core-driven mesh, an edge-assisted mesh, and a flexible-following mesh is as follows: setting a first radial threshold. With the second radial threshold When radial distance When the balance wheel is located in the core pressure zone surrounding the center of gravity, it is assigned to the core drive grid, allowing full-power main drive lateral thrust to be output; when When assigned as an edge auxiliary mesh, its output thrust upper limit decreases linearly with respect to the distance from the center of gravity, and is used only for output attitude correction torque; when When the balance wheel is located at the edge of the wrap-around suspension or light pressure, it is assigned to a flexible following grid, forcing its steering angle to follow the core drive grid, but the motor output torque drops to a following state that approaches zero.

[0038] Traditional sorting machines typically control all sensing wheels under the package to move in the same direction at 100% full power when starting the sorting process. This can cause the wheels at the package edge, where the normal pressure is minimal, to slip due to excessive instantaneous thrust, even tearing soft packaging (such as plastic film express bags). Simultaneously, the forces at the edge can easily generate huge parasitic torques, causing the package to tumble. This embodiment addresses this by setting a first radial threshold. (Preferably set to 30%~40% of the equivalent radius of the actual contact imprint of the package) and the second radial threshold. (Preferred setting is 80%~90%), a differential soft landing potential field is constructed from the inside out inside the force-bearing surface.

[0039] Located in the core pressure zone ( Because the balance wheel bears the greatest positive gravity and has the most stable friction, it bears the heavy responsibility of outputting the main basic translational force (i.e., full power) to ensure the efficiency of the change of direction.

[0040] Located in the intermediate transition zone ( The edge-assisted grid of the non-uniform electronic differential soft thrust mechanism, whose output thrust is forced to decrease linearly (e.g., smoothly reduced from full power to about 15%), provides fine-tuning attitude correction torque through the gentle edge support of the non-uniform electronic differential soft thrust mechanism, much like the gentle edge support of a human hand when carrying something.

[0041] Located at the polar edge ( The flexible, following mesh, while physically in contact with the package, bears almost no weight; forcibly outputting power could easily tear the package. Therefore, this system instructs its deflection angle to follow the main group, but the servo motor output torque drops to a near-zero follow-up support state. This mechanism not only protects the package but also effectively avoids the redundant and useless power consumption and component wear mentioned in the background art.

[0042] In addressing the industry bottleneck mentioned in the background technology—the requirement to maintain a large physical safety distance between packages—this system breaks through the limitation of physical distance through step S4, namely: Step S4: When the effective grid sets of adjacent packages produce an intersection region, the boundary interference dynamic arbitration unit calculates the momentum contribution weight of the pendulum wheel to each package within the intersection region, performs control arbitration, or sets it as a non-powered isolation zone. Furthermore, the method for calculating the momentum contribution and performing arbitration in step S4 further includes: For any balance wheel within the intersection region Get the package With adjacent packages Expected target shunt motion vector and ; Calculate the balance wheel separately For the package Momentum contribution weight and the package Momentum contribution weight ,in This represents the vector of the maximum static friction force of the balance wheel. This is the distance attenuation coefficient. and Balance wheel Distance from package With packages The physical distance of the center of gravity projection coordinates; Setting an arbitration tolerance threshold And execute the comparison logic: when At that time, the balance wheel Control is assigned to the effective grid set with the higher momentum contribution weight; when If the system determines that a struggle over the intersection could easily cause the package to tear in the opposite direction, the system will forcibly strip both parties of their control and move the balance wheel. Set it to a non-powered freewheel state, so that it acts as a physical buffer zone between the two packages in the intersection area.

[0043] For example, in high-density parallel distribution, two packages are very likely to be connected end to end or placed side by side. At this time, if the balance wheel in the overlapping area receives conflicting commands to the left and right at the same time, it will cause the bottom of the package to be subjected to a huge shearing and pulling force, which may result in the trajectory going out of control or even tearing the bottom cover of the carton.

[0044] The arbitration formula of this invention introduces a nonlinear exponential decay term. In industrial calibration, the distance attenuation coefficient... The optimal setting is 0.08. The physical meaning of this is that the farther the balance wheel is from the center of gravity, the more its small lateral force will be converted into a huge, harmful yaw torque due to the leverage effect of its long lever arm. Therefore, the system, through exponential decay, deprives the far-end balance wheel of its competitive weight, prioritizing control over another grid closer to the center of gravity.

[0045] In addition, an arbitration tolerance threshold is set. (Preferably set at 5%~8% of the sum of the momentum of the two packages). When the difference At this point, the balance wheel is at the absolute physical boundary between the two points; either side taking it away would damage the other. The system then triggers hardware-level fallback protection, cutting off power to the electromagnetic deflection clutch at that point. The balance wheel, acting like a ball bearing, provides only upward support, smoothly sliding across the boundary due to its own inertia. This transforms rigid grid conflicts into flexible physical isolation, fundamentally solving the mechanical interference problem in high-density concurrent processing.

[0046] After resolving the interference problem, the system enters the final stage of driving torque distribution.

[0047] Step S5: The torque balance differential speed calculation unit calculates the torque balance differential speed based on the center of gravity projection coordinates. coefficient of friction with the bottom A moment balance equation is constructed with the constraint of eliminating yaw spin, and the equation is solved to generate a solution for the effective mesh set. The differential thrust matrix.

[0048] Furthermore, in step S5, based on the centroid projection coordinates... The specific methods for constructing torque balance equations include: Obtain the macroscopic basic translational force vector required to divert packages to the target compartment. And the target correction torque required to correct attitude yaw ; Projected coordinates by the center of gravity Using the origin as the starting point, calculate the effective mesh set. Inner The positional lever vector of the balance wheel ; Establish a system of matrix equilibrium equations: ; ; in, The allocation to the first [unit] obtained by the overall control module The differential thrust vector of each balance wheel; By solving the above set of matrix equilibrium equations, a differential thrust matrix is ​​generated to construct a non-uniform force field, ensuring that the resultant force vector output for the package accurately passes through the dynamic center of gravity and generates the desired torque.

[0049] Traditional sorting machines apply the same thrust to all balance wheels, causing the point of application of the resultant force to deviate from the true center of gravity, inevitably resulting in parasitic spin.

[0050] To solve for the optimal differential thrust vector set In this embodiment of the invention, the overall control module incorporates a weighted Moore-Penrose pseudo-inverse matrix solver. Furthermore, during the solution process, the algorithm must call the friction coefficient matrix estimated in step S1. As a hard boundary constraint (i.e., the upper limit of the thrust of any single wheel cannot exceed...), ,in (This is the positive pressure at that point). Thus, the differential matrix output by the system not only perfectly eliminates the yaw torque mathematically, but also ensures that it will never exceed the grip limit of the rubber balance wheel and slip, thus preventing the irregularly shaped eccentric component from changing direction as smoothly as if it were gliding on a track.

[0051] Step S6: The matrix-style independent execution module executes the final drive commands for multiple packages in parallel at the same time according to the differential thrust matrix.

[0052] Finally, the matrix-style independent execution module located at the execution layer converts the calculated differential thrust matrix into the underlying electrical signals of each rubber balance wheel micro-unit via a high-speed industrial bus. Simultaneously (synchronous jitter) The array of hundreds of servo direct-drive motors each outputs a specified speed and torque. Combined with the precise angle of the electromagnetic deflection clutch, this instantly creates multiple non-uniform driving potential fields above the conveyor belt, each perfectly matching its own wrapping dynamics. This completely overturns the traditional one-size-fits-all, crude approach, achieving truly high-density, multi-objective, safe, parallel flow distribution.

[0053] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0054] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0055] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0056] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. A multi-objective parallel diversion system for a logistics swing wheel sorting machine, characterized in that, The system includes a multi-dimensional perception module, a unified control module, and a matrix-style independent execution module; The multi-dimensional sensing module is configured to acquire the three-dimensional depth point cloud data and dynamic quality distribution characteristics of the package in real time before the package enters the diversion area. The overall control module, network-connected to the multi-dimensional perception module, is configured to extract the actual physical contact imprints between the package and the conveyor belt based on the three-dimensional depth point cloud data through threshold truncation, and calculate the three-dimensional dynamic center of gravity projection coordinates of the package in combination with the dynamic mass distribution characteristics. Then, using the center of gravity projection coordinates as the pole, the area covered by the actual physical contact imprints is mapped and downgraded into an effective set of grids including core driving grids, edge auxiliary grids, and flexible following grids. The overall control module is also used to establish a torque balance equation based on the center of gravity coordinates and calculate and solve the differential thrust matrix allocated to each grid. The matrix-style independent execution module, connected to the overall control module, is used to independently and non-uniformly drive different effective grid sets at the bottom of multiple concurrent packages simultaneously, based on the differential thrust matrix.

2. The multi-target parallel diversion system for a logistics swing wheel sorting machine according to claim 1, characterized in that: The multi-dimensional sensing module includes a three-dimensional depth camera array and a matrix-type dynamic weighing unit; the three-dimensional depth camera array is configured to scan a conveyor belt from above, and its output includes spatial coordinates. The high-density point cloud tensor; the matrix-type dynamic weighing unit is embedded below the conveyor belt in the preparation area to collect dynamic pressure distribution heat map of the package as it passes by, and to perform spatiotemporal registration with the high-density point cloud tensor to output multidimensional feature data containing mass distribution weights.

3. The multi-target parallel diversion system for a logistics swing wheel sorting machine according to claim 1, characterized in that: The overall control module includes a physical imprint and center of gravity calculation unit, a grid degradation allocation unit, a boundary interference dynamic arbitration unit, and a torque balance differential calculation unit. The physical imprint and center of gravity calculation unit is used to filter out suspended geometry in the three-dimensional point cloud, extract only the point cloud that fits the surface of the conveyor belt to generate real physical contact imprints, and calculate the three-dimensional dynamic center of gravity projection coordinates. The grid degradation allocation unit is used to divide the balance wheel covered by the real physical contact imprint into the core driving grid, the edge auxiliary grid and the flexible following grid according to the radial distance gradient from the center of gravity, based on the physical coordinates of the balance wheel. The boundary interference dynamic arbitration unit is used to calculate the momentum contribution weight of the pendulum wheel in the overlapping area to each package when the actual physical contact imprints of two adjacent packages are extremely close, causing mesh overlap, and to perform arbitration on control ownership or non-powered isolation. The torque balance differential speed calculation unit is used to generate the differential thrust matrix that makes the resultant force vector pass through the center of gravity based on the target deflection angle and the center of gravity projection coordinates.

4. The multi-target parallel diversion system for a logistics swing wheel sorting machine according to claim 3, characterized in that: The matrix-type independent execution module includes several rubber balance wheel micro-units arranged in a two-dimensional matrix array. Each rubber balance wheel micro-unit is equipped with an independent servo direct drive motor and an electromagnetic deflection clutch. The electromagnetic deflection clutch supports independent control of angle and speed, and has a free wheel energy release mode to cut off power in response to the non-powered isolation zone command issued by the boundary interference dynamic arbitration unit.

5. The multi-target parallel diversion system for a logistics swing wheel sorting machine according to claim 1, characterized in that: The overall control module further includes a material friction inversion and prediction unit, which is used to match a preset material library according to the surface texture features of the three-dimensional depth point cloud data, and invert and predict the non-uniform sliding friction coefficient matrix of the bottom of the package in combination with the dynamic mass distribution features, and input the matrix as a basic constraint condition into the torque balance equation.

6. A multi-objective parallel diversion method for a logistics swing wheel sorting machine using the system described in any one of claims 1-5, characterized in that: The method includes the following steps: Step S1: The multi-dimensional sensing module collects real-time data on the first [item] entering the diversion preparation area. The high-density point cloud tensor and dynamic pressure distribution data of each package were analyzed, and the bottom friction coefficient was obtained based on surface material matching. and current speed of movement ; Step S2: Set the bottom surface contact height threshold for the physical imprint and center of gravity calculation unit. In the high-density point cloud tensor Slice the point set and project it to generate a real physical contact imprint. Simultaneously, combining the dynamic pressure distribution data, the three-dimensional dynamic center of gravity of the package is calculated and projected onto the balance wheel plane to obtain the center of gravity projection coordinates. ; Step S3: The mesh degradation allocation unit projects coordinates based on the centroid. As the extreme point, calculate the true physical contact imprint. radial distance from each balance wheel to the pole The mesh is downgraded and divided into core-driven meshes, edge-assisted meshes, and flexible follower meshes, which together constitute an effective mesh set. ; Step S4: When the effective grid sets of adjacent packages produce an intersection region, the boundary interference dynamic arbitration unit calculates the momentum contribution weight of the pendulum wheel to each package within the intersection region, performs control arbitration, or sets it as a non-powered isolation zone. Step S5: The torque balance differential speed calculation unit calculates the torque balance differential speed based on the center of gravity projection coordinates. coefficient of friction with the bottom A moment balance equation is constructed with the constraint of eliminating yaw spin, and the equation is solved to generate a solution for the effective mesh set. The differential thrust matrix; Step S6: The matrix-style independent execution module executes the final drive commands for multiple packages in parallel at the same time according to the differential thrust matrix.

7. A multi-objective parallel diversion method for a logistics swing wheel sorting machine according to claim 6, characterized in that: In step S2, the actual physical contact imprint is calculated. With the projection coordinates of the center of gravity The specific method is as follows: By traversing the high-density point cloud tensor, filtering out The data characterizing the overhanging or unattached portions of the package allows for the effective delineation of the actual physical contact imprint. Strictly smaller than the two-dimensional bounding rectangle of the package; Calculate the projected coordinates of the centroid using the formula ; ; in, The total number of data points within the point cloud tensor. For the first The spatial coordinates of the points; This represents the mass weight density mapped to this coordinate. For contact step function, when hour Otherwise .

8. A multi-objective parallel diversion method for a logistics swing wheel sorting machine according to claim 7, characterized in that: The method for calculating the momentum contribution and performing arbitration in step S4 further includes: For any balance wheel in the intersection region , obtaining a package With adjacent packages Expected target split motion vector And ; Calculate the balance wheel separately For the package Momentum contribution weight and the package Momentum contribution weight ,in This represents the vector of the maximum static friction force of the balance wheel. This is the distance attenuation coefficient. and Balance wheel Distance from package With packages The physical distance of the center of gravity projection coordinates; Setting an arbitration tolerance threshold And execute the comparison logic: when At that time, the balance wheel Control is assigned to the effective grid set with the higher momentum contribution weight; when If the system determines that a struggle over the intersection could easily cause the package to tear in the opposite direction, the system will forcibly strip both parties of their control and move the balance wheel. Set it to a non-powered freewheel state, so that it acts as a physical buffer zone between the two packages in the intersection area.

9. A multi-objective parallel diversion method for a logistics swing wheel sorting machine according to claim 6, characterized in that: In step S5, based on the centroid projection coordinates The specific methods for constructing torque balance equations include: Obtain the macroscopic basic translational force vector required to divert packages to the target slot. And the target correction torque required to correct attitude yaw ; Projected coordinates by the center of gravity Using the origin as the starting point, calculate the effective mesh set. Inner The positional lever vector of the balance wheel ; Establish a system of matrix equilibrium equations: ; ; in, The allocation to the first [unit] obtained by the overall control module The differential thrust vector of each balance wheel; By solving the above matrix equilibrium equations, a differential thrust matrix is ​​generated to construct a non-uniform force field, ensuring that the resultant force vector output for the package accurately passes through the dynamic center of gravity and generates the desired torque.

10. A multi-objective parallel diversion method and system for a logistics swing wheel sorting machine according to claim 6, characterized in that: In step S3, the specific method for dividing the mesh downgrade into a core-driven mesh, an edge-assisted mesh, and a flexible-following mesh is as follows: Set the first radial threshold With the second radial threshold ,and ; When radial distance At that time, it is determined that the balance wheel is located in the core pressure zone of the center of gravity, and is assigned as the core drive grid, allowing the output of full-power main drive lateral thrust; when When it is assigned to the edge auxiliary grid, its output thrust upper limit decreases linearly with the proportion of distance from the center of gravity, and is only used to output attitude correction torque; when When the balance wheel is located at the edge of the wrap-around suspension or light pressure, it is assigned to a flexible following grid, forcing its steering angle to follow the core drive grid, but the motor output torque drops to a following state that approaches zero.