Full-automatic precise positioning and spraying system for liquor bottles

Abstract

The application discloses a kind of liquor bottle full-automatic precision positioning spray systems, belong to the technical field of spraying equipment, it includes the surface topography data and space pose data of liquor bottle are acquired and are carried out feature fusion, generate multidimensional topographic feature vector;From multidimensional topographic feature vector analysis extraction positioning deviation feature, generate attitude correction instruction and dynamic clamping parameter;According to attitude correction instruction and dynamic clamping parameter, establish standard spraying pose;Combining standard spraying pose with multidimensional topographic feature vector carries out trajectory planning and parameter mapping, generates dynamic spraying trajectory and real-time spraying parameter;According to dynamic spraying trajectory and real-time spraying parameter, adaptive spraying operation is carried out, and the liquor bottle that surface coating is completed is obtained.The application generates multidimensional topographic feature vector by intelligentization, attitude correction and clamping parameter decision-making, standard spraying pose establishment, trajectory planning and parameter optimization and adaptive spraying function, realizes full-automatic and high-precision spraying operation to liquor bottle.

Description

Technical Field

[0001] This invention relates to the field of spraying equipment technology, and in particular to a fully automatic precision positioning spraying system for liquor bottles. Background Technology

[0002] Automated spray coating systems are key equipment in modern manufacturing for coating workpiece surfaces, widely used in industries such as automotive, furniture, consumer electronics, and packaging. Spray coating glass products with complex curved surfaces and artistic shapes, such as liquor bottles, not only enhances their aesthetic value but also imparts unique physical properties. The core of this technology lies in using automated devices such as robotic arms to drive spray guns, uniformly covering the workpiece surface with coating.

[0003] Among related technologies, Chinese invention patent CN115283172B discloses a robot automatic spraying method based on point cloud processing, including a six-axis robot, a three-dimensional reconstruction device, a target freeform surface, and a spray gun, used to perform automatic spraying actions according to the point cloud processing results and process parameter requirements; the data processing process includes preprocessing and template making, point cloud feature extraction and description, point cloud registration, spraying template alignment, acquisition of spraying trajectory, and adjustment of spraying parameters and posture, used to process freeform surface point cloud data and template matching, can update the template database in real time and automatically match suitable spraying templates according to point cloud features, guiding the configuration of spraying process and spray gun motion parameters.

[0004] However, the aforementioned existing technical solutions have the following technical defects: They only achieve trajectory planning through point cloud registration and template alignment, without performing closed-loop posture correction and flexible clamping control for fragile glass products like liquor bottles. This fails to eliminate material positioning deviations, easily leading to stress concentration in the bottle or clamping damage. Furthermore, they only match spraying parameters based on curved surface features, without constructing a multi-dimensional morphological feature vector that integrates surface topography, texture, and spatial posture. This results in insufficient accuracy in recognizing complex curvature areas and stress-sensitive areas of the bottle, easily leading to uneven spraying, drips, and missed spraying. The lack of a standard closed-loop verification mechanism for spraying posture means that minor posture disturbances generated during clamping cannot be compensated, causing deviations between the actual and theoretical spraying postures, affecting spraying consistency and finished product yield. Finally, they fail to integrate post-processing steps such as paint mist treatment, VOCs catalytic decomposition, and hot air circulation curing into the liquor bottle spraying scenario, making it impossible to achieve full-process automation and green production. Summary of the Invention

[0005] To address the aforementioned issues, this invention provides a fully automated and precise positioning and spraying system for liquor bottles. Through intelligent multi-dimensional morphological feature vector generation, posture correction and clamping parameter decision-making, standard spraying pose establishment, trajectory planning and parameter optimization, and adaptive spraying functions, it achieves fully automated and high-precision spraying operations on liquor bottles.

[0006] The above objectives can be achieved through the following approach:

[0007] A fully automated precision positioning and spraying system for liquor bottles includes: acquiring surface topography data and spatial pose data of the liquor bottle and performing feature fusion to generate a multi-dimensional topography feature vector; analyzing and extracting positioning deviation features from the multi-dimensional topography feature vector to generate attitude correction commands and dynamic clamping parameters; establishing a standard spraying pose based on the attitude correction commands and dynamic clamping parameters; performing trajectory planning and parameter mapping by combining the standard spraying pose and the multi-dimensional topography feature vector to generate a dynamic spraying trajectory and real-time spraying parameters; and performing adaptive spraying operations according to the dynamic spraying trajectory and real-time spraying parameters to obtain a liquor bottle with a completed surface coating.

[0008] Furthermore, the multidimensional shape feature vector generation module includes: a three-dimensional model fusion unit, a shape feature encoding unit, and a shape feature vector generation unit; wherein, the three-dimensional model fusion unit is used to acquire a visual image of the liquor bottle and perform image denoising and point cloud registration with the depth point cloud to obtain a fused three-dimensional model; the shape feature encoding unit is connected to the three-dimensional model fusion unit and is used to extract the contour curvature and surface texture of the fused three-dimensional model for feature encoding to generate a shape encoding sequence; the shape feature vector generation unit is connected to the shape feature encoding unit and is used to perform vector concatenation of the shape encoding sequence and the spatial pose data to generate a multidimensional shape feature vector.

[0009] Furthermore, the attitude correction and clamping parameter decision module includes: a deviation feature extraction unit, an attitude correction command generation unit, and a clamping parameter decision unit; wherein, the deviation feature extraction unit is connected to the shape feature vector generation unit, and is used to compare the multi-dimensional shape feature vector with a preset standard bottle shape feature library to extract spatial offset and rotation angle, and generate positioning deviation features; the attitude correction command generation unit is connected to the deviation feature extraction unit, and is used to perform inverse kinematics solution based on the positioning deviation features to generate attitude correction commands; the clamping parameter decision unit is connected to the attitude correction command generation unit, and is used to calculate the stress distribution in the force-sensitive area of ​​the multi-dimensional shape feature vector to generate dynamic clamping parameters.

[0010] Furthermore, the standard spraying pose establishment module includes: a pose correction drive unit, a flexible clamping control unit, and a pose verification and confirmation unit; wherein, the pose correction drive unit is connected to the pose correction command generation unit, and is used to parse and extract the lifting displacement and rotation step size in the pose correction command, drive the lifting turntable to perform spatial displacement, and obtain the target correction pose; the flexible clamping control unit is connected to the clamping parameter decision unit and the pose correction drive unit respectively, and is used to adjust the output torque and contact area of ​​the clamping mechanism according to the dynamic clamping parameters, and flexibly clamp the liquor bottle in the target correction pose to obtain a stable clamping state; the pose verification and confirmation unit is connected to the flexible clamping control unit, and is used to perform closed-loop compensation verification on the real-time pose data in the stable clamping state to establish a standard spraying pose.

[0011] Furthermore, the trajectory planning and parameter optimization module includes: a surface mesh mapping unit, a dynamic spraying trajectory generation unit, and a spraying parameter optimization unit; wherein, the surface mesh mapping unit is connected to the pose verification and confirmation unit, and is used to map the multi-dimensional topographic feature vector to the coordinate system where the standard spraying pose is located to perform surface mesh division, thereby obtaining a mesh surface to be sprayed; the dynamic spraying trajectory generation unit is connected to the surface mesh mapping unit, and is used to perform normal offset and interference checks along the normal vector of the mesh surface to be sprayed, thereby generating a dynamic spraying trajectory; the spraying parameter optimization unit is connected to the dynamic spraying trajectory generation unit, and is used to perform hydrodynamic mapping along the rate of curvature change of the mesh surface to be sprayed based on the dynamic spraying trajectory, thereby generating real-time spraying parameters.

[0012] Furthermore, the adaptive spraying execution module includes: a spraying trajectory analysis unit, a spray gun parameter control unit, and a continuous surface coating unit; wherein, the spraying trajectory analysis unit, connected to the spraying parameter optimization unit, is used to extract the spatial coordinates and attitude angles of each trajectory node from the dynamic spraying trajectory analysis, and control the six-degree-of-freedom spraying robot arm to move to the corresponding node to obtain the current spraying node; the spray gun parameter control unit, connected to the spraying trajectory analysis unit, is used to adjust the atomization pressure and paint flow rate of the intelligent spray gun according to the real-time spraying parameters corresponding to the current spraying node to generate a dynamic atomization cone; the continuous surface coating unit, connected to the spray gun parameter control unit, is used to continuously cover the surface of the liquor bottle with the dynamic atomization cone to obtain a liquor bottle with a completed surface coating.

[0013] Furthermore, the step of performing hydrodynamic mapping along the rate of curvature change of the grid surface to be sprayed based on the dynamic spraying trajectory to generate real-time spraying parameters includes: performing gradient analysis on the curvature difference between adjacent grids in the grid surface to be sprayed to obtain a curvature gradient distribution map; identifying curvature abrupt change regions and smooth regions based on the curvature gradient distribution map to generate region attribute labels; and combining the region attribute labels with preset paint rheological properties to perform parameter matching to generate real-time spraying parameters including atomization pressure, paint flow rate, and spraying distance.

[0014] Furthermore, the step of acquiring visual images of the liquor bottle and performing image denoising and point cloud registration with the depth point cloud to obtain a fused 3D model includes: acquiring multi-angle 2D images of the liquor bottle for edge detection and feature point extraction to obtain 2D contour features; scanning the surface of the liquor bottle to acquire the original depth point cloud for filtering, smoothing, and downsampling to obtain the target depth point cloud; and mapping the 2D contour features to the target depth point cloud for spatial alignment and texture mapping to obtain the fused 3D model.

[0015] Furthermore, after continuously covering the surface of the liquor bottle with the dynamic atomizing cone to obtain a liquor bottle with a completed surface coating, the process further includes: collecting the free paint mist and volatile organic compounds generated during the adaptive spraying operation and performing gas-liquid separation to obtain liquid paint mist and gaseous volatiles; performing flocculation and sedimentation treatment on the liquid paint mist to obtain recycled paint residue, and performing catalytic oxidation decomposition on the gaseous volatiles to obtain emission gases that meet standards; and subjecting the liquor bottle with the completed surface coating to hot air circulation baking to obtain the finished liquor bottle.

[0016] Compared with the prior art, the present invention has the following advantages: This invention constructs a multi-dimensional morphological feature vector to digitally model the three-dimensional geometry, surface texture, and spatial pose of each liquor bottle. This enables the system to accurately perceive the individual differences and real-time status of each object to be processed, providing a high-dimensional decision-making basis for subsequent precise control and improving the system's perception accuracy and adaptability.

[0017] This invention establishes a posture correction and clamping parameter decision-making module as well as a standard spraying posture establishment module. By actively identifying positioning deviations and performing closed-loop correction, it completely solves the core problem of unstable spraying quality caused by inconsistent material postures in traditional assembly line operations. At the same time, the dynamically generated clamping parameters ensure a stable and undamaged fixation of the bottle, guaranteeing the accuracy and reliability of the spraying operation benchmark.

[0018] The trajectory planning and parameter optimization module of this invention can dynamically generate a spraying trajectory based on the actual three-dimensional shape of the liquor bottle and along its complex surface curvature, and optimize the spraying process parameters in real time. This ensures that the spray gun always maintains the best distance and posture with the spraying surface, and matches the optimal paint flow rate and atomization effect, thereby improving the uniformity of the coating and the yield, and avoiding common defects such as sagging and missed spraying.

[0019] This invention, through an integrated adaptive spraying execution module, tightly couples perception, decision-making, and physical execution, achieving full-process automation and intelligence from liquor bottle identification, positioning, clamping to final spraying. This reduces reliance on manual skills, improves production efficiency and process consistency, and enhances the flexibility of the production line to cope with changes in different bottle shapes.

[0020] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a framework diagram of a fully automatic precision positioning and spraying system for liquor bottles according to an embodiment of the present invention.

[0023] Figure 2 This is a schematic diagram of the structure of a fully automatic precision positioning and spraying system for liquor bottles according to an embodiment of the present invention.

[0024] Figure 3 This is a flowchart of the attitude correction and clamping parameter decision module according to an embodiment of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0026] Reference Figure 1One embodiment of the present invention proposes a fully automatic and precise positioning and spraying system for liquor bottles. Through intelligent multi-dimensional morphological feature vector generation, posture correction and clamping parameter decision-making, standard spraying posture establishment, trajectory planning and parameter optimization, and adaptive spraying function, the system realizes fully automatic and high-precision spraying operations on liquor bottles.

[0027] like Figure 2 As shown, the system in this embodiment specifically includes: a multi-dimensional shape feature vector generation module, a posture correction and clamping parameter decision module, a standard spraying pose establishment module, a trajectory planning and parameter optimization module, and an adaptive spraying execution module; wherein, S1. The multi-dimensional morphological feature vector generation module is used to acquire the surface morphological data and spatial pose data of the liquor bottle and perform feature fusion to generate a multi-dimensional morphological feature vector. Furthermore, the multi-dimensional shape feature vector generation module includes: a three-dimensional model fusion unit, a shape feature encoding unit, and a shape feature vector generation unit; wherein, The three-dimensional model fusion unit is used to acquire the visual image and depth point cloud of the liquor bottle, perform image denoising and point cloud registration, and obtain a fused three-dimensional model. The shape feature encoding unit is connected to the three-dimensional model fusion unit and is used to extract the contour curvature and surface texture of the fused three-dimensional model for feature encoding to generate a shape encoding sequence. The shape feature vector generation unit is connected to the shape feature encoding unit and is used to concatenate the shape encoding sequence with the spatial pose data to generate a multidimensional shape feature vector.

[0028] Specifically, the 3D model fusion unit begins operation. A structured light scanner and a multi-angle industrial camera, deployed above the workstation, simultaneously acquire visual images and depth point clouds of the liquor bottle. The structured light scanner obtains the raw depth point cloud data of the bottle surface with a scanning accuracy of 0.1 to 0.5 mm. For the acquired visual images, a Gaussian filtering algorithm is used for image denoising to eliminate sensor noise; the Gaussian kernel size is set between 3×3 and 7×7 depending on the image resolution. For the depth point cloud, a voxel downsampling algorithm is executed to reduce data redundancy; the sampling resolution is set between 0.2 and 0.8 mm. An iterative nearest-point algorithm is used to precisely register the multi-view point clouds, ensuring the integrity and accuracy of the point cloud model. Finally, the denoised visual images are used as texture information and applied to the registered depth point cloud surface using UV mapping technology to generate a fused 3D model with realistic surface details.

[0029] The topographic feature encoding unit then processes the data. This unit receives the fused 3D model as input and its task is to convert its geometric and texture information into a quantifiable numerical sequence. The unit first performs principal curvature analysis on the fused 3D model. Principal curvature analysis is a differential geometric method used to quantify the degree of curvature of a surface at a given point, sampling along the model surface. At key points, the maximum and minimum principal curvatures are calculated for each point to form a curvature feature set. Simultaneously, a local binary mode operator is used to extract texture features from the model surface. This operator generates a texture descriptor insensitive to illumination changes by comparing the grayscale values ​​of the center pixel with those of its neighbors. The system then normalizes the calculated curvature sequence and texture feature sequence and concatenates them to form a shape encoding sequence.

[0030] The shape feature vector generation unit completes data integration. The input to this unit is the shape encoding sequence generated in the previous step, and the spatial pose data synchronously measured by the 3D scanner during the initial scanning phase. The spatial pose data is a six-dimensional vector describing the position and orientation of the bottle in the device coordinate system, including three translational components and three rotational components, such as Euler angles. This unit performs a vector concatenation operation, merging this spatial pose data with the shape encoding sequence to generate the final multi-dimensional shape feature vector. This concatenation process can be represented by the following formula: , in, This represents the final generated multidimensional morphological feature vector; It is a normalized morphology coding sequence, composed of the curvature and texture features of the bottle surface; This data describes the spatial pose of the liquor bottle, obtained through 3D scanning. The multi-dimensional shape feature vector generation module outputs a high-dimensional digital descriptor. This vector not only includes the fine geometric features of the bottle surface, such as contour curvature and surface texture, but also accurately records its instantaneous state in three-dimensional space. This unified data structure provides a precise and unified data source for the subsequent pose correction and clamping parameter decision modules to accurately calculate positioning deviations, generate correction commands, and plan spraying paths. It is the core foundation for achieving fully automated and precise positioning spraying.

[0031] For example, regarding the specific implementation process of the multi-dimensional shape feature vector generation module, the 3D model fusion unit first acquires the original depth point cloud data of the surface of the liquor bottle using a structured light scanner with a scanning accuracy of 0.5 mm, while simultaneously acquiring visual images using a multi-angle industrial camera. The system uses a Gaussian filtering algorithm with a kernel size of 5×5 to denoise the visual images and performs a voxel downsampling algorithm with a sampling resolution of 0.5 mm on the depth point cloud to reduce data redundancy. Then, an iterative nearest-point algorithm is used to perform fine registration of the multi-view point clouds. Finally, the denoised visual images are attached to the registered depth point cloud surface using UV mapping technology to generate a fused 3D model. Subsequently, the shape feature encoding unit performs principal curvature analysis on the fused 3D model, samples key points, and calculates the maximum principal curvature to be 0.8 and the minimum principal curvature to be 0.2. At the same time, the local binary mode operator is used to extract surface texture features, resulting in a normalized feature value of 0.6. The system normalizes the calculated curvature sequence and texture feature sequence and concatenates them to form a shape encoding sequence. =[0.8,0.2,0.6]. Next, the shape feature vector generation unit acquires the spatial pose data simultaneously measured by the 3D scanner. =[10,20,30,0,0,90], this data precisely contains three translational components 10, 20, and 30, and three rotational components 0, 0, and 90. The shape feature vector generation unit substitutes the above two sets of vectors into the concatenation formula. Perform vector concatenation to obtain a complete multidimensional feature vector. This successfully integrated the fine geometric features of the liquor bottle surface with the real-time state in three-dimensional space into a unified high-dimensional digital descriptor.

[0032] Furthermore, the process of obtaining a fused 3D model by acquiring a visual image of the liquor bottle and performing image denoising and point cloud registration with the depth point cloud includes: Multi-angle two-dimensional images of the liquor bottle were acquired for edge detection and feature point extraction to obtain two-dimensional contour features. The surface of the liquor bottle is scanned to obtain the original depth point cloud, which is then filtered, smoothed, and downsampled to obtain the target depth point cloud. The two-dimensional contour features are mapped to the target depth point cloud for spatial alignment and texture mapping to obtain a fused three-dimensional model.

[0033] Specifically, multiple industrial cameras deployed around the workstation simultaneously capture multi-angle two-dimensional images of the liquor bottles from different angles. These cameras typically have a resolution of 2 to 5 megapixels. For each captured image, the system uses the Canny edge detection algorithm, which effectively identifies edge contours with drastic brightness changes. By setting two thresholds (high and low), noise is suppressed and edges are connected, thus extracting clear two-dimensional contour features. Simultaneously, feature point extraction algorithms such as SIFT or SURF are applied to find key feature points in the image that are insensitive to scale, rotation, and illumination changes. These feature points provide a stable matching basis for subsequent image-point cloud registration.

[0034] A structured light scanner is used to perform a 3D scan of the surface of the liquor bottle, acquiring the raw depth point cloud. The raw point cloud data is massive and noisy, thus requiring preprocessing. The system first applies a statistical outlier removal algorithm to eliminate isolated noise points caused by measurement errors or environmental reflections. Next, bilateral filtering or moving least squares filtering is used to smooth the point cloud, effectively removing minor surface undulations while preserving the sharp features of the bottle surface. Subsequently, to improve subsequent processing efficiency, a voxel grid downsampling algorithm is executed to reduce the point cloud density to between 100 and 500 points per square centimeter, obtaining a target depth point cloud that maintains the geometric shape and facilitates computation.

[0035] The system fuses 2D features with 3D point clouds. The core of this step is to accurately map 2D contour features and texture information onto the target depth point cloud, achieving spatial alignment and texture mapping. The system uses camera calibration parameters and scanner intrinsic and extrinsic parameters to correlate the pixel coordinates of the 2D image with the spatial coordinates of the 3D point cloud. By matching the 2D feature points extracted in previous steps with the corresponding geometric features on the 3D model, the system calculates an accurate transformation matrix. This process can be described by the following formula: , in, Represents the coordinates of the feature point in three-dimensional space; These are the homogeneous coordinates of the feature point in the two-dimensional image; It is a 3×4 projection transformation matrix that contains the camera's intrinsic and extrinsic parameters, as well as the rotation and translation relationships between the 2D image plane and 3D space. By solving and optimizing this matrix, the system achieves precise spatial alignment between the 2D image and the 3D point cloud. After alignment, the 2D image from multiple angles is used as a texture source. Through UV mapping technology, the image pixel color information is applied to the corresponding surface of the target depth point cloud, ultimately generating a realistic and detailed fused 3D model.

[0036] For example, regarding the specific implementation process of the fused 3D model generation module, the system first simultaneously acquires multi-angle 2D images of the liquor bottle using multiple industrial cameras deployed around the workstation. The 3D model fusion unit applies the Canny edge detection algorithm to process one of the images with a resolution of 5 megapixels, identifying and extracting a clear outline of the bottle's edge. Simultaneously, a structured light scanner is activated to scan the surface of the liquor bottle to obtain the original depth point cloud. The system first applies a statistical outlier removal algorithm to remove isolated noise points, then uses moving least squares for filtering and smoothing, and executes a voxel grid downsampling algorithm to reduce the point cloud density to 300 points per square centimeter, obtaining the target depth point cloud. During the fusion of 2D features and 3D point clouds, the system uses camera calibration parameters to establish the relationship between the pixel coordinates of the 2D image and the spatial coordinates of the 3D point cloud. Assuming that a feature point extracted from the image has 2D homogeneous coordinates... The system is defined as [640, 480, 1], obtained through pre-calibration. Projection transformation matrix Perform spatial alignment calculations. Set the matrix. The first row is [0.5, 0.1, 0.2, 10], the second row is [0.1, 0.6, 0.3, 20], and the third row is [0.01, 0.02, 1.0, 5]. These values ​​are substituted into the projection transformation formula to perform the calculation. Through matrix multiplication, the first component of the feature point's coordinates in 3D space is... The second component is The third component is The three-dimensional spatial coordinates of the point are calculated. The values ​​are [378.2, 372.3, 22]. Finally, the system uses the optimized transformation matrix to take the multi-angle 2D image as a texture source, and uses UV mapping technology to attach the pixel color information to the corresponding surface of the target depth point cloud, generating a richly detailed fused 3D model.

[0037] S2. The attitude correction and clamping parameter decision module is used to analyze and extract positioning deviation features from the multi-dimensional morphological feature vector to generate attitude correction commands and dynamic clamping parameters.

[0038] Furthermore, the attitude correction and clamping parameter decision module includes: a deviation feature extraction unit, an attitude correction command generation unit, and a clamping parameter decision unit; wherein,

[0039] The deviation feature extraction unit is connected to the shape feature vector generation unit and is used to compare the multi-dimensional shape feature vector with a preset standard bottle shape feature library, extract spatial offset and rotation angle, and generate positioning deviation features.

[0040] The attitude correction command generation unit is connected to the deviation feature extraction unit and is used to perform inverse kinematics solution based on the positioning deviation features to generate attitude correction commands.

[0041] The clamping parameter decision unit is connected to the attitude correction command generation unit and is used to calculate the stress distribution of the force-sensitive area in the multi-dimensional topographic feature vector to generate dynamic clamping parameters.

[0042] Specifically, the workflow diagram of the attitude correction and clamping parameter decision module is as follows: Figure 3 As shown, the deviation feature extraction unit receives the multidimensional shape feature vector output by the shape feature vector generation unit and initiates comparison analysis. The system pre-stores a standard bottle shape feature library, which contains standard multidimensional shape feature vectors of the target liquor bottle at an ideal workstation. This library is a pre-established database containing standard multidimensional shape feature vectors of the target liquor bottle in a standard spraying pose. These standard multidimensional shape feature vectors include a standard contour curvature sequence, a standard surface texture feature sequence, and a standard spatial pose six-dimensional vector. The standard bottle shape feature library is generated by performing multiple 3D scans and visual acquisitions on standard sample bottles to obtain surface shape, texture, and standard pose data. These data are then processed through feature fusion and normalization, with one standard feature data point corresponding to each bottle shape. This unit accurately quantifies the positioning error by calculating the difference between the current liquor bottle's multidimensional shape feature vector and the standard vector. This difference is mainly calculated by extracting the spatial pose data components from the vectors to obtain the spatial offset and rotation angle, which together constitute the positioning deviation feature. Spatial offset refers to the difference in three-dimensional coordinates between the current center of the bottle and the standard position center, while rotation angle refers to the angular difference between the current bottle posture and the standard posture, usually expressed as quaternions or Euler angles. Positioning deviation characteristics. It can be described by the following formula: , in, Represents the feature vector of positioning deviation; This represents a standard spatial pose six-dimensional vector retrieved from a standard bottle-shaped feature library, containing three translational components and three rotational components; This represents the actual spatial pose six-dimensional vector extracted from the current multidimensional topographic feature vector. The tolerance range of this calculation result is usually set between 0.1-5 mm for translation and 0.1-2 degrees for rotation; exceeding this range will trigger correction.

[0043] The attitude correction command generation unit performs inverse kinematics solving based on the extracted positioning deviation features. Inverse kinematics solving here means, based on the known final desired pose correction amount, deducing the specific motion parameters of each joint required to achieve that correction. For the lifting turntable in this system, this unit needs to convert the vertical offset and rotation angle around the vertical axis in the positioning deviation features into the number of pulses or the stroke of the servo motor driving the lifting turntable. For example, the vertical offset is converted into the lifting displacement of the lifting platform, and the rotation angle difference is converted into the rotation step size of the turntable. Finally, a series of precise low-level hardware control codes, i.e., attitude correction commands, are generated.

[0044] The clamping parameter decision unit operates in parallel, analyzing the geometric information in the multi-dimensional morphological feature vector. The purpose of this unit is to generate a set of dynamic clamping parameters that ensure clamping stability while avoiding damage to the bottle. It first uses curvature and texture data from the morphological encoding sequence to identify stress-sensitive areas on the bottle surface, such as raised parts of the bottle's pattern, thin areas of the neck, or sharp corners with drastic curvature changes. Next, it performs simplified stress distribution calculations on these areas, evaluating stress concentration at key points on the bottle surface under virtual clamping force. Through an optimization algorithm, it seeks a combination of clamping force and contact point position that minimizes the maximum stress value, while ensuring the clamping force is sufficient to overcome gravity and inertial forces during subsequent processing. The optimization algorithm uses minimizing the maximum stress value on the bottle surface as the optimization objective, with constraints including a clamping force not less than a safe clamping threshold, contact points avoiding stress-sensitive areas, and the maximum stress not exceeding the allowable stress of the glass. It employs a numerical iterative optimization method to find the optimal combination. The calculation results are ultimately quantified into dynamic clamping parameters such as the output torque of the driving clamping mechanism, the clamping speed, and the contact area of ​​the flexible gripper. As a result, the attitude correction and clamping parameter decision module outputs two sets of key, mutually coordinated decision data: attitude correction commands for driving the lifting turntable to perform physical position correction, and dynamic clamping parameters for controlling the clamping mechanism to perform safe and stable fixation.

[0045] For example, in the specific implementation process of the attitude correction and clamping parameter decision module, the deviation feature extraction unit receives the multi-dimensional shape feature vector output by the shape feature vector generation unit and initiates a comparison analysis. The system retrieves the standard spatial pose six-dimensional vector from the internally pre-stored standard bottle-shaped feature library. Assuming the translation components are set to 100 mm, 200 mm, and 300 mm, and the rotation components are set to 0 degrees, 0 degrees, and 90 degrees, the system simultaneously extracts the actual spatial pose six-dimensional vector from the current multidimensional topographic feature vector. The translational components were detected to be 100.5 mm, 199 mm, and 302 mm, and the rotational components were 0 degrees, 0 degrees, and 91.5 degrees. The unit then substituted these two sets of vectors into the formula. The spatial offset and rotation angle are calculated. After subtracting the values ​​in the corresponding dimensions, the positioning deviation feature vector is obtained. The translational component deviations are -0.5 mm, 1 mm, and -2 mm, while the rotational component deviations are 0 degrees, 0 degrees, and -1.5 degrees. Since both the -2 mm translational deviation and the -1.5 degree rotational deviation exceed the preset tolerance range for error correction, the system transmits this positioning deviation feature to subsequent stages for motion compensation. The attitude correction command generation unit performs inverse kinematics solving based on the extracted positioning deviation features, converting the -2 mm vertical offset into a 2 mm downward displacement to drive the lifting turntable servo motor, and converting the -1.5 degree yaw angle difference into a 1.5 degree counterclockwise rotational step to drive the turntable, thus generating a series of precise underlying hardware control codes, i.e., attitude correction commands. The parallel-working clamping parameter decision unit simultaneously analyzes the geometric information in the multi-dimensional topographic feature vector, using curvature and texture data to identify the sharp corners of the bottle surface with drastic curvature changes as force-sensitive areas. The unit then performs a simplified stress distribution calculation on the aforementioned stress-sensitive areas. With the goal of minimizing the maximum stress value on the bottle surface, and under the constraint that the clamping force is not less than the safe clamping threshold, the optimal combination is solved by numerical iteration. Finally, the calculation results are quantified into dynamic clamping parameters with a drive clamping mechanism output torque of 2.5 N·m and a flexible gripper contact area of ​​15 square centimeters. These two sets of data, working together, ultimately provide a quantitative basis for subsequent modules to achieve accurate correction and safe fixation.

[0046] S3. The standard spraying posture establishment module is used to drive the lifting turntable and the clamping mechanism to perform posture adjustment and clamping according to the posture correction command and dynamic clamping parameters, and establish a standard spraying posture.

[0047] Furthermore, the standard spraying pose establishment module includes: a pose correction drive unit, a flexible clamping control unit, and a pose verification and confirmation unit; wherein,

[0048] The pose correction driving unit is connected to the pose correction command generation unit and is used to parse and extract the lifting displacement and rotation step size in the pose correction command, drive the lifting turntable to perform spatial displacement, and obtain the target corrected pose.

[0049] The flexible clamping control unit is connected to the clamping parameter decision unit and the pose correction drive unit respectively. It is used to adjust the output torque and contact area of ​​the clamping mechanism according to the dynamic clamping parameters, so as to flexibly clamp the liquor bottle in the target correction pose and obtain a stable clamping state.

[0050] The pose verification and confirmation unit is connected to the flexible clamping control unit and is used to perform closed-loop compensation verification on the real-time pose data under the stable clamping state to establish a standard spraying pose.

[0051] Specifically, the pose correction drive unit receives attitude correction commands from the attitude correction command generation unit. This unit first parses the commands, extracting specific motion control parameters such as lifting displacement and rotation step size. Lifting displacement is typically accurate to 0.05 to 0.1 millimeters, while rotation step size is accurate to 0.1 to 0.5 degrees. Subsequently, the unit converts these parameters into control signals for the lifting turntable's servo motors, driving the vertical and rotational axes of the lifting turntable with high precision until the liquor bottle reaches the target corrected pose after eliminating positioning deviations. The entire drive process is performed under closed-loop control to ensure positioning accuracy.

[0052] After the lifting turntable completes its displacement, the flexible clamping control unit is activated. This unit synchronously receives dynamic clamping parameters from the clamping parameter decision unit and a status completion signal from the pose correction drive unit. Based on the dynamic clamping parameters, this unit precisely adjusts the output torque and stroke of the servo motor or pneumatic components in the clamping mechanism, thereby controlling the clamping force and contact area of ​​the flexible gripper. For example, for areas with complex reliefs on the bottle, the system uses lower torque and a larger contact area to distribute pressure and prevent damage. After performing the flexible clamping operation on the liquor bottle in the target correction pose, the bottle reaches a mechanically stable clamping state.

[0053] The pose verification and confirmation unit performs final pose confirmation on the liquor bottle in a stable clamping state. This unit's task is to eliminate minor displacement disturbances that may be introduced during clamping through closed-loop compensation verification. It remeasures the real-time pose data of the liquor bottle after clamping using a high-precision encoder mounted on the lifting turntable or a separate vision sensor. This real-time data is compared with the target corrected pose; if a deviation exceeds a preset threshold, a fine-tuning compensation amount is generated. This compensation process can be expressed by the following formula: , in, The final established standard spraying pose is represented by a six-dimensional vector. It is the target correction pose before clamping, which is also a six-dimensional vector and is implemented by the pose correction driving unit; The pose verification and confirmation unit calculates a six-dimensional compensation vector based on the difference between the real-time pose data and the target pose. This compensation vector is then sent back to the pose correction drive unit for one or more iterative fine-tuning iterations until the error between the real-time pose and the target pose converges within the allowable range. Only then does the system finally confirm the establishment of the standard spraying pose. The output of the standard spraying pose establishment module is a liquor bottle that has been physically executed and verified through closed-loop control, and is in a precise and stable state. It is no longer an abstract data or instruction, but a physical entity whose position and attitude height in the device coordinate system are determined and repeatable. This determined standard spraying pose provides an absolute, error-free reference benchmark for the subsequent trajectory planning and parameter optimization modules, and is the physical basis for achieving high-precision adaptive spraying.

[0054] For example, in the specific implementation process of the standard spraying pose establishment module, the pose correction drive unit first receives the pose correction command and parses out the specific motion control parameters. Assuming the parsed lifting displacement is 2 mm and the rotation step is -1.5 degrees, this unit drives the servo motor of the lifting turntable to perform displacement and rotation actions, bringing the liquor bottle to the target correction pose. Subsequently, the flexible clamping control unit adjusts the output torque of the clamping mechanism to 2.5 N·m based on the received dynamic clamping parameters, and makes the contact area of ​​the flexible gripper reach 15 square centimeters, performing flexible clamping on the liquor bottle in the target correction pose to achieve a stable clamping state. The pose verification and confirmation unit uses a vision sensor to remeasure the real-time pose data of the liquor bottle after it is clamped. Assuming the measured real-time pose six-dimensional vector has a vertical height component of 301.8 mm and a rotation angle around the vertical axis of 89.9 degrees, and the target correction pose vector realized by the pose correction drive unit... The corresponding components are 302 mm and 90 degrees, respectively. This unit is defined by the formula... Closed-loop compensation verification is performed. In this calculation process, the system first calculates the compensation vector. The vertical displacement compensation is 302 - 301.8 = 0.2 mm, and the rotation angle compensation is 90 - 89.9 = 0.1 degrees. The system superimposes this compensation vector onto the target correction pose to calculate the final established standard spraying pose vector. The pose correction drive unit then drives the motor again to make fine adjustments based on the compensation amount until the error between the real-time pose and the target pose converges within the allowable range, thereby finally physically establishing the standard spraying pose and providing an accurate reference benchmark for subsequent trajectory planning.

[0055] S4. The trajectory planning and parameter optimization module is used to combine the standard spraying pose and the multi-dimensional morphological feature vector to perform trajectory planning and parameter mapping, and generate dynamic spraying trajectory and real-time spraying parameters.

[0056] Furthermore, the trajectory planning and parameter optimization module includes: a surface mesh mapping unit, a dynamic spraying trajectory generation unit, and a spraying parameter optimization unit; wherein, The surface mesh mapping unit is connected to the pose verification and confirmation unit and is used to map the multi-dimensional morphology feature vector to the coordinate system where the standard spraying pose is located to perform surface mesh division and obtain the mesh surface to be sprayed. The dynamic spraying trajectory generation unit is connected to the surface mesh mapping unit and is used to perform normal offset and interference checks along the normal vector of the mesh surface to be sprayed, thereby generating a dynamic spraying trajectory. The spraying parameter optimization unit is connected to the dynamic spraying trajectory generation unit and is used to perform fluid dynamic mapping along the rate of curvature change of the grid surface to be sprayed based on the dynamic spraying trajectory to generate real-time spraying parameters.

[0057] Specifically, the surface mesh mapping unit receives standard spraying pose information established by the pose verification and confirmation unit, as well as multi-dimensional topographic feature vectors. The core task of this unit is coordinate system unification and model discretization. It first applies a rigid body transformation matrix to accurately map the 3D bottle model described by the multi-dimensional topographic feature vectors from the initial scanning coordinate system to the robot's working coordinate system defined by the standard spraying pose. Subsequently, a moving cube algorithm is used to triangularly mesh the mapped model surface. The mesh density is a key parameter, typically set between 1 and 5 millimeters for the triangle side length based on spraying accuracy requirements, thus generating a mesh surface composed of thousands of triangular faces for path planning.

[0058] The dynamic spraying trajectory generation unit performs path calculations based on the mesh surface to be sprayed. Its goal is to generate a motion trajectory that ensures the spray gun tip maintains the optimal spraying distance and posture with the bottle surface at all times. This unit traverses all mesh vertices on the mesh surface to be sprayed, calculating the normal vector of each vertex. The normal vector is a unit vector perpendicular to the tangent plane at that point, representing the orientation of the surface at that location. Then, it offsets outward along the normal vector direction of each vertex by a preset spraying distance, which is set between 150 and 250 mm depending on the paint characteristics and spraying process, forming a series of trajectory candidate points. The system connects these candidate points to initially form the center point trajectory of the spray gun. Simultaneously, the system performs interference checks, using a collision simulation between the robot's kinematics model and the environment model to ensure that the generated trajectory will not cause the robotic arm to collide with the bottle, gripper, or itself. If interference is detected, the attitude angle of the trajectory points is locally adjusted or transition points are added to avoid it. This process can be described by the following formula: , in, Represents the first on the dynamic spraying trajectory Coordinates of a trajectory point; It is the corresponding number on the grid surface to be sprayed. Vertex coordinates; The optimal spraying distance is a scalar value. yes The unit normal vector at the point. The final generated dynamic spraying trajectory is a sequence of six-dimensional pose points with a certain time order.

[0059] The spraying parameter optimization unit matches optimal spraying process parameters to each point on the dynamic spraying trajectory. The unit's input consists of the generated dynamic spraying trajectory and the geometric information of the mesh surface to be sprayed. It synchronously analyzes the rate of curvature change on the corresponding mesh surface along the dynamic spraying trajectory. The rate of curvature change reflects the smoothness of the surface; for example, the rate of curvature change is close to zero in flat areas of the bottle body, while it is larger at corners of the bottle shoulder or bottom. This unit maps different ranges of curvature change rates to different combinations of spraying parameters using a pre-established fluid dynamics mapping model. This model, based on extensive experimental data, correlates the rheological properties of the coating with surface geometry, generating real-time spraying parameters including core parameters such as atomization pressure and coating flow rate. For example, in areas with drastic curvature changes, the system reduces the coating flow rate and atomization pressure to prevent coating accumulation and splashing.

[0060] For example, regarding the specific implementation process of the trajectory planning and parameter optimization module, the surface mesh mapping unit receives standard spraying pose information and multi-dimensional topographic feature vectors established by the pose verification and confirmation unit. First, it applies a rigid body transformation matrix to map the bottle's 3D model from the initial scanning coordinate system to the robot's working coordinate system. Then, it uses a moving cube algorithm to divide the model surface into triangular meshes, setting the side length of each triangle to 2 mm to obtain a mesh surface to be sprayed composed of thousands of triangular facets. The dynamic spraying trajectory generation unit traverses the vertices of this mesh surface and calculates the normal vectors. Assuming the coordinates of the first vertex on the mesh surface to be sprayed... The value is [50, 60, 100], and its corresponding unit normal vector is... Set the optimal spraying distance to [0,0,1]. The value is 200 mm. Substitute this value into the trajectory calculation formula. Perform the calculation. The coordinates of the first trajectory point on the dynamic spraying trajectory are obtained through calculation. The coordinates are [50, 60, 300]. The system then performs an interference check on this point and connects the trajectory points to form a six-dimensional pose point sequence. Simultaneously, the spraying parameter optimization unit performs fluid dynamics mapping, calculates the curvature difference between adjacent meshes for gradient analysis, assuming the mesh... The representative curvature value 0.15, adjacent grid The representative curvature value The value is 0.12, which is then substituted into the curvature gradient calculation formula. To obtain an approximate value of the curvature gradient The gradient value is 0.03. Since this gradient value is lower than the preset threshold, the system assigns a flat area attribute label to the region and generates real-time spraying parameters including an atomization pressure of 0.4 MPa, a paint flow rate of 200 ml / min, and a spraying distance of 200 mm by matching with the process database. This provides precise path and process guidance for the robotic arm's adaptive operation.

[0061] Furthermore, the step of performing hydrodynamic mapping along the rate of curvature change of the grid surface to be sprayed based on the dynamic spraying trajectory to generate real-time spraying parameters includes: Gradient analysis is performed on the curvature difference between adjacent grids in the grid surface to be sprayed to obtain a curvature gradient distribution map; Based on the curvature gradient distribution map, identify regions with abrupt curvature changes and regions with gentle curvature, and generate region attribute labels; By combining the regional attribute tags with the preset rheological properties of the coating, real-time spraying parameters including atomization pressure, coating flow rate and spraying distance are generated.

[0062] Specifically, curvature gradient analysis is performed on the mesh surface to be sprayed generated in the previous steps. For each triangular mesh patch on the mesh surface, the Gaussian curvature or average curvature at its centroid is calculated as the representative curvature value of that mesh. Then, all adjacent mesh pairs are traversed, and the curvature difference between them is calculated. Gradient analysis here means quantifying the geometric complexity of the surface by calculating the rate and direction of curvature change on the surface. Areas with large curvature differences indicate abrupt changes in surface shape. These calculated curvature difference data are integrated into a curvature gradient distribution map covering the entire bottle surface, which visually demonstrates the degree of curvature in different regions of the bottle surface. The calculation of the curvature gradient can be expressed by the following simplified formula: , in, Representing adjacent grids and Approximate values ​​of the curvature gradient between them; and They are grids and grid The representative curvature value is obtained by averaging or interpolating the principal curvatures of the mesh vertices.

[0063] The system classifies regions based on the generated curvature gradient distribution map. By setting a curvature gradient threshold, regions with gradient values ​​greater than a specific value are identified as regions with abrupt curvature changes, such as bottle shoulders, bottom corners, or embossed edges. Regions with gradient values ​​lower than this threshold are identified as smooth regions, such as the main body of the bottle. The system assigns a region attribute label, such as abrupt or smooth, to each grid on the surface to be coated, thereby discretizing the continuous geometric surface into characteristic regions with different process requirements.

[0064] The system performs parameter matching to generate the final real-time spraying parameters. This step is based on a pre-set process parameter database or expert system rule base. The process parameter database or expert system rule base is a process knowledge base established in advance through offline calibration experiments. The database pre-stores the corresponding matching relationships between curvature region attribute tags and spraying process parameters, including three core sets of process parameters: atomization pressure, paint flow rate, and spraying distance. This database obtains the optimal spraying parameters for each region through numerous trial spraying experiments on different surface areas of a standard liquor bottle, and summarizes these into rules. Smooth areas are matched with higher atomization pressure and larger paint flow rate to improve spraying efficiency and coverage uniformity, while areas with abrupt curvature changes are matched with lower atomization pressure and smaller paint flow rate to avoid paint dripping, accumulation, and splashing at edges, corners, and embossed areas. Based on the region attribute tags of the grid surface to be sprayed, the system retrieves and calls the corresponding process parameter combinations from the database, thereby generating real-time spraying parameters that are adapted to the surface morphology in real time. This library incorporates the rheological properties of the paints used, such as viscosity and surface tension, to pre-determine optimal spraying parameter ranges for different region attribute labels. For example, for regions marked as abrupt changes, the system will match lower paint flow rate and atomization pressure, as well as a potentially longer spraying distance, to prevent paint buildup or sagging at sharp edges due to surface tension. Conversely, for gentler regions, higher paint flow rate and atomization pressure can be used to improve spraying efficiency. Through this matching mechanism, the system generates a specific set of real-time spraying parameters, including atomization pressure, paint flow rate, and spraying distance, for each path point on the dynamic spraying trajectory, based on its corresponding grid region attribute label.

[0065] For example, in the specific implementation process of generating real-time spraying parameters, the system first performs gradient analysis on the curvature difference between adjacent grids on the grid surface to be sprayed to obtain a curvature gradient distribution map. Assume that adjacent grids on the grid surface to be sprayed... With grid The representative curvature values ​​were calculated as follows: and Substitute the values ​​into the curvature gradient calculation formula and perform the calculation. An approximate value of the curvature gradient is obtained through numerical subtraction. The system identifies regional attributes based on the curvature gradient distribution map, setting the curvature gradient threshold to 0.05. Since the currently calculated... Below this threshold, the system identifies the area as a flat region and generates a corresponding region attribute label. The system then combines the region attribute label with the coating rheological properties to perform parameter matching, retrieving the corresponding process parameter combinations for the flat region from a pre-set process database. In this calculation, the system ultimately matches and generates real-time spraying parameters including an atomization pressure of 0.4 MPa, a coating flow rate of 200 ml / min, and a spraying distance of 200 mm, thereby achieving precise quantitative control of the spraying process parameters based on changes in the bottle surface morphology.

[0066] S5. The adaptive spraying execution module is used to control the six-degree-of-freedom spraying robot arm and the intelligent spray gun to perform adaptive spraying operations according to the dynamic spraying trajectory and the real-time spraying parameters, so as to obtain a liquor bottle with a completed surface coating.

[0067] Furthermore, the adaptive spraying execution module includes: a spraying trajectory analysis unit, a spray gun parameter control unit, and a continuous surface coating unit; wherein, The spraying trajectory analysis unit is connected to the spraying parameter optimization unit. It is used to extract the spatial coordinates and attitude angles of each trajectory node from the dynamic spraying trajectory analysis, control the six-degree-of-freedom spraying robot arm to move to the corresponding node, and obtain the current spraying node. The spray gun parameter control unit is connected to the spray trajectory analysis unit and is used to adjust the atomization pressure and paint flow of the intelligent spray gun according to the real-time spray parameters corresponding to the current spray node to generate a dynamic atomization cone. The continuous surface coating unit is connected to the spray gun parameter control unit and is used to continuously cover the surface of the liquor bottle with the dynamic atomizing cone to obtain a liquor bottle with a completed surface coating.

[0068] Specifically, the spraying trajectory parsing unit receives the dynamic spraying trajectory and real-time spraying parameter data packets output by the trajectory planning and parameter optimization module. As the interface to the robot control system, this unit's core task is to transform abstract trajectory data into motion commands that the robot can understand. It reads the trajectory nodes in the dynamic spraying trajectory one by one in sequence. Each node contains six-dimensional pose information, i.e., three spatial coordinates. The unit sends this six-dimensional pose information to the controller of the six-DOF painting robot. The controller calculates the rotation angles required to drive the six joints of the robot to the specified pose using inverse kinematics, and then drives the motors of each joint to move precisely. When the end effector of the robot reaches a node in the trajectory sequence, that node becomes the current painting node.

[0069] As the robotic arm moves to the current spraying node, the spray gun parameter control unit operates synchronously. This unit reads real-time spraying parameters corresponding to the current spraying node from the data packet, including atomization pressure and paint flow rate. It converts these numerical parameters into control signals for the actuators inside the intelligent spray gun, such as adjusting the compressed air pressure entering the spray gun via a proportional valve and controlling the paint flow rate delivered to the spray gun via a precision metering pump. The atomization pressure is typically adjusted between 0.2 and 0.5 MPa, and the paint flow rate is controlled between 50 and 300 ml per minute as needed. Through real-time, dynamic adjustment of these parameters, the intelligent spray gun can generate a dynamic atomization cone whose shape, density, and range vary with the trajectory, ensuring optimal paint deposition on bottle surfaces with different curvatures. This process can be represented by the following functional relationship: , in, This represents the dynamic characteristics of the generated atomizing cone, including the cone's shape, size, and particle density. The atomization pressure, representing the real-time spraying parameters corresponding to the current spraying node, is generated in real time by the spraying parameter optimization unit based on the curvature gradient of the bottle surface, regional attributes, and the rheological properties of the coating. The coating flow rate, which originates from the atomization pressure, is generated synchronously and in real-time by the spraying parameter optimization unit. The coefficients and bias terms of the nonlinear mapping function are obtained through offline spray gun calibration and least squares fitting: Within a preset process range, multiple different combinations of atomization pressure and coating flow rate are selected, and the corresponding atomization cone angle, particle size distribution, and coverage width are measured using an atomization detection device to obtain an atomization characteristic calibration dataset; the calibration dataset is substituted into the nonlinear mapping model to construct an overdetermined system of equations; the least squares method is used to solve this system of equations to obtain the coefficients and bias terms that minimize the fitting error; these coefficients are pre-stored in the system for real-time calculation of dynamic atomization cone characteristics during online operations.

[0070] The continuous surface coating unit integrates the movement of the robotic arm with the dynamic atomization of the spray gun to complete the final coating operation. As the six-degree-of-freedom robotic arm moves smoothly and continuously along the dynamic spraying trajectory, the spray gun parameter control unit also adjusts the spraying parameters synchronously. This high degree of coordination between the two actions allows the dynamic atomization cone to continuously and evenly cover the surface of the liquor bottle below the moving path. The movement speed of the robotic arm, the density of trajectory points, and the spraying parameters together determine the thickness and uniformity of the final coating. The system ensures that the coating thickness is within the design requirements and that there are no runs or missed areas by precisely controlling the dwell time and spraying parameters of each spraying node. The entire process is repeated until the robotic arm has completed all trajectory nodes, achieving continuous coverage of the entire surface of the liquor bottle.

[0071] For example, in the specific implementation process of the adaptive spraying execution module, the spraying trajectory parsing unit receives the dynamic spraying trajectory and real-time spraying parameter data packet output by the trajectory planning and parameter optimization module. This unit, acting as the interface of the robot control system, parses and extracts the spatial coordinates and attitude angles of each trajectory node, and controls the six-degree-of-freedom spraying robot arm to move to the corresponding node. Assume the spatial coordinates of the current spraying node are 50 mm, 60 mm, and 300 mm. Simultaneously with the robot arm's movement, the spray gun parameter control unit adjusts the atomization pressure and paint flow rate of the intelligent spray gun according to the real-time spraying parameters corresponding to the current spraying node, generating a dynamic atomization cone. The system applies a nonlinear mapping function to perform real-time calculations on the characteristics of the dynamic atomization cone, assuming the atomization pressure corresponding to the current spraying node... The paint flow rate is 0.4 MPa. The speed is 200 ml / min. The calibration coefficients obtained through least squares fitting are set. It is 0.5. It is 0.1. It is 0.01. It is 0.002. The bias term is 0.005. The value is 1.2. Substitute this value into the formula and perform the calculation. The result was obtained through numerical calculation. This numerical value quantifies the morphology and particle density of the current dynamic atomizing cone. The continuous surface coating unit uses this dynamic atomizing cone to continuously cover the surface of the liquor bottle. As the robotic arm moves smoothly along the trajectory, the system ensures that the coating thickness is within the design requirements and there is no dripping, ultimately resulting in a liquor bottle with a completed surface coating.

[0072] Furthermore, after continuously covering the surface of the liquor bottle with the dynamic atomizing cone to obtain a liquor bottle with a completed surface coating, the process further includes: The free paint mist and volatile organic compounds generated during the adaptive spraying process are collected and separated into liquid and gaseous states to obtain liquid paint mist and gaseous volatiles. The liquid paint mist is subjected to flocculation and sedimentation treatment to obtain recycled paint residue, and the gaseous volatiles are subjected to catalytic oxidation and decomposition to obtain emission gas that meets the standards. The coated liquor bottle is then baked with hot air circulation to obtain the finished liquor bottle.

[0073] Specifically, during and after the adaptive spraying operation, the negative pressure ventilation system in the spray booth is immediately activated. Its task is to capture and treat the exhaust gases generated during spraying. This system draws all free paint mist particles and volatile organic compounds mixed in the working area into the exhaust gas treatment unit. In this unit, the mixed gas first undergoes gas-liquid separation through a wet scrubbing tower or multi-stage baffle filter. This process utilizes the principles of physical collision and interception to capture larger liquid paint mist particles in the airflow into the circulating liquid, forming wastewater containing paint mist, while the molecular-state gaseous volatiles continue to move forward with the airflow.

[0074] The two phases of pollutants separated are treated separately. For wastewater containing liquid paint mist, a flocculant, such as polyaluminum chloride or polyacrylamide, is added to the system. The flocculant disrupts the suspension stability of paint mist particles in water, causing them to aggregate into larger particles, i.e., flocs, which then settle in a sedimentation tank by gravity, forming reclaimed paint sludge that can be collected and treated. For the separated gaseous volatiles, the gas stream is introduced into a catalytic oxidation decomposition unit. In this unit, the gas is first preheated to 250 to 400 degrees Celsius and then passes through a catalytic bed filled with a precious metal catalyst. Under the catalytic action of the catalyst, the volatile organic compounds react with oxygen in the air at a relatively low temperature, decomposing into harmless carbon dioxide and water, ultimately forming compliant emission gases that can be directly discharged.

[0075] Meanwhile, the coated liquor bottles are conveyed from the spray booth to the subsequent curing unit via a conveyor belt. The system initiates a hot air circulation baking process, placing the bottles in a precisely temperature-controlled baking oven. The oven temperature is set between 60 and 120 degrees Celsius depending on the type of coating, and hot air is forced to circulate within the oven by a high-speed fan to ensure uniform heating of the bottle surface. Maintaining this temperature for 20 to 40 minutes allows the solvent in the coating to completely evaporate, and the resin to undergo a cross-linking and curing reaction, forming a strong, wear-resistant final coating with excellent adhesion. At this point, the liquor bottle becomes a finished liquor bottle ready for the next packaging process.

[0076] For example, regarding the post-processing after the adaptive spraying operation, after the adaptive spraying execution module continuously covers the surface of the liquor bottle with a dynamic atomizing cone and obtains a coated liquor bottle, the system initiates the environmental purification and finished product curing process. First, the paint mist treatment unit collects the free paint mist and volatile organic compounds generated during the adaptive spraying operation, and separates them into liquid paint mist and gaseous volatiles through a gas-liquid separator. The system adds a flocculant to the liquid paint mist for flocculation and sedimentation treatment, thereby obtaining recycled paint residue for waste utilization, and simultaneously sends the gaseous volatiles into a catalytic combustion chamber for catalytic oxidation and decomposition, converting them into carbon dioxide and water vapor, obtaining emission gases that meet standards. Subsequently, the finished product processing unit sends the coated liquor bottle into a baking tunnel for hot air circulation baking. During the curing process, the system applies a temperature compensation model to adjust the heating power in real time, and calculates the actual real-time temperature inside the baking tunnel by obtaining the preset target curing temperature and combining it with the temperature compensation term caused by external environmental fluctuations. Based on this, the finished product processing unit drives the electric heating component to stabilize the temperature at the calculated target value and continue baking to ensure that the coating is fully cross-linked and cured, ultimately resulting in the finished liquor bottle. This achieves full life-cycle management from spraying to green and environmentally friendly post-processing and finished product delivery.

[0077] It should be noted that the electrical connections between the various units described above do not necessarily represent direct or indirect connections. Any indirect connection method can be applied to the embodiments of the present invention as long as it achieves the purpose of the present invention. The above descriptions are merely exemplary embodiments of the present invention and should not be construed as limiting the scope of the present invention.

[0078] All equivalent changes and modifications made in accordance with the teachings of this invention are still within the scope of this invention. Those skilled in the art will readily conceive of other embodiments of this invention upon considering the specification and the disclosure of practical truth. This application is intended to cover any variations, uses, or adaptations of this invention that follow the general principles of this invention and include common knowledge or conventional techniques in the art not described herein.

Claims

1. A fully automatic precision positioning spraying system for liquor bottles, characterized in that, The system includes: a multi-dimensional morphology feature vector generation module, a posture correction and clamping parameter decision module, a standard spraying pose establishment module, a trajectory planning and parameter optimization module, and an adaptive spraying execution module; wherein... The multidimensional morphological feature vector generation module is used to acquire the surface morphological data and spatial pose data of the liquor bottle and perform feature fusion to generate a multidimensional morphological feature vector. The attitude correction and clamping parameter decision module is used to analyze and extract positioning deviation features from the multi-dimensional morphological feature vector, and generate attitude correction commands and dynamic clamping parameters. The standard spraying posture establishment module is used to drive the lifting turntable and the clamping mechanism to perform posture adjustment and clamping according to the posture correction command and dynamic clamping parameters, and establish a standard spraying posture. The trajectory planning and parameter optimization module is used to combine the standard spraying pose and the multi-dimensional morphological feature vector to perform trajectory planning and parameter mapping, and generate dynamic spraying trajectory and real-time spraying parameters. The adaptive spraying execution module is used to control the six-degree-of-freedom spraying robot arm and the intelligent spray gun to perform adaptive spraying operations according to the dynamic spraying trajectory and the real-time spraying parameters, so as to obtain a liquor bottle with a completed surface coating.

2. The fully automatic precision positioning and spraying system for liquor bottles according to claim 1, characterized in that, The multidimensional shape feature vector generation module includes: a three-dimensional model fusion unit, a shape feature encoding unit, and a shape feature vector generation unit; wherein... The three-dimensional model fusion unit is used to acquire the visual image and depth point cloud of the liquor bottle, perform image denoising and point cloud registration, and obtain a fused three-dimensional model. The shape feature encoding unit is connected to the three-dimensional model fusion unit and is used to extract the contour curvature and surface texture of the fused three-dimensional model for feature encoding to generate a shape encoding sequence. The shape feature vector generation unit is connected to the shape feature encoding unit and is used to concatenate the shape encoding sequence with the spatial pose data to generate a multidimensional shape feature vector.

3. The fully automatic precision positioning and spraying system for liquor bottles according to claim 2, characterized in that, The attitude correction and clamping parameter decision module includes: a deviation feature extraction unit, an attitude correction command generation unit, and a clamping parameter decision unit; wherein... The deviation feature extraction unit is connected to the shape feature vector generation unit and is used to compare the multi-dimensional shape feature vector with a preset standard bottle shape feature library, extract spatial offset and rotation angle, and generate positioning deviation features. The attitude correction command generation unit is connected to the deviation feature extraction unit and is used to perform inverse kinematics solution based on the positioning deviation features to generate attitude correction commands. The clamping parameter decision unit is connected to the attitude correction command generation unit and is used to calculate the stress distribution of the force-sensitive area in the multi-dimensional topographic feature vector to generate dynamic clamping parameters.

4. The fully automatic precision positioning spraying system for liquor bottles according to claim 3, characterized in that, The standard spraying pose establishment module includes: a pose correction drive unit, a flexible clamping control unit, and a pose verification and confirmation unit; wherein... The pose correction driving unit is connected to the pose correction command generation unit and is used to parse and extract the lifting displacement and rotation step size in the pose correction command, drive the lifting turntable to perform spatial displacement, and obtain the target corrected pose. The flexible clamping control unit is connected to the clamping parameter decision unit and the pose correction drive unit respectively. It is used to adjust the output torque and contact area of ​​the clamping mechanism according to the dynamic clamping parameters, so as to flexibly clamp the liquor bottle in the target correction pose and obtain a stable clamping state. The pose verification and confirmation unit is connected to the flexible clamping control unit and is used to perform closed-loop compensation verification on the real-time pose data under the stable clamping state to establish a standard spraying pose.

5. The fully automatic precision positioning and spraying system for liquor bottles according to claim 4, characterized in that, The trajectory planning and parameter optimization module includes: a surface mesh mapping unit, a dynamic spraying trajectory generation unit, and a spraying parameter optimization unit; wherein... The surface mesh mapping unit is connected to the pose verification and confirmation unit and is used to map the multi-dimensional morphology feature vector to the coordinate system where the standard spraying pose is located to perform surface mesh division and obtain the mesh surface to be sprayed. The dynamic spraying trajectory generation unit is connected to the surface mesh mapping unit and is used to perform normal offset and interference checks along the normal vector of the mesh surface to be sprayed, thereby generating a dynamic spraying trajectory. The spraying parameter optimization unit is connected to the dynamic spraying trajectory generation unit and is used to perform fluid dynamic mapping along the rate of curvature change of the grid surface to be sprayed based on the dynamic spraying trajectory to generate real-time spraying parameters.

6. The fully automatic precision positioning and spraying system for liquor bottles according to claim 5, characterized in that, The adaptive spraying execution module includes: a spraying trajectory analysis unit, a spray gun parameter control unit, and a continuous surface coating unit; wherein... The spraying trajectory analysis unit is connected to the spraying parameter optimization unit. It is used to extract the spatial coordinates and attitude angles of each trajectory node from the dynamic spraying trajectory analysis, control the six-degree-of-freedom spraying robot arm to move to the corresponding node, and obtain the current spraying node. The spray gun parameter control unit is connected to the spray trajectory analysis unit and is used to adjust the atomization pressure and paint flow of the intelligent spray gun according to the real-time spray parameters corresponding to the current spray node to generate a dynamic atomization cone. The continuous surface coating unit is connected to the spray gun parameter control unit and is used to continuously cover the surface of the liquor bottle with the dynamic atomizing cone to obtain a liquor bottle with a completed surface coating.

7. The fully automatic precision positioning and spraying system for liquor bottles according to claim 5, characterized in that, The process of generating real-time spraying parameters by performing hydrodynamic mapping along the rate of curvature change of the grid surface to be sprayed based on the dynamic spraying trajectory includes: Gradient analysis is performed on the curvature difference between adjacent grids in the grid surface to be sprayed to obtain a curvature gradient distribution map; Based on the curvature gradient distribution map, identify regions with abrupt curvature changes and regions with gentle curvature, and generate region attribute labels; By combining the regional attribute tags with the preset rheological properties of the coating, real-time spraying parameters including atomization pressure, coating flow rate and spraying distance are generated.

8. The fully automatic precision positioning spraying system for liquor bottles according to claim 2, characterized in that, The process of obtaining a visual image of the liquor bottle and performing image denoising and point cloud registration to obtain a fused 3D model includes: Multi-angle two-dimensional images of the liquor bottle were acquired for edge detection and feature point extraction to obtain two-dimensional contour features. The surface of the liquor bottle is scanned to obtain the original depth point cloud, which is then filtered, smoothed, and downsampled to obtain the target depth point cloud. The two-dimensional contour features are mapped to the target depth point cloud for spatial alignment and texture mapping to obtain a fused three-dimensional model.

9. The fully automatic precision positioning and spraying system for liquor bottles according to claim 6, characterized in that, The process of continuously covering the surface of the liquor bottle with the dynamic atomizing cone to obtain a liquor bottle with a completed surface coating also includes: The free paint mist and volatile organic compounds generated during the adaptive spraying process are collected and separated into liquid and gaseous states to obtain liquid paint mist and gaseous volatiles. The liquid paint mist is subjected to flocculation and sedimentation treatment to obtain recycled paint residue, and the gaseous volatiles are subjected to catalytic oxidation and decomposition to obtain emission gas that meets the standards. The coated liquor bottle is then baked with hot air circulation to obtain the finished liquor bottle.

Images