Packaging method and system based on millimeter wave radar

Abstract

The invention relates to a packaging method and system based on millimeter wave radar. The method comprises the steps that three-dimensional contour scanning is conducted on packaged goods on a tray through a millimeter wave radar, and echo signal data including the tray size, the stacking height, the surface flatness and the corner position are obtained; generating a three-dimensional point cloud model of the cargo based on the echo signal; analyzing the three-dimensional point cloud model, and extracting cargo parameters; the cargo parameters comprise the overall length, the width, the height, the corner position and the surface flatness, and points with the curvature larger than a preset curvature threshold value are recognized to serve as corner points; based on the cargo parameters, a path planning strategy combining longitudinal layering and a transverse track is adopted, and a film winding packaging track is generated; and when it is detected that the goods shake or the film winding head deviates, the film winding packaging track is dynamically corrected, so that self-adaptive adjustment of the packaging path is achieved. The packaging quality and stability are improved, and material consumption and manual intervention are reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of packaging equipment and industrial automation control technology, in particular to a packaging method and system based on millimeter wave radar. BACKGROUND

[0002] With the development of e-commerce logistics, warehouse distribution and intelligent factory, the wrapping film packaging of pallet goods has become an important process to prevent goods from scattering, getting wet and being damaged by bumps during handling, turnover and transportation.

[0003] In the prior art, one type of equipment basically does not have the perception ability of the three-dimensional contour of the goods, and relies on manual estimation of the size of the pallet and the height of the goods to perform packaging through a pre-set fixed program. This type of scheme cannot dynamically adjust the wrapping film path according to the actual stacking height, the shape of the goods and the surface unevenness, and is prone to problems such as incomplete coverage of the top, too high or too low local overlap rate, and insufficient protection of the corner area, resulting in poor packaging, material waste and insufficient transportation safety.

[0004] Another improved scheme attempts to introduce photoelectric sensors, laser ranging or two-dimensional cameras and other devices to detect the height or contour of the goods to assist in determining the wrapping film start and stop height or the approximate contour range. However, optical and laser sensors are sensitive to environmental light, dust, smoke and reflective surfaces, and in complex lighting, dusty and metal pallet, reflective film and other working conditions in warehouse, loading and unloading sites, measurement signal saturation, shadow blocking or reflection interference may occur, resulting in unstable contour recognition. In particular, for mixed stacking of goods of different materials such as cartons, plastic bags and metal boxes, existing optical solutions cannot reliably detect goods of multiple materials and multiple shapes. SUMMARY

[0005] Therefore, the present application provides a packaging method and system based on millimeter wave radar, which can realize three-dimensional contour perception, clutter suppression, adaptive path planning and dynamic deviation correction control of pallet goods based on millimeter wave radar in complex goods stacking and complex environmental conditions, to improve packaging quality and stability, and reduce material consumption and manual intervention.

[0006] To solve the above technical problems, the present application provides a packaging method based on millimeter wave radar, characterized in that it is applied to a packaging system, the packaging system includes a pallet loaded with goods, a wrapping head controlled by a mechanical arm, a millimeter wave radar arranged on the wrapping head, and a rotating structure supporting the rotation of the pallet, and the packaging method comprises: S1, scanning the three-dimensional contour of the packaging goods on the pallet by the millimeter wave radar to obtain echo signal data containing the size of the pallet, the stacking height, the surface flatness and the corner position; including: S11. Perform an initial scan of the packaging scene in a targetless state to build an initial background model; S12. Based on the initial background model, a difference model is constructed through inter-frame difference to update the background and reduce static clutter weights; S13. Based on the difference model, candidate target regions are obtained; wherein, the newly acquired data of each frame is differentially analyzed with the current background, the background weight is maintained or enhanced for regions with a difference amplitude lower than a preset difference threshold, and regions with a difference amplitude not less than the difference threshold are selected as candidate target regions. S14. Calculate the mean and standard deviation of the signal strength within the candidate target area; S15. Calculate the detection threshold based on the mean and standard deviation, and based on the reflectance coefficient of the cargo material in the candidate target area; S16. Dynamically adjust the detection threshold according to the material of the goods in the candidate target area; S17. Retain the region within the candidate target area whose signal intensity is higher than the detection threshold as the purified target signal; S18. After performing signal smoothing and denoising and phase adjustment on the target signal, the cargo contour phase signal used for subsequent three-dimensional point cloud modeling is obtained. S2. Based on the echo signal, generate a three-dimensional point cloud model of the cargo; S3. Analyze the three-dimensional point cloud model and extract cargo parameters; the cargo parameters include overall length, width, height, corner positions and surface flatness, and identify points with curvature greater than a preset curvature threshold as corner points; S4. Based on the cargo parameters, a path planning strategy combining longitudinal layering and lateral trajectory is adopted to generate a wrap-around packaging trajectory; including: S41. Divide the packaging path into multiple height levels according to the height of the goods, and raise the film wrapping head layer by layer along the lifting direction so that the height of the film wrapping head is adjusted with the change of the level. S42. In each height level, a spiral trajectory of the wrapping head relative to the goods is generated based on the center coordinates of the pallet, so that the wrapping head forms a transverse spiral winding trajectory relative to the goods in the transverse direction. S43. Increase the number of transverse windings of the film-wrapping head in the height level that includes the corner points; S5. When the shaking of goods or the offset of the wrapping head is detected, the wrapping trajectory is dynamically corrected to achieve adaptive adjustment of the packaging path.

[0007] In one embodiment of the present invention, based on the initial background model, a differential model is constructed through inter-frame differencing to update the background and reduce static clutter weights. Based on the differential model, candidate target regions are obtained, including: An initial scan of the packaging scene is performed without a target to construct an initial background model. : ; in, This refers to continuous scan frame data from a millimeter-wave radar. t represents a time frame; The coordinates are spatial coordinates, and the values ​​are signal strength. N is the number of frames; When new data arrives in each frame, the background is updated through inter-frame differencing to reduce static clutter weights. ; in, , is the background retention coefficient; For indicator functions; The preset difference threshold; ; Difference between current frame and background for: ; in, ,when The value is 1 if the condition is met; otherwise, it is 0, meaning only the static region is updated. Candidate target regions are obtained through background subtraction. ; .

[0008] In one embodiment of the present invention, a detection threshold is calculated based on the mean and standard deviation, and on the reflectance coefficient of the cargo material in the candidate target region. The detection threshold is then dynamically adjusted according to the cargo material in the candidate target region, including: The average signal strength within the candidate target region and standard deviation : ; ; in, The number of pixels in the candidate target region; Candidate target regions; Based on material reflectivity Dynamic adjustment detection : ; in, It dynamically adjusts based on the signal-to-noise ratio, especially at low signal-to-noise ratios. ; Regions within the candidate target area with signal intensity higher than the detection threshold are retained as purified target signals. : .

[0009] In one embodiment of the present invention, signal smoothing and denoising, and phase adjustment are performed on the target signal, including: Smoothing of multiple consecutive frames of radar data in the target signal over time to suppress transient fluctuations includes: For continuous The frame data is weighted and averaged to suppress instantaneous fluctuations, resulting in the weighted average target signal. : ; in , where is the weight, which decays over time; To retrieve the frame number; This corresponds to taking the frame from the current frame to the earliest frame within the same frame; The purified target signal; In the spatial dimension, median filtering is applied to the weighted averaged target signal to remove isolated clutter points and preserve contour edge details, including: right The signal strength is obtained by taking the median of the neighborhood and removing isolated clutter points. : ; in, for of Neighborhood; Represent the median function; The signal is filtered in the spatiotemporal domain, and t is a time frame; These are the spatial coordinates of the radar scanning plane; The phase information of millimeter-wave signals is used in conjunction with a pre-defined material phase feature library to demodulate the target echo, eliminating interference signals whose phase characteristics do not match the target, including: Calculate the difference between the phase and the background phase at the same location. : ; in, The background phase model is obtained through the initial scan; The phase signal of the radar echo, in radians; A preset material phase feature library includes metal phase offset and cardboard box phase offset; among which, the metal phase offset is: ; The phase offset of the cardboard box is: ; Preserve the signal that matches the target phase : ; in, The phase range of the target cargo.

[0010] In one embodiment of the present invention, generating a three-dimensional point cloud model of the cargo based on the echo signal includes: Input radar simulated echo signal ; Among them, carrier frequency ; Signal strength; For carrier frequency; It is a time variable; The instantaneous phase is the angle that changes over time; By sampling rate Discretization yields: ; in, For different sampling points; The distance calculation for a 3D point cloud is as follows: Based on signal flight time ,distance Frequency shift extracted by Fourier transform ,calculate And thus obtain ;in: ; ; ; in, For relative velocity, For the speed of light, when dealing with a static target, it simplifies to a direct mapping between distance and frequency; Angle calculation of 3D point cloud: Assuming a 4×4 antenna array with horizontal and vertical phase difference and Horizontal angle and vertical angle Calculation yields: ; ; in, For wavelength, Antenna spacing; 3D point cloud data output: Based on the above calculations, the three-dimensional coordinates of each point can be obtained as follows: .

[0011] In one embodiment of the present invention, the three-dimensional point cloud model is analyzed to extract cargo parameters; the cargo parameters include overall length, width, height, corner positions, and surface flatness, and points with curvature greater than a preset curvature threshold are identified as corner points, including: The 3D point cloud model is analyzed to obtain the 3D point cloud. , This represents the number of point clouds; Size calculation: length ; width ; high ; Corner points are identified by point cloud curvature calculation; points with curvature exceeding a threshold are also identified. ; in, For points in a point cloud, For the number of neighboring points, for normal vector, curvature It is determined to be a corner point; For neighboring points; For metal box goods with sharp corners, mark the corner coordinates to provide a positioning basis for subsequent lateral reinforcement coverage; for goods with uneven surfaces, identify raised or recessed areas, and adjust the height and tension of the robotic arm and wrapping head to avoid missing packages or overly tight packaging that could cause the goods to deform.

[0012] In one embodiment of the present invention, the packaging path is divided into multiple height levels according to the height of the goods, including: height Divided into Layers, height of each layer That is, each layer ≤10cm, the first Floor height ; Generate a spiral trajectory of the wrapping head relative to the goods based on the center coordinates of the pallet, including: The film wrapping head is centered on the tray in each layer. The polar equation for a spiral motion centered on a circle is: ; in, , where is the initial radius, covering the maximum lateral dimension of the cargo; , where is the spiral step size, controlling the film overlap rate; The number of wraps. .

[0013] Increasing the number of lateral windings of the film head in the height level including the corner points includes: opposite corner points By increasing the number of wrapping cycles at the corresponding height level, the trajectory is corrected as follows: ,exist Add one more wrap within the range.

[0014] In one embodiment of the present invention, it further includes: The lateral step distance between adjacent winding tracks is calculated based on the width of the film used by the winding head, so that the overlap rate between adjacent film strips is within a preset range, so as to avoid excessive overlap causing material waste or insufficient overlap causing missing packages. The tension of the winding head is controlled within the range of 0.3N-5N according to the preset tension parameters, and correction is made when the tension is detected to exceed the range.

[0015] In one embodiment of the present invention, the dynamic correction of the wrapping trajectory when cargo shaking or wrapping head deviation is detected includes: When the millimeter-wave radar detects in real time that the positional offset of the cargo exceeds the positioning accuracy threshold of the robotic arm driving the film-wrapping head, the offset of the cargo in the x and y directions is calculated. , and the trajectory center was changed from Revised to ; Based on the corrected trajectory center, a position adjustment command is sent to the robotic arm to correct the posture of the wrapping head and ensure that the wrapping head is continuously aligned with the outline of the goods; at the same time, a speed adjustment command is sent to the rotating structure supporting the pallet rotation to reduce its rotation speed, reduce the amplitude of goods swaying, and avoid packaging deviation caused by swaying. The film covering position is detected in real time. When the lateral offset of the film winding head exceeds the predetermined value of the film width, the central axis of the robotic arm wrist is controlled to adjust the angle of the film winding head, and the movement trajectory of the robotic arm is corrected simultaneously so that the film covering position conforms to the planned path again. If the lateral offset of the wrapping head exceeds the set safety threshold, the packaging operation will be suspended and a warning will be triggered. The film tension F is detected in real time. When the film tension F deviates from the preset tension F0 and exceeds the range of 0.3N-5N, the rotation speed of the film winding head motor is adjusted. , Set the initial rotation speed to ensure the tension remains stable between 0.3N and 5N; In response to the detection of localized film thickness anomalies leading to insufficient strength, the winding head linear speed is reduced, and the number of winding layers is increased in the abnormal area to compensate for material defects and improve packaging quality.

[0016] The present invention also provides a packaging system based on millimeter-wave radar, comprising: The column is equipped with a lifting mechanism. The robotic arm, configured as a six-axis robotic hand, is connected to the column and achieves lifting and lowering via the lifting mechanism; A film-winding head is located at the end of the robotic arm. The film-winding head includes a mounting frame and a film-winding placement roller, a film-winding head fixed transmission roller, a film-winding head central shaft, and a film-winding head movable adjustment roller arranged in parallel with each other on the mounting frame. The film-winding placement roller is driven by a film-winding motor located on the mounting frame. Millimeter-wave radar is mounted on the mounting bracket; A tray is positioned beside the robotic arm; A rotating structure drives the tray to rotate; A control system for implementing the millimeter-wave radar-based packaging method includes: The echo signal data acquisition module is used to perform three-dimensional contour scanning of packaged goods located on a pallet using millimeter-wave radar, obtaining echo signal data including pallet size, stacking height, surface flatness, and corner positions; including: The initial background model building unit is used to perform an initial scan of the packaging scene in a targetless state and build an initial background model; The background update unit is used to construct a difference model based on the initial background model through inter-frame difference, so as to update the background and reduce the static clutter weights. The candidate target region acquisition unit is used to obtain candidate target regions based on the difference model; wherein, the newly acquired data of each frame is differentially analyzed with the current background, the background weight is maintained or enhanced for regions with a difference amplitude lower than a preset difference threshold, and regions with a difference amplitude not less than the difference threshold are used as candidate target regions. The mean and standard deviation calculation unit is used to calculate the mean and standard deviation of the signal intensity within the candidate target area; The detection threshold calculation unit is used to calculate the detection threshold based on the mean and standard deviation, and based on the reflectance coefficient of the cargo material in the candidate target area; A dynamic adjustment unit is used to dynamically adjust the detection threshold according to the material of the goods in the candidate target area; A purification unit is used to retain regions within the candidate target area where the signal intensity is higher than the detection threshold, as purified target signals. The phase signal acquisition unit is used to perform signal smoothing and denoising and phase adjustment on the target signal to obtain the cargo contour phase signal for subsequent 3D point cloud modeling. A three-dimensional point cloud model acquisition unit is used to generate a three-dimensional point cloud model of the cargo based on the echo signal. The cargo parameter acquisition module is used to parse the three-dimensional point cloud model and extract cargo parameters. The cargo parameters include overall length, width, height, corner positions and surface flatness, and points with curvature greater than a preset curvature threshold are identified as corner points. A wrap-around packaging trajectory generation module is used to generate a wrap-around packaging trajectory based on the cargo parameters, employing a path planning strategy that combines longitudinal layering and lateral trajectory; including: The height adjustment unit is used to divide the packaging path into multiple height levels according to the height of the goods. The wrapping head rises layer by layer along the lifting direction, so that the height of the wrapping head is adjusted according to the level. The transverse spiral winding trajectory generation unit is used to generate a spiral trajectory of the wrapping head relative to the goods based on the center coordinates of the pallet in each height level, so that the wrapping head forms a transverse spiral winding trajectory relative to the goods in the transverse direction. A transverse winding number increasing unit is used to increase the transverse winding number of the film head in the height level including the corner points; An adaptive adjustment module is used to dynamically correct the wrapping trajectory when cargo shaking or wrapping head deviation is detected, so as to achieve adaptive adjustment of the packaging path.

[0017] The technical solution of the present invention has the following advantages compared with the prior art: This invention discloses a packaging method and system based on millimeter-wave radar. The invention constructs an initial background model in a target-free state, updates the background using inter-frame differential to reduce static clutter weights, obtains candidate target regions using background differential, and calculates the mean and standard deviation of signal intensity in these candidate target regions. A detection threshold is dynamically calculated based on the reflectivity of the cargo material. By using an adaptive detection threshold, target regions above the threshold are retained, significantly suppressing static interference such as fixed metal supports and ground reflections. Simultaneously, it considers the reflectivity of different cargo materials such as metal boxes and cardboard boxes, avoiding contour distortion caused by strong reflection overflow and the omission of weak reflection edge signals. This allows for stable acquisition of accurate cargo contour signals even in complex environments.

[0018] This invention employs a weighted average of continuous multi-frame radar data in the temporal dimension and median filtering in the spatial dimension to remove isolated clutter points. Spatiotemporal filtering suppresses instantaneous fluctuations and random noise. Simultaneously, a pre-defined material phase feature library is used to demodulate the radar echo phase, eliminating interference signals whose phase characteristics do not match the target, retaining only echo data within the target's phase range. The final result is a cargo contour phase signal used for 3D point cloud modeling. This multi-stage signal processing significantly improves the signal-to-noise ratio of millimeter-wave echoes, providing high-quality input for subsequent distance and angle calculations based on time-of-flight, frequency offset, and multi-antenna phase differences. This results in a more accurate and continuous 3D point cloud model at cargo edges, corners, and surface details.

[0019] This invention analyzes a 3D point cloud model to calculate the overall length, width, and height of the goods. Based on the point cloud curvature, it identifies corner points with curvature exceeding a preset threshold, while also recognizing protruding and recessed areas. Furthermore, the goods height is divided into multiple height levels. Within each height level, the wrapping head spirals around the center of the pallet, increasing the number of lateral wrapping cycles in height levels containing corner points. This achieves path planning that combines vertical layering with a lateral spiral trajectory. By reinforcing the corners and impact-prone areas, and adjusting the trajectory and wrapping head posture on uneven surfaces, the overall rigidity and impact resistance of the packaging are improved. This prevents goods from scattering or breaking at corners due to vibration or collision during transportation, while also avoiding deformation caused by excessively tight wrapping in certain areas.

[0020] This invention integrates a tension sensor and a film thickness detection device based on the phase difference principle of millimeter-wave reflected waves into the film winding head. By detecting the film tension, the rotation speed of the film winding head motor is dynamically adjusted to stabilize the film tension within a set range. Simultaneously, based on the film thickness detection results, when areas with insufficient thickness or strength are detected, compensation is achieved by reducing the linear speed and increasing the number of winding layers in the abnormal areas. This closed-loop control strategy avoids both excessive tension leading to film breakage or damage to goods, and insufficient tension leading to loose packaging. Furthermore, by rationally controlling the overlap rate, unnecessary multi-layer winding is reduced, thereby effectively reducing film material consumption while ensuring packaging quality.

[0021] This invention utilizes millimeter-wave radar to detect the real-time offset of goods in the x and y directions. When the goods' position offset exceeds the positioning accuracy threshold of the robotic arm, the system automatically corrects the center coordinates of the wrapping trajectory and issues attitude adjustment and speed adjustment commands to the robotic arm and rotating pallet, respectively, reducing the pallet's rotation speed and minimizing goods swaying. Simultaneously, it corrects the wrapping head's attitude to ensure the wrapping head remains aligned with the goods' outline. When the lateral offset of the wrapping head exceeds a predetermined film width, the system adjusts the robotic arm's wrist center axis and movement trajectory to bring the film coverage back to the planned path. If the offset exceeds a safety threshold, the system automatically stops and alarms. This dynamic correction mechanism enables the system to maintain high packaging accuracy and safety even under conditions of unstable goods stacking and significant external interference, reducing the risk of missing or misaligned packages and equipment collisions caused by swaying and offset.

[0022] This invention integrates a multi-antenna array millimeter-wave radar with a six-axis robotic arm, a lifting column, a rotating pallet, and an industrial-grade controller. It uses a high-speed ADC and FPGA to perform real-time digitization and parallel computation on the radar echoes, generating a 3D point cloud which is then transmitted to the control unit via an EtherCAT bus. Compared to traditional wrapping machines, this system offers better structural compatibility with existing pallet rotation and robotic arm platforms, facilitating upgrades to existing production lines. It significantly reduces manual intervention and debugging workload, improving packaging efficiency and equipment intelligence. Attached Figure Description

[0023] To make the content of this invention easier to understand, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings.

[0024] Figure 1 This is a flowchart of the packaging method based on millimeter-wave radar according to the present invention.

[0025] Figure 2 This is a front view of the packaging system based on millimeter-wave radar according to the present invention.

[0026] Figure 3 This is a top view of the packaging system based on millimeter-wave radar according to the present invention.

[0027] Figure 4 This is a schematic diagram of the main structure of the robotic arm of the present invention.

[0028] Figure 5 This is a top view of the robotic arm of the present invention.

[0029] Explanation of reference numerals on the accompanying drawings: 1-1. Lifting column top plate; 1-2. Height limit sensor; 1-3. Lifting mechanism; 1-4. Robotic arm; 1-5, Columns; 1-6. Rotational structure; 1-7. Pallets; 1-8. Rotary pallet side guards; 1-9. Connectors; 1-10. Bottom guard plate of the packaging machine; 1-11. Fixing screws; 1-12. Control box; 1-13. Rotary electric motor; 1-14. Rotary motor signal connector; 1-15. Rotary motor speed change gearbox; 1-16. PLC electrical control quick operation buttons; 1-17. Touchscreen human-computer interaction display; 1-18. Motor protective cover; 1-19. Protective covers for wire harnesses and connectors; 2-1. Stretch film take-up and discharge machine; 2-2, The first millimeter-wave radar; 2-3. Mounting bracket; 2-4. Rotary motor of the central axis of the robotic arm; 2-5. Stretch film placement roller; 2-6. Second millimeter-wave radar; 2-7. The winding head is fixed to the drive roller; 2-8. Winding head central axis; 2-9. Adjustable roller for winding head; 2-10. Drive arm seat fixing base; 2-11. Chassis rotating turbine shaft; 2-12. Chassis rotating turbine box; 2-13. Box base; 2-14. Fixing flange for the base of the enclosure; 2-15. Wrapping film head fixing flange; 2-16. Wrist central axis; 2-17. Gearbox of the robotic arm wrist motor; 2-18. Mechanical arm forearm motor; 2-19. Robotic arm forearm mounting base; 2-20, Connecting rod; 2-21. The large arm of the robotic arm; 2-22. Connecting rod shaft; 2-23. Robotic arm linkage shaft drive shaft; 2-24. Cycloidal reducer for robotic arms; 2-25. Linkage servo motor; 2-26. Drive arm seat connecting plate; 2-27. Chassis fixing flange cover; 2-28. Chassis rotating motor; 2-29. Fixed flange cover under the chassis rotating turbine box; 2-30. Turbine box mechanical inspection hole. Detailed Implementation

[0030] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present invention. However, the embodiments described are not intended to limit the present invention.

[0031] In this invention, when directions (up, down, left, right, front, and back) are described, it is only for the convenience of describing the technical solution of this invention, and does not indicate or imply that the technical features referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0032] In this invention, "several" means one or more, "multiple" means two or more, "greater than," "less than," "exceeding," etc., are understood to exclude the stated number; "above," "below," "within," etc., are understood to include the stated number. In the description of this invention, the terms "first" and "second" are used only to distinguish technical features and should not be construed as indicating or implying relative importance, or implicitly indicating the number of indicated technical features, or implicitly indicating the order of the indicated technical features.

[0033] In this invention, unless otherwise explicitly defined, the terms "setting," "installing," and "connecting" should be interpreted broadly. For example, they can refer to a direct connection or an indirect connection through an intermediate medium; a fixed connection, a detachable connection, or an integrally formed connection; a mechanical connection, an electrical connection, or a connection capable of mutual communication; or the internal connection of two components or the interaction between two components. Those skilled in the art can reasonably determine the specific meaning of the above terms in this invention based on the specific content of the technical solution.

[0034] Example 1 Reference Figure 1 As shown, this invention provides a packaging method based on millimeter-wave radar, applied to a packaging system, with reference to... Figure 2 As shown, the packaging system includes a pallet 1-7 for loading goods, a film-wrapping head controlled by a robotic arm 1-4, a millimeter-wave radar mounted on the film-wrapping head, and a rotating structure 1-6 supporting the rotation of the pallet 1-7. In addition, it includes a column 1-5 with a lifting mechanism 1-3. The robotic arm 1-4 is configured as a six-axis robotic arm, which is connected to the column 1-5 and is lifted and lowered by the lifting mechanism 1-3.

[0035] The packaging method includes: S1. Perform a three-dimensional contour scan of the packaged goods located on pallets 1-7 using millimeter-wave radar to obtain echo signal data including the dimensions of pallets 1-7, stacking height, surface flatness, and corner positions; including: S11. Perform an initial scan of the packaging scene in a targetless state to build an initial background model.

[0036] S12. Based on the initial background model, construct a difference model through inter-frame difference to update the background and reduce static clutter weights.

[0037] It should be noted that the algorithm initially distinguishes between background clutter and the target cargo signal. In packaging scenarios, there are static interferences such as fixed metal supports and ground reflections. These are included in the background model, and the algorithm automatically filters out these fixed clutters during subsequent scanning to prevent them from affecting cargo contour recognition. Real-time background modeling is the basic process of actively constructing a dynamically updated background model and filtering out fixed interferences (such as metal supports and ground).

[0038] S13. Based on the difference model, candidate target regions are obtained; wherein, the newly acquired data of each frame is differentially analyzed with the current background, the background weight is maintained or enhanced for regions with a difference amplitude lower than a preset difference threshold, and regions with a difference amplitude not less than the difference threshold are selected as candidate target regions.

[0039] Specifically, it includes: An initial scan of the packaging scene is performed without a target to construct an initial background model. : ; in, This refers to continuous scan frame data from a millimeter-wave radar. t represents a time frame; The coordinates are spatial coordinates, and the values ​​are signal strength. N is the number of frames; When new data arrives in each frame, the background is updated through inter-frame differencing to reduce static clutter weights. ; in, , is the background retention coefficient; For indicator functions; The preset difference threshold; ; Difference between current frame and background for: ; in, ,when The value is 1 if the condition is met; otherwise, it is 0, meaning only the static region is updated. Candidate target regions are obtained through background subtraction. ; .

[0040] S14. Calculate the mean and standard deviation of the signal strength within the candidate target area.

[0041] The average signal strength within the candidate target region and standard deviation : ; ; in, The number of pixels in the candidate target region; This is the candidate target region.

[0042] S15. Calculate the detection threshold based on the mean and standard deviation, and based on the reflectance coefficient of the cargo material in the candidate target area.

[0043] S16. Dynamically adjust the detection threshold according to the material of the goods in the candidate target area.

[0044] Based on material reflectivity Dynamic adjustment detection For example, the reflectivity of metal materials. The reflectivity of the cardboard box material ).

[0045] ; in, It dynamically adjusts based on the signal-to-noise ratio, especially at low signal-to-noise ratios. ; S17. Retain the region within the candidate target area where the signal intensity is higher than the detection threshold as the purified target signal.

[0046] Regions within the candidate target area with signal intensity higher than the detection threshold are retained as purified target signals. : .

[0047] It should be noted that, based on the differences in the reflective characteristics of goods and packaging materials, the algorithm dynamically adjusts the signal detection threshold through statistical learning. For example, for metal goods, which have strong radar reflection signals, the algorithm appropriately increases the threshold to prevent contour distortion caused by the overflow of strong reflection signals; for goods made of weakly reflective materials such as cardboard boxes, the threshold is lowered to ensure that weak signals at the edges of the goods are captured, avoiding contour loss. Simultaneously, the threshold adjustment also incorporates factors such as ambient light and humidity to further enhance anti-interference capabilities. Adaptive threshold adjustment dynamically adjusts the threshold according to the goods' material, filtering out dynamic interference (such as dust and packaging film fragments).

[0048] S18. After performing signal smoothing and denoising and phase adjustment on the target signal, the cargo contour phase signal used for subsequent 3D point cloud modeling is obtained.

[0049] It should be noted that the radar signal is filtered in both time and space dimensions. In the time dimension, a sliding window filter is used to smooth multiple consecutive frames of radar scan data, eliminating instantaneous data fluctuations caused by slight shaking of the cargo or random noise in the radar signal. In the spatial dimension, median filtering is used to remove isolated clutter points, preserve the edge details of the cargo outline, and ensure the integrity and accuracy of the three-dimensional contour.

[0050] The purpose of spatiotemporal filtering is to smooth signals in both time and space dimensions and remove isolated noise.

[0051] Phase modulation uses phase information to distinguish goods from interference (such as metal pallets 1-7 and cardboard boxes). Phase demodulation uses the phase information of millimeter-wave radar signals to further distinguish target signals from clutter signals. Objects of different materials exhibit different phase reflection characteristics of millimeter waves. For example, the phase shift of metal pallets 1-7 is significantly different from that of cardboard boxes. The algorithm demodulates the phase data and combines it with a pre-set material phase database to accurately remove phase components of interference signals such as the metal frame of pallets 1-7 and reflections from packaging film, ultimately outputting a clean cargo contour phase signal, providing high-quality data for subsequent 3D point cloud modeling.

[0052] Specifically, smoothing is performed on multiple consecutive frames of radar data in the target signal over time to suppress instantaneous fluctuations, including: For continuous The frame data is weighted and averaged to suppress instantaneous fluctuations, resulting in the weighted average target signal. : ; in , where is the weight, which decays over time; To obtain the frame count, it is generally taken as... , If the value is too small, the purification and smoothing effect will be insufficient. An excessively large value will increase data processing latency and affect the real-time performance of path correction.

[0053] This corresponds to taking the frame from the current frame to the earliest frame within the same frame.

[0054] The purified target signal; In the spatial dimension, median filtering is applied to the weighted averaged target signal to remove isolated clutter points and preserve contour edge details, including: right The signal strength is obtained by taking the median of the neighborhood and removing isolated clutter points. : ; in, for of Neighborhood; Represent the median function; The signal is filtered in the spatiotemporal domain, and t is a time frame; These are the spatial coordinates of the radar scanning plane; The phase information of millimeter-wave signals is used in conjunction with a pre-defined material phase feature library to demodulate the target echo, eliminating interference signals whose phase characteristics do not match the target, including: Calculate the difference between the phase and the background phase at the same location. : ; in, The background phase model is obtained through the initial scan; The phase signal of the radar echo, in radians; A preset material phase feature library includes metal phase offset and cardboard box phase offset; among which, the metal phase offset is: ; The phase offset of the cardboard box is: ; Preserve the signal that matches the target phase : ; in, The phase range of the target cargo.

[0055] S2. Based on the echo signal, generate a three-dimensional point cloud model of the cargo.

[0056] Input radar simulated echo signal ; Among them, carrier frequency ; Signal strength; For carrier frequency; It is a time variable; The instantaneous phase is the angle that changes over time; By sampling rate Discretization yields: ; in, For different sampling points; 3D point clouds can be generated in real time through FPGA programming. The distance calculation for the 3D point cloud is as follows: Based on signal flight time ,distance Frequency shift extracted by Fourier transform ,calculate And thus obtain ;in: ; ; ; in, For relative velocity, For the speed of light, when dealing with a static target, it simplifies to a direct mapping between distance and frequency; Angle calculation of 3D point cloud: Assuming a 4×4 antenna array with horizontal and vertical phase difference and Horizontal angle and vertical angle Calculation yields: ; ; in, For wavelength, Antenna spacing; 3D point cloud data output: Based on the above calculations, the three-dimensional coordinates of each point can be obtained as follows: The data is transmitted to the control unit via the EtherCAT bus.

[0057] S3. Analyze the three-dimensional point cloud model and extract cargo parameters; the cargo parameters include overall length, width, height, corner positions and surface flatness, and identify points with curvature greater than a preset curvature threshold as corner points.

[0058] The 3D point cloud model is analyzed to obtain the 3D point cloud. , This represents the number of point clouds; Size calculation: length ; width ; high ; Corner points are identified by point cloud curvature calculation; points with curvature exceeding a threshold are also identified. ; in, For points in a point cloud, For the number of neighboring points, for normal vector, curvature It is determined to be a corner point; For neighboring points; For metal box goods with sharp corners, mark the corner coordinates to provide a positioning basis for subsequent lateral reinforcement coverage; for goods with uneven surfaces, identify raised or recessed areas, and adjust the height and tension of robotic arms 1-4 and the wrapping head to avoid missing packages or overly tight packaging that could cause the goods to deform.

[0059] S4. Based on the cargo parameters, a path planning strategy combining longitudinal layering and lateral trajectory is adopted to generate a wrapping packaging trajectory. According to the cargo stacking height, the packaging path is divided into multiple longitudinal layers. The robotic arms 1-4 spirally rise along the lifting columns 1-5 according to the layers, and the height of the wrapping head is dynamically adjusted with the layers to ensure that the coverage height of each layer of packaging film is uniform. In the lateral dimension, for the corner areas of the cargo, the lateral coverage frequency of the wrapping head is increased, and the angle of the wrapping head is adjusted to ensure that the packaging film at the corners is tightly attached to prevent damage to the corners during transportation.

[0060] Step S4 includes: S41. Divide the packaging path into multiple height levels according to the height of the goods, and raise the film wrapping head layer by layer along the lifting direction so that the height of the film wrapping head is adjusted with the change of the level. S42. In each height level, a spiral trajectory of the wrapping head relative to the goods is generated based on the center coordinates of pallet 1-7, so that the wrapping head forms a transverse spiral winding trajectory relative to the goods in the transverse direction. S43. Increase the number of transverse windings of the film head in the height level that includes the corner points.

[0061] Specifically, the packaging path is divided into multiple height levels based on the height of the goods, including: height Divided into Layers, height of each layer That is, each layer ≤10cm, the first Floor height ; Based on the center coordinates of pallet 1-7, a spiral trajectory of the wrapping head relative to the goods is generated, including: The film wrapping head centers on trays 1-7 in each layer. The polar equation for a spiral motion centered on a circle is: ; in, , where is the initial radius, covering the maximum lateral dimension of the cargo; , where is the spiral step size, controlling the film overlap rate; The number of wraps. .

[0062] Increasing the number of lateral windings of the film head in the height level including the corner points includes: opposite corner points By increasing the number of wrapping cycles at the corresponding height level, the trajectory is corrected as follows: ,exist Add one more wrap within the range.

[0063] Specifically, it also includes: The lateral step distance between adjacent winding tracks is calculated based on the width of the film used by the winding head, so that the overlap rate between adjacent film strips is within a preset range, so as to avoid excessive overlap causing material waste or insufficient overlap causing missing packages. The tension of the winding head is controlled within the range of 0.3N-5N according to the preset tension parameters, and correction is made when the tension is detected to exceed the range.

[0064] S5. When the shaking of goods or the offset of the wrapping head is detected, the wrapping trajectory is dynamically corrected to achieve adaptive adjustment of the packaging path.

[0065] Specifically, it includes: When the millimeter-wave radar detects in real time that the positional offset of the cargo exceeds the positioning accuracy threshold of the robotic arms 1-4 that drive the film-wrapping head, the offset of the cargo in the x and y directions is calculated. , and the trajectory center was changed from Revised to ; Based on the corrected trajectory center, a position adjustment command is sent to the robotic arm 1-4 to correct the posture of the wrapping head and ensure that the wrapping head is continuously aligned with the outline of the goods; at the same time, a speed adjustment command is sent to the rotating structure 1-6 that supports the rotation of the pallet 1-7 to reduce its rotation speed, reduce the swaying amplitude of the goods, and avoid packaging deviation caused by swaying. The film thickness detector (±3μm accuracy, based on the phase difference principle of millimeter wave reflected waves) mounted on the film winding head detects the film coverage position in real time. In response to the lateral offset of the film winding head exceeding the predetermined value of the film width (such as the lateral offset exceeding 10% of the film width), the central axis 2-16 of the robotic arm wrist is controlled to adjust the angle of the film winding head, and the movement trajectory of the robotic arm 1-4 is corrected simultaneously, so that the film coverage position conforms to the planned path again; If the lateral offset of the wrapping head exceeds the set safety threshold, the packaging operation will be suspended and a warning will be triggered (a fault prompt will be sent to the human-machine interface). The film winding head is equipped with a tension sensor to detect the film tension F in real time. When the film tension F deviates from the preset tension F0 and exceeds the range of 0.3N-5N (such as due to increased tension caused by a decrease in film roll diameter), the speed of the film winding head motor is adjusted. , Set the initial rotation speed to ensure the tension remains stable between 0.3N and 5N; When the thickness sensor (set on the film winding head) detects a local thickness abnormality in the film that results in insufficient strength, the linear speed of the film winding head is reduced, and the number of winding layers is increased in the abnormal area to compensate for material defects and improve packaging quality.

[0066] Through multi-layered processing strategies including real-time background modeling, adaptive threshold adjustment, spatiotemporal filtering optimization, and phase demodulation, clutter signals caused by environmental reflections, pallet 1-7 structural reflections, and the characteristics of the goods themselves are suppressed, outputting clean target contour data. This avoids path deviations caused by clutter interference and ensures the accuracy of dynamic path planning. Based on the 3D model of the goods acquired by millimeter-wave acquisition, the packaging trajectory is automatically generated. The height of the robotic arms 1-4 is adjusted longitudinally according to the height of the goods, and the coverage of the wrapping head is strengthened for the corners laterally. It also has a dynamic correction function. When the shaking of the goods or the deviation of the wrapping head is detected, adjustment commands can be sent to the actuator in real time to ensure packaging quality.

[0067] Example 2 Based on the same inventive concept, this embodiment provides a packaging system based on millimeter-wave radar. The principle of solving the problem is similar to that of the packaging method based on millimeter-wave radar, and the repetitions will not be repeated.

[0068] This embodiment provides a packaging system based on millimeter-wave radar, referring to... Figures 2 to 5 As shown, it includes: Columns 1-5 are equipped with lifting mechanisms 1-3. The robotic arm 1-4 is configured as a six-axis robotic hand, which is connected to the column 1-5 and is lifted and lowered by the lifting mechanism 1-3; A film winding head is located at the end of the robotic arm 1-4. The film winding head includes a mounting frame 2-3 and a film placement roller 2-5, a winding head fixed transmission roller 2-7, a winding head central shaft 2-8, and a winding head movable adjustment roller 2-9 arranged in parallel on the mounting frame 2-3. The film placement roller 2-5 is driven by a film winding motor 2-1 located on the mounting frame 2-3 to realize the winding and unwinding of the film. Millimeter-wave radar is mounted on the mounting bracket 2-3; Trays 1-7 are positioned beside the robotic arms 1-4; Rotating structures 1-6 drive the trays 1-7 to rotate; A control system for implementing the millimeter-wave radar-based packaging method includes: The echo signal data acquisition module is used to perform a three-dimensional contour scan of the packaged goods located on pallets 1-7 using millimeter-wave radar, obtaining echo signal data including the dimensions of pallets 1-7, stacking height, surface flatness, and corner positions; including: The initial background model building unit is used to perform an initial scan of the packaging scene in a targetless state and build an initial background model; The background update unit is used to construct a difference model based on the initial background model through inter-frame difference, so as to update the background and reduce the static clutter weights. The candidate target region acquisition unit is used to obtain candidate target regions based on the difference model; wherein, the newly acquired data of each frame is differentially analyzed with the current background, the background weight is maintained or enhanced for regions with a difference amplitude lower than a preset difference threshold, and regions with a difference amplitude not less than the difference threshold are used as candidate target regions. The mean and standard deviation calculation unit is used to calculate the mean and standard deviation of the signal intensity within the candidate target area; The detection threshold calculation unit is used to calculate the detection threshold based on the mean and standard deviation, and based on the reflectance coefficient of the cargo material in the candidate target area; A dynamic adjustment unit is used to dynamically adjust the detection threshold according to the material of the goods in the candidate target area; A purification unit is used to retain regions within the candidate target area where the signal intensity is higher than the detection threshold, as purified target signals. The phase signal acquisition unit is used to perform signal smoothing and denoising and phase adjustment on the target signal to obtain the cargo contour phase signal for subsequent 3D point cloud modeling. A three-dimensional point cloud model acquisition unit is used to generate a three-dimensional point cloud model of the cargo based on the echo signal. The cargo parameter acquisition module is used to parse the three-dimensional point cloud model and extract cargo parameters. The cargo parameters include overall length, width, height, corner positions and surface flatness, and points with curvature greater than a preset curvature threshold are identified as corner points. A wrap-around packaging trajectory generation module is used to generate a wrap-around packaging trajectory based on the cargo parameters, employing a path planning strategy that combines longitudinal layering and lateral trajectory; including: The height adjustment unit is used to divide the packaging path into multiple height levels according to the height of the goods. The wrapping head rises layer by layer along the lifting direction, so that the height of the wrapping head is adjusted according to the level. The transverse spiral winding trajectory generation unit is used to generate a spiral trajectory of the wrapping head relative to the goods based on the center coordinates of pallet 1-7 in each height level, so that the wrapping head forms a transverse spiral winding trajectory relative to the goods in the transverse direction. A transverse winding number increasing unit is used to increase the transverse winding number of the film head in the height level including the corner points; An adaptive adjustment module is used to dynamically correct the wrapping trajectory when cargo shaking or wrapping head deviation is detected, so as to achieve adaptive adjustment of the packaging path.

[0069] Specifically, a millimeter-wave radar operating at 77GHz is employed, configured with a 4×4 multi-antenna array structure. Two millimeter-wave radars (first millimeter-wave radar 2-2 and second millimeter-wave radar 2-6) are installed on both sides of the wrapping head, maintaining a horizontal distance of approximately 30cm, forming a dual-view cross-positioning arrangement. The millimeter-wave radar is used to perform real-time three-dimensional contour scanning of the packaged goods, acquiring spatial information such as the dimensions of pallets 1-7, the stacking height of the goods, surface flatness, and corner positions. Based on the phase difference of the multi-antenna array, interference from the metal frame of pallets 1-7 and film reflections is suppressed, achieving an overall positioning accuracy of ±0.05mm, thus providing a high-precision data foundation for subsequent packaging path planning.

[0070] For example, the lifting mechanism 1-3 includes a lifting motor and a lifting screw connected to its drive end. The lifting screw is movably connected to a connector 1-9 connected to the robotic arm 1-4. A height limit sensor 1-2 is provided on the column 1-5, and a lifting column 1-5 top plate is installed on the top of the column 1-5.

[0071] For example, the rotating structure 1-6 includes a rotating motor 1-13, and a pallet 1-7 is installed at the center of the bottom of the equipment. The pallet 1-7 is connected to the sprocket connecting shaft of the rotating motor 1-13 through a sprocket transmission mechanism and is driven by the rotating motor 1-13, serving as a working platform for carrying and rotating goods.

[0072] The side guards of the rotating pallet 1-7 are located on the outside of the pallet 1-7 and the sprocket drive mechanism to protect the transmission components from foreign objects entering, and to ensure that the stretch film is not rolled up during high-speed operation.

[0073] Fixing screws 1-11 are used to rigidly fix the tray 1-7 to the sprocket connecting shaft to ensure stable power transmission.

[0074] The rotary motor 1-13 is fixed inside the bottom guard plate 1-10 of the packaging machine. The rotary motor speed change gearbox 1-15 is installed at the output end of the rotary motor 1-13. Power is output through the rotary motor speed change gearbox 1-15 and drives the tray 1-7 to rotate via the sprocket connecting shaft.

[0075] The rotating motor signal connector 1-14 is connected to the control box 1-12 via a wire harness and is used to transmit the working commands and feedback signals of the rotating motor 1-13.

[0076] The bottom tray 1-7 rotates the motor 1-13 with a protective cover to protect the motor and wiring harness from external impact, entanglement or liquid intrusion.

[0077] The control box 1-12 is installed on the side of the chassis and is electrically connected to the height limit sensor 1-2, the rotary motor 1-13, the six-axis robotic arm 1-4, and the lifting mechanism 1-3.

[0078] PLC electrical control quick operation buttons 1-16 are located on the outside of control box 1-12 and are electrically connected to the PLC controller. They are used for on-site operations such as emergency stop, manual jogging, reset, start and stop.

[0079] The touch-screen human-machine interface display 1-17 is installed outside the control box 1-12 and is connected to the signal processing center via a communication line. It is used to display system status, set parameters, read alarm information, and perform manual debugging.

[0080] The protective cover plate for the wire harness and connectors 1-9 is used to cover the wire harness channel between the robotic arm 1-4, the column 1-5 and the electrical control box to prevent the wire harness from being worn, falling off or being interfered with by external forces.

[0081] The bottom guard plates 1-10 of the packaging machine are located on the outside of the chassis. They are used to enclose the bottom structure and protect the internal transmission, motor and wiring harness, thereby improving the stability and protection level of the whole machine.

[0082] Motor protective covers 1-18 cover the rotary motor 1-13 to prevent mechanical impact, film entanglement and dust intrusion.

[0083] For example, the robotic arms 1-4 have a load capacity of 50kg, a positioning accuracy of ±0.1mm, and a maximum working radius of 2.5m. The end effector robotic arms 1-4 are equipped with a film-wrapping head at the hand position. It can adapt to various specifications of items with lengths of 1m-2.5m and widths of 0.8-1.5m. Through the coordinated movement of multiple motors, the film-wrapping head can achieve multi-angle, high-precision displacement to complete the packaging action.

[0084] For example, the film winding head is equipped with a 0.3-5N adjustable tension sensor and a film thickness detector with an accuracy of ±3μm, supporting various materials such as PE film and stretch film; it can monitor film tension and thickness in real time to avoid missing packages or material waste, and at the same time work with robotic arms 1-4 to complete film winding and cutting actions.

[0085] For example, refer to Figure 4As shown, the robotic arm 1-4 includes a drive arm base fixing seat 2-10, a chassis rotary turbine shaft 2-11, a chassis rotary turbine box 2-12, a box base 2-13, a box base fixing flange 2-14, a wrapping film head fixing flange 2-15, a robotic arm wrist central shaft 2-16, a mechanical wrist motor gearbox 2-17, a robotic arm forearm motor 2-18, a robotic arm forearm fixing seat 2-19, a connecting rod 2-20, a robotic arm upper arm 2-21, a connecting rod connecting shaft 2-22, a robotic arm 1-4 connecting rod 2-20 shaft drive shaft, a robotic arm cycloidal reducer 2-24, a connecting rod servo motor 2-25, a drive arm base connecting plate 2-26, a chassis fixing flange cover 2-27, a robotic arm 1-4 chassis rotary motor 1-13, a chassis rotary turbine box lower fixing flange cover 2-29, and a turbine box mechanical inspection hole 2-30, etc.

[0086] Among them, the wrapping head fixing flange 2-15 is used to fix the wrapping head to the wrist of the robotic arm 1-4 as a whole, so as to realize the rigid connection between the wrapping head and the robotic arm 1-4; The central shaft 2-16 of the robotic arm wrist is connected to the wrapping head fixing flange 2-15, and is driven by the central shaft rotary motor 2-4 of the robotic arm to adjust the angle of the wrapping head; The gearbox 2-17 of the robotic arm wrist motor is connected to the wrist central shaft 2-16. The robotic arm forearm motor 2-18 is installed in the robotic arm forearm fixing seat 2-19 and drives the robotic arm forearm to lift and pitch through the internal reduction mechanism. The robotic arm forearm mounting base 2-19 is connected to the link 2-20 and the robotic arm big arm 2-21. The link 2-20 is connected to the robotic arm forearm mounting base 2-19 through the link connecting shaft 2-22, which is used to transmit the movement between the forearm and the big arm. The robotic arm's large arm 2-21 is connected to the connecting rod 2-20 via the robotic arm's connecting rod shaft transmission shaft 2-23, and is driven by the robotic arm's cycloidal reducer 2-24; The robotic arm linkage shaft drive shaft 2-23 connects linkage 2-20 to the robotic arm's main arm 2-21. The cycloidal reducer 2-24 of the robotic arm is installed on the drive arm base fixed seat 2-10 and connected to the connecting rod servo motor 2-25; The linkage servo motor 2-25 drives the upper arm 2-21 of the robotic arm through the cycloidal reducer 2-24 of the robotic arm, which is used to realize the lifting and lowering and angle adjustment of the upper arm; The drive arm base connecting plate 2-26 is used to fix the execution components of the robotic arm 1-4 onto the chassis rotating platform, thereby achieving a rigid connection between the upper part of the robotic arm 1-4 and the chassis. The drive arm mounting base 2-10 serves as the bottom support for the robotic arm 1-4 and is fixedly connected to the chassis rotating structure 1-6. The chassis rotary turbine shaft 2-11 is connected to the rotary turbine box, providing rotational freedom for the robotic arm 1-4; The chassis rotating turbine box 2-12 is equipped with a worm gear mechanism, which is connected to the chassis rotating motor 2-28 to drive the base of the robotic arm 1-4 to rotate. The base 2-13 supports the entire robotic arm 1-4 and the winding platform, providing an installation platform for the chassis rotary motor 2-28, the turbine box, and the control system. The mounting flange 2-14 of the housing base is located at the lower end of the base and is used for fixed connection with the ground or mounting platform.

[0087] The chassis mounting flange cover 2-27 is used to cover the mounting flange area, providing safety protection and structural sealing. The rotary motor 1-13 of the robotic arm 1 chassis is connected to the turbine box and drives the robotic arm chassis to rotate through the turbine shaft; The chassis rotating turbine housing lower fixed flange cover 2-29 is installed under the turbine housing to protect the bottom components of the turbine housing; Turbine box mechanical inspection holes 2-30 are used for repairing the internal mechanisms of the turbine box, facilitating lubrication, maintenance, and component replacement.

[0088] Specifically, the control system includes touch-screen human-machine interface displays 1-17 and a controller. The controller uses an industrial-grade PLC, supports EtherCAT bus, and can link millimeter-wave data processing modules with robotic arms 1-4, with a response cycle of ≤20ms. It can quickly analyze 3D point cloud data, generate packaging strategies, and send action commands to the actuators, while simultaneously receiving sensor feedback signals for real-time adjustments.

[0089] The touch-screen human-machine interaction display 1-17 can preset parameters such as packaging type (wooden pallet, plastic pallet) and goods material (cardboard box, metal box), and supports one-click recall of historical packaging schemes; at the same time, it displays the equipment operating status, packaging progress and fault information, which facilitates manual operation and packaging process monitoring.

[0090] The controller integrates a high-speed ADC with a sampling rate of no less than 1 GHz and an FPGA chip. The high-speed ADC is used to convert the analog echo of the millimeter-wave radar into digital data. The FPGA uses its parallel processing capabilities to quickly calculate the digital signal, generate a real-time 3D point cloud model, and send the results to the integrated control unit, thereby significantly reducing the overall data processing latency.

[0091] The controller has an ADC analog-to-digital converter module, which is used to discretize and sample the input analog signal, so that the radar echo signal enters the FPGA processing flow in digital form, ensuring the data accuracy and timeliness of subsequent path planning and attitude control.

[0092] In addition to millimeter-wave radar, it is also equipped with auxiliary sensors such as height limit sensors 1-2 and infrared sensors, which can help monitor information such as the lifting height of robotic arms 1-4 and the position of goods to prevent overtravel. At the same time, it also integrates environmental sensors such as temperature sensors to help monitor environmental changes (such as high temperature and high humidity) and provide data support for multimodal fusion positioning.

[0093] The packaging system first uses millimeter-wave radar sensors to collect the outline signal of the goods, which is then converted into three-dimensional point cloud data by a signal processing unit. After the integrated control unit analyzes the data, it generates a packaging strategy by combining preset parameters or adaptive algorithms, and sends action commands to the six-axis robotic arm 1-4, the film wrapping head, the lifting mechanism 1-3, and the rotating tray 1-7. While the actuators complete the packaging process, they provide real-time status feedback through auxiliary sensors such as tension sensors and thickness detectors. The control unit adjusts the commands according to the feedback parameters, ultimately achieving a high-precision, automated packaging process without human intervention. The packaging process can be monitored in real time or traced historically through a human-machine interface.

[0094] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0095] 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.

[0096] 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 1The function specified in one or more boxes.

[0097] 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.

[0098] Finally, it should be noted that the above specific embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to examples, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A packaging method based on millimeter-wave radar, characterized in that, The packaging system, comprising a pallet for loading goods, a film-wrapping head controlled by a robotic arm, a millimeter-wave radar mounted on the film-wrapping head, and a rotating structure supporting the rotation of the pallet, includes the following packaging method: S1. Perform a three-dimensional contour scan of the packaged goods on the pallet using millimeter-wave radar to obtain echo signal data including pallet dimensions, stacking height, surface flatness, and corner positions; including: S11. Perform an initial scan of the packaging scene in a targetless state to build an initial background model; S12. Based on the initial background model, a difference model is constructed through inter-frame difference to update the background and reduce static clutter weights; S13. Based on the difference model, candidate target regions are obtained; wherein, the newly acquired data of each frame is differentially analyzed with the current background, the background weight is maintained or enhanced for regions with a difference amplitude lower than a preset difference threshold, and regions with a difference amplitude not less than the difference threshold are selected as candidate target regions. S14. Calculate the mean and standard deviation of the signal strength within the candidate target area; S15. Calculate the detection threshold based on the mean and standard deviation, and based on the reflectance coefficient of the cargo material in the candidate target area; S16. Dynamically adjust the detection threshold according to the material of the goods in the candidate target area; S17. Retain the region within the candidate target area whose signal intensity is higher than the detection threshold as the purified target signal; S18. After performing signal smoothing and denoising and phase adjustment on the target signal, the cargo contour phase signal used for subsequent three-dimensional point cloud modeling is obtained. S2. Based on the echo signal, generate a three-dimensional point cloud model of the cargo; S3. Analyze the three-dimensional point cloud model and extract cargo parameters; the cargo parameters include overall length, width, height, corner positions and surface flatness, and identify points with curvature greater than a preset curvature threshold as corner points; S4. Based on the cargo parameters, a path planning strategy combining longitudinal layering and lateral trajectory is adopted to generate a wrap-around packaging trajectory; including: S41. Divide the packaging path into multiple height levels according to the height of the goods, and raise the film wrapping head layer by layer along the lifting direction so that the height of the film wrapping head is adjusted with the change of the level. S42. In each height level, a spiral trajectory of the wrapping head relative to the goods is generated based on the center coordinates of the pallet, so that the wrapping head forms a transverse spiral winding trajectory relative to the goods in the transverse direction. S43. Increase the number of transverse windings of the film-wrapping head in the height level that includes the corner points; S5. When the shaking of goods or the offset of the wrapping head is detected, the wrapping trajectory is dynamically corrected to achieve adaptive adjustment of the packaging path.

2. The packaging method based on millimeter-wave radar according to claim 1, characterized in that, Based on the initial background model, a difference model is constructed through inter-frame differencing to update the background and reduce static clutter weights. Based on the difference model, candidate target regions are obtained, including: An initial scan of the packaging scene is performed without a target to construct an initial background model. : ； in, This refers to continuous scan frame data from a millimeter-wave radar. t represents a time frame; The coordinates are spatial coordinates, and the value is the signal strength. N is the number of frames; When new data arrives in each frame, the background is updated through inter-frame differencing to reduce static clutter weights. ； in, , is the background retention coefficient; For indicator functions; The preset difference threshold; ； Difference between current frame and background for: ； in, ,when The value is 1 if the condition is met; otherwise, it is 0, meaning only the static region is updated. Candidate target regions are obtained through background subtraction. ; 。 3. The packaging method based on millimeter-wave radar according to claim 1, characterized in that, Based on the mean and standard deviation, and based on the reflectance coefficient of the cargo material in the candidate target region, a detection threshold is calculated. The detection threshold is then dynamically adjusted according to the cargo material in the candidate target region, including: The average signal strength within the candidate target region and standard deviation : ； ； in, The number of pixels in the candidate target region; Candidate target regions; Based on material reflectivity Dynamic adjustment detection : ； in, It dynamically adjusts based on the signal-to-noise ratio, especially at low signal-to-noise ratios. ; Regions within the candidate target area with signal intensity higher than the detection threshold are retained as purified target signals. : 。 4. The packaging method based on millimeter-wave radar according to claim 1, characterized in that, Performing signal smoothing and denoising, and phase adjustment on the target signal, includes: Smoothing of multiple consecutive frames of radar data in the target signal over time to suppress transient fluctuations includes: For continuous The frame data is weighted and averaged to suppress instantaneous fluctuations, resulting in the weighted average target signal. : ； in , where is the weight, which decays over time; To retrieve the frame number; This corresponds to taking the frame from the current frame to the earliest frame within the same frame; The purified target signal; In the spatial dimension, median filtering is applied to the weighted averaged target signal to remove isolated clutter points and preserve contour edge details, including: right The signal strength is obtained by taking the median of the neighborhood and removing isolated clutter points. : ； in, for of Neighborhood; Represent the median function; The signal is filtered in the spatiotemporal domain, and t is a time frame; These are the spatial coordinates of the radar scanning plane; The phase information of millimeter-wave signals is used in conjunction with a pre-defined material phase feature library to demodulate the target echo, eliminating interference signals whose phase characteristics do not match the target, including: Calculate the difference between the phase and the background phase at the same location. : ； in, The background phase model is obtained through the initial scan; The phase signal of the radar echo, in radians; A preset material phase feature library includes metal phase offset and cardboard box phase offset; among which, the metal phase offset is: ； The phase offset of the cardboard box is: ； Preserve the signal that matches the target phase : ； in, The phase range of the target cargo.

5. The packaging method based on millimeter-wave radar according to claim 1, characterized in that, Based on the echo signal, a three-dimensional point cloud model of the cargo is generated, including: Input radar simulated echo signal ; Among them, carrier frequency ; Signal strength; For carrier frequency; It is a time variable; The instantaneous phase is the angle that changes over time; By sampling rate Discretization yields: ； in, For different sampling points; The distance calculation for a 3D point cloud is as follows: Based on signal flight time ,distance Frequency shift extracted by Fourier transform ,calculate And thus obtain ;in: ； ； ； in, For relative velocity, For the speed of light, when dealing with a static target, it simplifies to a direct mapping between distance and frequency; Angle calculation of 3D point cloud: Assuming a 4×4 antenna array with horizontal and vertical phase difference and Horizontal angle and vertical angle Calculation yields: ； ； in, For wavelength, Antenna spacing; 3D point cloud data output: Based on the above calculations, the three-dimensional coordinates of each point can be obtained as follows: .

6. The packaging method based on millimeter-wave radar according to claim 1, characterized in that, The 3D point cloud model is analyzed to extract cargo parameters. These parameters include overall length, width, height, corner positions, and surface flatness. Points with curvature greater than a preset curvature threshold are identified as corner points. The 3D point cloud model is analyzed to obtain the 3D point cloud. , This represents the number of point clouds; Size calculation: length ; width ; high ; Corner points are identified by point cloud curvature calculation; points with curvature exceeding a threshold are also identified. ； in, For points in a point cloud, For the number of neighboring points, for normal vector, curvature , used to determine if it is a corner point; For neighboring points; For metal box goods with sharp corners, mark the corner coordinates to provide a positioning basis for subsequent lateral reinforcement coverage; for goods with uneven surfaces, identify raised or recessed areas, and adjust the height and tension of the robotic arm and wrapping head to avoid missing packages or overly tight packaging that could cause the goods to deform.

7. A packaging method based on millimeter-wave radar according to claim 1, characterized in that, The packaging path is divided into multiple height levels based on the height of the goods, including: height Divided into Layers, height of each layer That is, each layer ≤10cm, the first Floor height ; 3D point cloud The point in the middle; Generate a spiral trajectory of the wrapping head relative to the goods based on the center coordinates of the pallet, including: The film wrapping head is centered on the tray in each layer. The polar equation for a spiral motion centered on a circle is: ； in, , where is the initial radius, covering the maximum lateral dimension of the cargo; , where is the spiral step size, controlling the film overlap rate; The number of wraps. , Increasing the number of lateral windings of the film head in the height level including the corner points includes: opposite corner points By increasing the number of wrapping cycles at the corresponding height level, the trajectory is corrected as follows: ,exist Add one more wrap within the range.

8. The packaging method based on millimeter-wave radar according to claim 1, characterized in that, Also includes: The lateral step distance between adjacent winding tracks is calculated based on the width of the film used by the winding head, so that the overlap rate between adjacent film strips is within a preset range, so as to avoid excessive overlap causing material waste or insufficient overlap causing missing packages. The tension of the winding head is controlled within the range of 0.3N-5N according to the preset tension parameters, and correction is made when the tension is detected to exceed the range.

9. A packaging method based on millimeter-wave radar according to claim 1, characterized in that, Dynamically correcting the wrapping trajectory when cargo shaking or wrapping head deviation is detected, including: When the millimeter-wave radar detects in real time that the positional offset of the cargo exceeds the positioning accuracy threshold of the robotic arm driving the film-wrapping head, the offset of the cargo in the x and y directions is calculated. , and the trajectory center was changed from Revised to ; Based on the corrected trajectory center, a position adjustment command is sent to the robotic arm to correct the posture of the wrapping head and ensure that the wrapping head is continuously aligned with the outline of the goods; at the same time, a speed adjustment command is sent to the rotating structure supporting the pallet rotation to reduce its rotation speed, reduce the amplitude of goods swaying, and avoid packaging deviation caused by swaying. The film covering position is detected in real time. When the lateral offset of the film winding head exceeds the predetermined value of the film width, the central axis of the robotic arm wrist is controlled to adjust the angle of the film winding head, and the movement trajectory of the robotic arm is corrected simultaneously so that the film covering position conforms to the planned path again. If the lateral offset of the wrapping head exceeds the set safety threshold, the packaging operation will be suspended and a warning will be triggered. The film tension F is detected in real time. When the film tension F deviates from the preset tension F0 and exceeds the range of 0.3N-5N, the rotation speed of the film winding head motor is adjusted. , Set the initial rotation speed to ensure the tension remains stable between 0.3N and 5N; In response to the detection of localized film thickness anomalies leading to insufficient strength, the winding head linear speed is reduced, and the number of winding layers is increased in the abnormal area to compensate for material defects and improve packaging quality.

10. A packaging system based on millimeter-wave radar, characterized in that, include: The column is equipped with a lifting mechanism. The robotic arm, configured as a six-axis robotic hand, is connected to the column and achieves lifting and lowering via the lifting mechanism; A film-winding head is located at the end of the robotic arm. The film-winding head includes a mounting frame and a film-winding placement roller, a film-winding head fixed transmission roller, a film-winding head central shaft, and a film-winding head movable adjustment roller arranged in parallel with each other on the mounting frame. The film-winding placement roller is driven by a film-winding motor located on the mounting frame. Millimeter-wave radar is mounted on the mounting bracket; A tray is positioned beside the robotic arm; A rotating structure drives the tray to rotate; A control system for implementing the millimeter-wave radar-based packaging method according to any one of claims 1-9, comprising: The echo signal data acquisition module is used to perform three-dimensional contour scanning of packaged goods located on a pallet using millimeter-wave radar, obtaining echo signal data including pallet size, stacking height, surface flatness, and corner positions; including: The initial background model building unit is used to perform an initial scan of the packaging scene in a targetless state and build an initial background model; The background update unit is used to construct a difference model based on the initial background model through inter-frame difference, so as to update the background and reduce the static clutter weights. The candidate target region acquisition unit is used to obtain candidate target regions based on the difference model; wherein, the newly acquired data of each frame is differentially analyzed with the current background, the background weight is maintained or enhanced for regions with a difference amplitude lower than a preset difference threshold, and regions with a difference amplitude not less than the difference threshold are used as candidate target regions. The mean and standard deviation calculation unit is used to calculate the mean and standard deviation of the signal intensity within the candidate target area; The detection threshold calculation unit is used to calculate the detection threshold based on the mean and standard deviation, and based on the reflectance coefficient of the cargo material in the candidate target area; A dynamic adjustment unit is used to dynamically adjust the detection threshold according to the material of the goods in the candidate target area; A purification unit is used to retain regions within the candidate target area where the signal intensity is higher than the detection threshold, as purified target signals. The phase signal acquisition unit is used to perform signal smoothing and denoising and phase adjustment on the target signal to obtain the cargo contour phase signal for subsequent 3D point cloud modeling. A three-dimensional point cloud model acquisition unit is used to generate a three-dimensional point cloud model of the cargo based on the echo signal. The cargo parameter acquisition module is used to parse the three-dimensional point cloud model and extract cargo parameters. The cargo parameters include overall length, width, height, corner positions and surface flatness, and points with curvature greater than a preset curvature threshold are identified as corner points. A wrap-around packaging trajectory generation module is used to generate a wrap-around packaging trajectory based on the cargo parameters, employing a path planning strategy that combines longitudinal layering and lateral trajectory; including: The height adjustment unit is used to divide the packaging path into multiple height levels according to the height of the goods. The wrapping head rises layer by layer along the lifting direction, so that the height of the wrapping head is adjusted according to the level. The transverse spiral winding trajectory generation unit is used to generate a spiral trajectory of the wrapping head relative to the goods based on the center coordinates of the pallet in each height level, so that the wrapping head forms a transverse spiral winding trajectory relative to the goods in the transverse direction. A transverse winding number increasing unit is used to increase the transverse winding number of the film head in the height level including the corner points; An adaptive adjustment module is used to dynamically correct the wrapping trajectory when cargo shaking or wrapping head deviation is detected, so as to achieve adaptive adjustment of the packaging path.

Images