Intelligent ink jet control system and method, robot and storage medium

Abstract

The invention relates to the technical field of ink jet control, and discloses an intelligent ink jet control system and method, a robot and a storage medium. The method comprises the following steps: firstly, acquiring surface contour data and color information of an object to be jetted through a laser ranging device, and constructing an object surface feature vector; a target ink jet area is determined through area spray adaptability evaluation; then, semantic information of ink jet content is obtained, semantic element weights are calculated, and ink jet parameter constraints including an ink droplet size threshold value, an information level presentation rule and the like are generated in combination with a target area range; generating an intelligent ink-jet scheme according to the constraint, and controlling an ink-jet device to form a target pattern; and finally, a moving path of the moving mechanical arm is generated based on a cubic spline interpolation algorithm, and the target pattern acts on the target area. The method can adapt to complex surface forms and various materials, the ink jet process is accurately controlled, patterns are accurately presented and stably attached, and the ink jet effect and the information transmission effectiveness are improved.

Description

Technical Field

[0001] The present invention relates to the field of inkjet control technology, and in particular to an intelligent inkjet control system, method, robot and storage medium. Background Art

[0002] In the practical application of inkjet printing technology, the surface morphology of the objects to be inkjetted often varies significantly, ranging from planar structures to complex three-dimensional structures such as curved and concave-convex surfaces. Traditional inkjet control methods are mostly based on preset planar printing parameters, which are difficult to adapt to the morphological characteristics of different object surfaces, resulting in problems such as deformation, blurring, and even shedding of inkjet patterns in non-planar areas. For example, when inkjet printing on a curved object, if the ink droplet size and spray angle used for planar printing are still used, it is very easy for the ink droplets to be unevenly distributed due to changes in surface curvature, affecting the integrity and clarity of the pattern.

[0003] There are significant differences in the adhesion properties of ink to objects made of different materials. The surface smoothness and porosity of materials such as metals, plastics, and ceramics vary. Traditional methods do not develop specific inkjet parameters based on the characteristics of the materials, and the ink often does not adhere firmly, cracks after drying, or falls off when wiped. In addition, the presentation of inkjet content is not sufficiently compatible with the surface area of the object. When the inkjet content contains multiple layers of semantic information, traditional control methods cannot reasonably allocate information levels according to the area range, resulting in the compression of important information or the over-presentation of secondary information, affecting the effectiveness of information transmission.

[0004] Existing inkjet systems rely heavily on manual judgment or simple area division rules when determining the target inkjet area, lacking an accurate assessment of the area's suitability for inkjet printing. Some areas may not be suitable for inkjet printing due to material properties or surface conditions, but traditional methods are unable to identify such areas. Forced inkjet printing not only wastes ink but may also damage the surface of the object. In addition, the path planning of mobile robotic arms often uses a linear interpolation algorithm. When faced with complex curved surfaces or large inkjet areas, path jamming and large positioning errors are prone to occur, making it impossible for the inkjet pattern to accurately fit the target area, reducing the consistency and stability of the inkjet effect.

[0005] With the increasing demand for personalized inkjet printing in industrial production, artwork processing and other fields, the presentation of semantic information in inkjet content is becoming increasingly important. When dealing with inkjet content containing multiple layers of semantics, traditional methods have difficulty in presenting it differently based on the importance of semantic elements, which weakens key information and affects the accuracy of information transmission. For example, in product identification inkjet printing, the importance of brand logos and explanatory text is different, but traditional methods often use the same printing parameters, which cannot highlight the core information. At the same time, the settings of parameters such as ink droplet size and pattern size lack correlation with semantic information, resulting in uneven presentation of inkjet content in different areas, making it difficult to meet diverse application needs.

[0006] Existing inkjet control methods lack close coordination between the inkjet device and the mobile robotic arm, and their positional information isn't effectively integrated into path planning. This can lead to positioning errors during the robotic arm's movement, affecting the final position of the inkjet pattern. This is especially true when working with large objects or multi-area inkjet printing tasks. Inappropriate path planning increases the robotic arm's movement time, reduces overall inkjet efficiency, and makes it difficult to meet the requirements of efficient production. Summary of the Invention

[0007] The object of the present invention is to provide an intelligent inkjet control system, method, robot and storage medium to solve the problems raised in the above background technology.

[0008] To achieve the above object, the present invention provides an intelligent inkjet control method, the method comprising:

[0009] The surface profile data and color information of the object to be inkjetted are obtained according to a preset laser distance measuring device, and used to construct a surface feature vector of the object including surface morphological features and material properties;

[0010] Performing a regional inkjet suitability evaluation on the surface feature vector of the object to obtain a suitable inkjet area distribution map including a probability distribution of inkjettable areas, and determining a target inkjet area according to a highest confidence value corresponding to the suitable inkjet area distribution map;

[0011] Acquiring semantic information corresponding to preset inkjet content, calculating semantic element weights corresponding to the semantic information, and generating inkjet parameter constraints corresponding to the inkjet content based on the area range corresponding to the target inkjet area, the semantic element weights, and the semantic information; the inkjet parameter constraints at least include an ink droplet size threshold, information hierarchy presentation rules, inkjet pattern size parameters, and material adhesion firmness parameters;

[0012] generating an intelligent inkjet scheme corresponding to the inkjet content according to the inkjet parameter constraints, so as to control a preset inkjet device to perform inkjet according to the intelligent inkjet scheme to form a target inkjet pattern;

[0013] Based on the cubic spline interpolation algorithm, a preset moving path corresponding to the mobile robotic arm is generated according to the first position information corresponding to the object to be inkjetted, the second position information of the preset inkjet device, and the area range corresponding to the target area, so as to control the mobile robotic arm according to the moving path to apply the target inkjet pattern to the target area.

[0014] Preferably, the surface profile data and color information of the object to be inkjetted are obtained by using a preset laser distance measuring device, and are used to construct a surface feature vector of the object including surface morphological features and material properties, including:

[0015] Inputting the color information into a preset recurrent neural network to extract color feature maps of multiple color channels;

[0016] Inputting the surface contour data into a preset point cloud processing network to perform surface morphology modeling on the contour data and outputting surface morphology features including spectral reflectance and material properties;

[0017] The color feature map and the surface morphology feature are fused to form a feature vector, and the fused feature vector is dimensionally compressed through normalization processing to generate a surface feature vector of the object with local morphology preservation.

[0018] Preferably, performing regional inkjet suitability evaluation on the surface feature vector of the object to obtain a suitable inkjet area distribution map including a probability distribution of inkjettable areas includes:

[0019] Constructing a deep region segmentation network; the deep region segmentation network adopts the encoding and decoding structure of the ResNet architecture, introduces a spatial attention module in the encoding stage to weight the importance of the object surface feature vector; embeds a hollow convolution layer in the decoding stage, and dynamically adjusts the receptive field to adapt to the contour characteristics of different morphological regions;

[0020] Inputting the surface feature vector of the object into a deep region segmentation network, and outputting a spraying probability map including the spraying suitability of each pixel point;

[0021] The spraying-suitable probability map is processed according to median filtering to eliminate discrete noise points, thereby generating a spraying-suitable area distribution map with regional connectivity.

[0022] Preferably, determining the target inkjet area according to the highest confidence value corresponding to the suitable inkjet area distribution map includes:

[0023] An evaluation index system is constructed based on the maximum coverage area, minimum surface roughness variance, and maximum edge clarity index corresponding to the target inkjet area;

[0024] Optimizing the inkjet-suitable area distribution map according to a preset evaluation index system and a particle swarm optimization algorithm to obtain a plurality of candidate inkjet areas;

[0025] The MOEA / D algorithm is used to perform non-dominated sorting on multiple candidate inkjet areas, and the analytic hierarchy process is used to calculate the comprehensive score of each sorted candidate inkjet area.

[0026] Among the multiple candidate inkjet areas, a continuous area with a comprehensive score greater than a critical value is obtained as the target inkjet area.

[0027] Preferably, the calculating of the semantic element weight corresponding to the semantic information to generate the inkjet parameter constraint corresponding to the inkjet content according to the area range corresponding to the target inkjet area, the semantic element weight and the semantic information includes:

[0028] Extract the semantic association graph corresponding to the semantic information based on the preset LSTM model;

[0029] Calculating the importance score of each semantic node in the semantic association graph as the semantic element weight based on a preset graph neural network;

[0030] The regional range parameters corresponding to the regional range are converted into nozzle arrangement constraints, and the semantic element weights are mapped into weighted coefficients of information hierarchy presentation rules; the regional range parameters include one or more of geometric size parameters, morphological feature parameters, edge feature parameters, material characteristic parameters, spatial position and direction parameters, topological parameters, and stability parameters;

[0031] The inkjet parameter constraints corresponding to the inkjet content are obtained according to the nozzle arrangement constraints and weighting coefficients based on a dynamic programming method.

[0032] Preferably, generating the intelligent inkjet solution corresponding to the inkjet content according to the inkjet parameter constraints includes:

[0033] Construct an intelligent inkjet architecture based on a reinforcement learning policy network; the policy network of the reinforcement learning policy network adopts a multi-scale feature fusion network with a multi-head attention mechanism, and the value network of the reinforcement learning policy network introduces a simulation rendering module to simulate the actual inkjet effect; the policy network parameters of the reinforcement learning policy network are optimized based on policy gradient, and the corresponding reward function of the reinforcement learning is generated according to information entropy, visual coordination and material matching;

[0034] Generating a style parameter vector according to preset visual specifications and inkjet parameter constraints, and inputting the style parameter vector into a feature space corresponding to the intelligent inkjet architecture;

[0035] The inkjet content is input into the smart inkjet architecture to generate the smart inkjet solution.

[0036] Preferably, based on a cubic spline interpolation algorithm, a preset moving path corresponding to the mobile robot arm is generated according to the first position information corresponding to the object to be inkjetted, the second position information of the preset inkjet device, and the area range corresponding to the target area, including:

[0037] Constructing a kinematic model based on Cartesian space for converting the first position information and the second position information into a posture transformation matrix in a preset space;

[0038] Performing trajectory planning based on the kinematic model and the posture transformation matrix to generate an initial movement path; wherein a velocity smoothing constraint of cubic spline interpolation is introduced into the configuration space corresponding to the initial movement path;

[0039] Calculating collision detection information of the initial movement path according to a preset dynamic simulation, adjusting the initial movement path according to the collision detection information, and generating the movement path; wherein the collision detection information is calculated based on the collision probability, obstacle position, and path energy consumption corresponding to the initial movement path;

[0040] The controlling the mobile mechanical arm to apply the target inkjet pattern to the target area according to the moving path includes:

[0041] Acquiring pressure feedback data of an end effector corresponding to the mobile robotic arm, and adjusting motion parameters of the mobile robotic arm according to the pressure feedback data;

[0042] The inkjet process of the mobile robotic arm is monitored in real time according to a preset real-time image acquisition device. When a position deviation and / or surface undulation change of the object to be inkjetted is detected, the moving path is dynamically adjusted to control the mobile robotic arm to complete the inkjet process according to the dynamically adjusted motion parameters and moving path.

[0043] Preferably, the present invention further includes an intelligent inkjet control system for implementing the intelligent inkjet control method as described above, the system comprising:

[0044] A feature construction module is used to obtain surface profile data and color information of the object to be inkjetted based on a preset laser ranging device, and to construct a surface feature vector of the object containing surface morphological features and material properties;

[0045] a region selection module, configured to perform a regional inkjet suitability assessment on the object surface feature vector, obtain a suitable inkjet region distribution map including a probability distribution of inkjettable regions, and determine a target inkjet region based on a highest confidence value corresponding to the suitable inkjet region distribution map;

[0046] a parameter generation module, configured to obtain semantic information corresponding to a preset inkjet content, calculate semantic element weights corresponding to the semantic information, and generate inkjet parameter constraints corresponding to the inkjet content based on the area range corresponding to the target inkjet area, the semantic element weights, and the semantic information; the inkjet parameter constraints at least include an ink droplet size threshold, information hierarchy presentation rules, inkjet pattern size parameters, and material adhesion strength parameters;

[0047] a scheme formulation module, configured to generate an intelligent inkjet scheme corresponding to the inkjet content according to the inkjet parameter constraints, so as to control a preset inkjet device to perform inkjet according to the intelligent inkjet scheme to form a target inkjet pattern;

[0048] The inkjet execution module is used to generate a preset moving path corresponding to the mobile robotic arm based on the cubic spline interpolation algorithm according to the first position information corresponding to the object to be inkjetted, the second position information of the preset inkjet device, and the area range corresponding to the target area, so as to control the mobile robotic arm according to the moving path to apply the target inkjet pattern to the target area.

[0049] Preferably, the present invention further includes an intelligent inkjet control robot for executing the intelligent inkjet control method as described above, wherein the intelligent inkjet control robot includes:

[0050] A laser distance measuring device for obtaining surface profile data and color information of the object to be inkjetted;

[0051] An inkjet device for performing an inkjet operation;

[0052] A mobile robotic arm is used to apply the target inkjet pattern to an object to be inkjetted;

[0053] A control device is used to construct an object surface feature vector including surface morphological features and material properties based on the surface contour data and color information; perform regional inkjet suitability evaluation on the object surface feature vector, obtain a suitable inkjet area distribution map including a probability distribution of inkjettable areas, and determine a target inkjet area according to a maximum confidence value corresponding to the suitable inkjet area distribution map; obtain semantic information corresponding to a preset inkjet content, calculate a semantic element weight corresponding to the semantic information, and generate an inkjet parameter constraint corresponding to the inkjet content according to the area range, semantic element weight, and semantic information corresponding to the target inkjet area; the inkjet parameter constraint At least it includes an ink drop size threshold, information level presentation rules, inkjet pattern size parameters and material adhesion firmness parameters; based on the inkjet parameter constraints, an intelligent inkjet scheme corresponding to the inkjet content is generated, so as to control a preset inkjet device to spray ink according to the intelligent inkjet scheme to form a target inkjet pattern; based on a cubic spline interpolation algorithm, a moving path corresponding to a preset mobile robotic arm is generated according to first position information corresponding to the object to be inkjetted, second position information of a preset inkjet device and an area range corresponding to a target area, so as to control the mobile robotic arm according to the moving path to apply the target inkjet pattern to the target area.

[0054] Preferably, the present invention also includes a computer-readable storage medium, which includes a stored computer program, wherein when the device where the computer-readable storage medium is located executes the computer program, the intelligent inkjet control method as described above is implemented.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] Using a laser rangefinder to capture the surface profile and color information of the object to be inkjetted, a surface feature vector is constructed, encompassing both surface morphology and material properties. This accurately captures subtle differences in the object's surface, providing detailed baseline data for subsequent regional inkjet suitability assessments. Regional inkjet suitability assessments based on this vector determine the target inkjet area based on the probability distribution of inkjet-compatible areas, avoiding inkjet operations in unsuitable areas and reducing ink waste and unnecessary damage to the object's surface.

[0057] When processing inkjet content, this method extracts semantic information and calculates the weight of semantic elements, and generates inkjet parameter constraints based on the range of the target inkjet area, so that the inkjet parameters can match the semantic importance of the content. For example, for core information with a higher semantic weight, more reasonable ink droplet size and adhesion parameters can be set to ensure that it is clearly presented on the surface of the object and not easy to fall off; for secondary information, the pattern size can be adjusted according to the area range to ensure the integrity of the information while avoiding taking up too much space. This method of generating parameter constraints allows inkjet content to be presented in an appropriate form in different areas and materials, improving the effectiveness of information transmission.

[0058] The generation of an intelligent inkjet solution based on these parameter constraints makes the operation of the inkjet device more targeted. The mobile arm path generated using a cubic spline interpolation algorithm fully considers the positional relationship between the object to be inkjetted and the inkjet device, as well as the scope of the target area. The path is smooth and the positioning is precise, reducing errors during the movement of the robot arm. This process achieves the coordinated cooperation between the inkjet device and the mobile robot arm, ensuring that the target inkjet pattern is accurately applied to the target area, even on complex surfaces such as curved and concave surfaces, and maintaining the integrity and consistency of the pattern.

[0059] Constraints in this method, such as droplet size thresholds and material adhesion parameters, can adapt to the surface characteristics of different materials. For smooth materials, parameters can be adjusted to enhance ink adhesion; for porous materials, droplet size can be controlled to prevent excessive penetration, resulting in a stable inkjet pattern on a variety of materials. Furthermore, the application of information hierarchy presentation rules allows for the orderly display of multiple layers of semantic information within a limited area, avoiding information clutter and improving the overall aesthetics and readability of the pattern. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 This is a working principle diagram of the intelligent inkjet control method of the present invention;

[0061] Figure 2 Flowchart for constructing surface feature vectors of objects;

[0062] Figure 3 Flowchart for regional spray suitability assessment;

[0063] Figure 4 A flow chart for determining a target inkjet area;

[0064] Figure 5 Flowchart generated for inkjet parameter constraints. DETAILED DESCRIPTION

[0065] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0066] See also Figure 1-Figure 5 The present invention provides an intelligent inkjet control system, method, robot and storage medium, wherein the method comprises:

[0067] Step 1: Surface profile data and color information of the object to be inkjetted are acquired using a pre-set laser ranging device. This data is used to construct a surface feature vector that encompasses both surface morphological features and material properties. The laser ranging device can be a high-precision laser scanner, which radiates a laser beam to scan the object in all directions, acquiring the three-dimensional coordinate data of each surface point as surface profile data. Simultaneously, a built-in color sensor collects RGB color information of the object's surface. The acquired surface profile data and color information are pre-processed to remove noise, and then feature extraction and fusion are performed to form a feature vector that comprehensively reflects the object's surface properties.

[0068] Step 2: Perform a regional inkjet suitability assessment on the object's surface feature vectors, obtaining a suitable inkjet area distribution map containing a probability distribution of inkjet-capable areas. Target inkjet areas are then determined based on the highest confidence value corresponding to the suitable inkjet area distribution map. A specialized evaluation model is constructed to analyze different regions of the object's surface, determining their suitability for inkjet printing and presenting them as a probability distribution. The continuous region with the highest confidence value in the suitable inkjet area distribution map is selected and designated as the target inkjet area.

[0069] Step 3: Obtain semantic information corresponding to the preset inkjet content and calculate the semantic element weights corresponding to this semantic information. This generates inkjet parameter constraints corresponding to the inkjet content based on the area range corresponding to the target inkjet area, the semantic element weights, and the semantic information. These inkjet parameter constraints include at least droplet size thresholds, information hierarchy presentation rules, inkjet pattern size parameters, and material adhesion parameters. The inkjet content may include text or patterns. Semantic parsing is performed to extract key semantic information, such as the importance of text and the structural hierarchy of patterns. The weights of each semantic element are then calculated. Combined with range information such as the size and shape of the target inkjet area, parameter constraints are generated to standardize the inkjet process.

[0070] Step 4: Generate an intelligent inkjet solution corresponding to the inkjet content based on the inkjet parameter constraints. This solution controls the preset inkjet device to spray ink according to the intelligent inkjet solution, forming the target inkjet pattern. Based on the inkjet parameter constraints, the operating mode and inkjet sequence of the inkjet device are determined, and a detailed inkjet solution is developed. This solution controls the opening and closing of the inkjet device's nozzles, the amount of ink droplets ejected, and other aspects to form the desired pattern in the target inkjet area.

[0071] Step 5: Based on the first position information corresponding to the object to be inkjetted, the second position information of the preset inkjet device, and the area range corresponding to the target area, a predetermined movement path for the mobile arm is generated. The mobile arm is controlled according to the movement path to apply the target inkjet pattern to the target area. The first position information of the object to be inkjetted and the second position information of the inkjet device are obtained by a positioning device. Combined with the range of the target area, the movement path of the mobile arm is planned using the cubic spline interpolation algorithm. The mobile arm drives the inkjet device according to this path to ensure that the target inkjet pattern is accurately applied to the target area.

[0072] Example 1:

[0073] In step 1, the surface profile data and color information of the object to be inkjetted are obtained according to the preset laser ranging device, and are used to construct the object surface feature vector containing the surface morphological features and material properties. The specific process is as follows:

[0074] The laser rangefinder can utilize a multi-line scanning LiDAR (LiDAR) system. By emitting multiple laser beams toward the surface of the object to be inkjetted, the laser reflection time difference is used to calculate the three-dimensional coordinates of each sampling point on the object's surface, generating surface profile data in the form of a point cloud. Simultaneously, an integrated hyperspectral camera simultaneously collects color information from the object's surface, acquiring multi-channel color data encompassing the visible spectrum. During the acquisition process, the laser rangefinder's scanning frequency is aligned with the camera's sampling frame rate to ensure precise spatial alignment between surface profile data and color information.

[0075] The color information is input into a pre-set recurrent neural network. This recurrent neural network uses a bidirectional LSTM structure, consisting of an input layer, two bidirectional LSTM layers, and an output layer. The input layer converts the RGB channel data of the color information into a three-dimensional tensor, where the first dimension corresponds to the image height, the second dimension corresponds to the image width, and the third dimension corresponds to the three RGB color channels. The first bidirectional LSTM layer extracts features along the rows of the tensor to capture horizontal color variations; the second bidirectional LSTM layer extracts features along the columns of the tensor to capture vertical color distribution characteristics. After these two layers of processing, the output layer uses convolution operations to map the extracted features into color feature maps for multiple color channels. Each feature map corresponds to the spatial distribution pattern of a specific color frequency band.

[0076] The surface contour data is input into a preset point cloud processing network. The network includes a point cloud preprocessing module, a feature extraction module and a morphological modeling module. The point cloud preprocessing module first denoises the original point cloud data, removes discrete points that are too far from the mean through a statistical filtering algorithm, and then uses a voxel grid downsampling method to reduce the number of point clouds and maintain the overall structure of the point cloud. The feature extraction module adopts the PointNet architecture, and uses a multi-layer perceptron to perform feature calculations on the local neighborhood of each point to obtain a point feature vector containing geometric features such as normal vectors and curvature. The morphological modeling module inputs the point feature vector into the graph convolutional network, constructs the topological structure of the object surface through message passing between graph nodes, and then calculates the spectral reflectance of each area, and derives the material characteristics in combination with the geometric features, and finally outputs the surface morphological features containing spectral reflectance and material characteristics.

[0077] The color feature map and the surface morphological features are fused to form feature vectors. The fusion process consists of two stages: spatial alignment and feature splicing. In the spatial alignment stage, the mapping relationship between the pixel coordinates of the color feature map and the point cloud coordinates of the surface morphological features is calculated. The surface morphological features are then projected onto the two-dimensional plane of the color feature map using the nearest neighbor interpolation method, ensuring that the two are consistent in spatial dimensions. In the feature splicing stage, the aligned color feature map and surface morphological features are spliced together according to the channel dimension to form a high-dimensional fused feature vector.

[0078] The fused feature vectors are compressed in dimension through normalization. A principal component analysis algorithm is used to reduce the dimensionality of the high-dimensional fused feature vectors. The covariance matrix of the feature vectors is calculated, and their eigenvalues and eigenvectors are solved. The top k eigenvectors with the largest eigenvalues are selected to form a projection matrix, and the high-dimensional feature vectors are projected into a low-dimensional space. During the projection process, by retaining the local variance information of the feature vectors, the reduced eigenvectors can still reflect the local morphological details of the object surface, ultimately generating a surface feature vector that preserves local morphology. The dimension of this feature vector is determined based on the accuracy requirements of the actual application scenario and is typically controlled between 256 and 1024 dimensions to balance feature expression capabilities and computational efficiency.

[0079] Throughout the entire processing process, the parameters of the recurrent neural network and point cloud processing network are pre-trained using a large-scale annotated dataset containing surface data of objects of varying materials and shapes, along with their corresponding feature vector labels. This pre-training process utilizes a stochastic gradient descent algorithm, optimizing the cross-entropy loss function. Through multiple rounds of iterative adjustments to the network parameters, the network is able to reliably extract color and morphological features consistent with actual physical properties. The parameters of the feature fusion and normalization processes are dynamically adjusted based on the distribution characteristics of the input data, ensuring that the generated surface feature vectors have consistent scale and distribution across different scenarios.

[0080] Example 2:

[0081] In step 2, regional inkjet suitability evaluation is performed on the surface feature vector of the object to obtain a distribution map of inkjet-suitable regions including a probability distribution of inkjet-suitable regions. The specific process is as follows:

[0082] A deep region segmentation network was constructed. This network adopts the encoder-decoder structure of the ResNet architecture. The encoding part consists of multiple convolutional modules, each of which includes a convolutional layer, a batch normalization layer, and an activation function. The convolutional layer uses a 3x3 convolution kernel and convolves the input surface feature vector with a sliding window to extract local features. The batch normalization layer normalizes the convolved feature data to maintain a stable data distribution. The activation function uses the ReLU function to enhance the network's nonlinear representation capabilities. A spatial attention module is introduced during the encoding phase. This module consists of a channel attention submodule and a spatial attention submodule. The channel attention submodule converts the feature map into a channel description vector through global average pooling and then calculates the importance weight of each channel through a fully connected layer. The spatial attention submodule performs channel-wise max pooling and average pooling on the feature map. The concatenated results are then passed through a convolutional layer to generate a spatial attention weight map. The channel weights are multiplied by the spatial weights to obtain the final attention weights, which are then weighted with the input feature vector to highlight the feature information of key regions.

[0083] The decoding stage consists of multiple deconvolution modules, each of which gradually restores the spatial resolution of the feature map through deconvolution operations. A dilated convolution layer is embedded in the decoding stage. This layer places spacing regions within the convolution kernel and adjusts the dilation ratio to change the receptive field size. For areas with complex surface contours, a larger dilation ratio is used to expand the receptive field and capture more contextual information; for relatively flat surfaces, a smaller dilation ratio is used to improve the recognition accuracy of local details. Alternating dilated convolution layers with standard convolution layers enables the network to adapt to the contour characteristics of regions with different morphologies, improving the accuracy of region segmentation.

[0084] The surface feature vector of the object is input into the deep region segmentation network. The feature vector is first processed by the convolution module and spatial attention module in the encoding stage to gradually extract high-level features and compress the spatial dimension; then it enters the decoding stage, through the processing of the deconvolution module and the void convolution layer, the spatial size of the feature map is gradually restored, and the feature information of different levels in the encoding stage is integrated to enhance the richness of feature expression. The output layer of the network uses the Sigmoid activation function to convert the numerical value of the feature map into a probability value between 0 and 1, and generate a suitable inkjet probability map containing the suitability of each pixel point for inkjet printing. The larger the value of each pixel point in the probability map, the more likely it is that the location is suitable for inkjet operation.

[0085] The spray-appropriate probability map is processed using median filtering. The window size of the median filter is determined according to the resolution of the probability map, and a window of 3x3 or 5x5 is usually selected. For each pixel in the probability map, the probability values of all pixels in the window are taken, and these values are sorted by size and the middle value is selected as the new value of the pixel. In this way, isolated noise points in the probability map can be effectively removed, avoiding the influence of a single outlier on regional judgment. For pixel points in the edge area, the boundary is processed by mirror filling to ensure the integrity of the edge contour after filtering.

[0086] After median filtering, the probability distribution in the jetting probability map becomes smoother, and the probability values of adjacent pixels change more continuously, forming a well-connected jetting region distribution map. In this distribution map, regions with higher probability values form a continuous block structure, each representing a potential jetting area. The boundaries between regions are distinguished by sudden changes in probability values, resulting in clear boundary lines, facilitating the subsequent identification and selection of target jetting areas.

[0087] The training process of the deep region segmentation network utilizes a dataset containing labeled printable regions for various surface types, including metal, plastic, and wood, as well as various surface morphologies, including flat, curved, and concave / convex surfaces. During training, the Adam optimization algorithm is used to adjust network parameters, using the cross-entropy loss between the predicted printable probability map and the labeled printable region map as the optimization target. Through multiple rounds of iterative training, the network accurately learns the mapping relationship between different surface features and printability, ensuring that the generated printable region distribution map meets actual inkjet requirements. The median filter parameters are adjusted based on the performance of the validation set during training to achieve optimal noise removal and region preservation.

[0088] Example 3:

[0089] In step 2, the target inkjet area is determined according to the highest confidence value corresponding to the suitable inkjet area distribution map. The specific process is as follows:

[0090] An evaluation index system is constructed based on the maximum coverage area, minimum surface roughness variance and maximum edge clarity index corresponding to the target inkjet area. The maximum coverage area refers to the area occupied by the candidate area on the surface of the object, which is obtained by calculating the two-dimensional projection area enclosed by the boundary of the area. The curvature of the object surface must be considered when calculating the area, and the curved area must be expanded before the area accumulation is performed. The minimum surface roughness variance is obtained by calculating the average value of the sum of the squares of the deviations between the roughness values of each point in the area and the average roughness value of the area. The roughness value is converted from the morphological feature parameters in the surface feature vector of the object and reflects the flatness of the surface of the area. The maximum edge clarity index is determined by analyzing the rate of change of the grayscale values of the pixel points at the boundary of the area. The larger the rate of change, the clearer the edge. The noise interference at the boundary must be excluded during the calculation, and the edge pixels must be smoothed before the gradient calculation is performed.

[0091] The distribution map of suitable inkjet areas is optimized based on a preset evaluation index system and a particle swarm optimization algorithm to obtain multiple candidate inkjet areas. During initialization of the particle swarm optimization algorithm, the position information of each candidate area is used as the initial position of the particle, and the particle's dimensions are determined by the boundary coordinate parameters of the area. During the algorithm iteration process, each particle adjusts its flight speed and direction based on its own historical optimal position and the swarm's global optimal position. An inertia weight factor is introduced during speed updates, and the inertia weight is gradually reduced as the number of iterations increases, allowing the particle to maintain its exploration capability in the early stages and enhance its local search capability in the later stages. The particle's fitness function is composed of a weighted combination of three indicators in the evaluation index system. The weights are set based on the relative importance of each indicator in different application scenarios, and the quality of the candidate area is judged by the size of the fitness value. After a preset number of iterations, the areas corresponding to multiple particles with higher fitness values are selected as candidate inkjet areas.

[0092] The MOEA / D algorithm performs non-dominated sorting on multiple candidate inkjet regions. During the non-dominated sorting process, each candidate region is considered a solution. If solution A outperforms solution B on all evaluation metrics, solution A dominates solution B, and solution B is marked as dominated. Through multiple rounds of comparison, all candidate regions are divided into different non-dominated levels. Lower level numbers indicate better overall performance of candidate regions at that level. After the non-dominated sorting is completed, the congestion degree of the candidate regions within each level is calculated. The congestion degree reflects the density of the candidate regions in the target space. Regions with higher congestion degrees are retained to maintain candidate region diversity.

[0093] The comprehensive score corresponding to each candidate inkjet area after sorting is calculated according to the hierarchical analysis method. The hierarchical analysis method first constructs a judgment matrix. The matrix elements represent the relative importance between the two evaluation indicators. The importance level is divided into 1 to 9 levels. The higher the level, the more important the indicator is. The judgment matrix is tested for consistency. The rationality of the matrix is ensured by calculating the consistency index and consistency ratio. If the consistency ratio exceeds the preset threshold, the judgment matrix element value is readjusted. After the consistency test passes, the eigenvector corresponding to the maximum eigenvalue of the judgment matrix is calculated as the weight coefficient of each evaluation indicator. The calculation formula for the comprehensive score is:

[0094] S=α×S1+β×S2+γ×S3

[0095] Where S represents the comprehensive score of the candidate inkjet area, α, β, and γ represent the weight coefficients for maximum coverage area, minimum surface roughness variance, and maximum edge definition index, respectively. S1 represents the normalized value of maximum coverage area, S2 represents the normalized value of minimum surface roughness variance, and S3 represents the normalized value of maximum edge definition index. During normalization, the original values of each indicator are mapped to the range of 0 to 1. For maximum coverage area and maximum edge definition index, positive normalization is used, that is, the larger the indicator value, the closer the normalized value is to 1; for minimum surface roughness variance, negative normalization is used, that is, the smaller the indicator value, the closer the normalized value is to 1.

[0096] Among multiple candidate inkjet regions, a continuous region with a comprehensive score greater than a critical value is selected as the target inkjet region. The critical value is determined based on actual inkjet requirements and is set by analyzing the distribution of comprehensive scores of qualified regions in historical inkjet data. Typically, the average of the comprehensive scores of all candidate regions is taken as the initial critical value, which is then adjusted based on the region's continuity requirements. Continuous regions are determined by analyzing the connectivity of each pixel within the region, using the 8-neighborhood connectivity criterion. That is, if two adjacent pixels are both suitable inkjet regions and their comprehensive scores are greater than the critical value, they are considered to belong to the same continuous region. Boundaries are extracted for these qualified continuous regions, and the region's contours are obtained using an edge detection algorithm. These contours are simplified using polygon fitting to reduce the computational complexity of subsequent path planning. If multiple qualified continuous regions exist, the region with the highest comprehensive score is selected as the target inkjet region. If the comprehensive scores of all candidate regions are less than the critical value, the process returns to the process of re-evaluating the regional inkjet suitability, adjusting the evaluation parameters, and regenerating the suitable inkjet region distribution map until a qualified target inkjet region is determined.

[0097] Throughout the entire process, the number of iterations of the particle swarm optimization algorithm, the population size of the MOEA / D algorithm, and the judgment matrix dimensions of the hierarchical analysis method are dynamically adjusted based on the number and complexity of the candidate regions. For objects with simple surface morphology, the number of iterations and population size can be appropriately reduced to improve processing efficiency; for objects with complex surface structures, the number of iterations and population size are increased to ensure that the selected target inkjet areas meet actual inkjet requirements. During the calculation of the comprehensive score, the weight coefficient can be dynamically adjusted based on the characteristics of the inkjet content. For example, for inkjet content that requires large-area presentation, the weight coefficient of the maximum coverage area is increased; for inkjet content that requires clear edges, the weight coefficient of the edge clarity index is increased.

[0098] Example 4:

[0099] In step 3, the semantic element weights corresponding to the semantic information are calculated to generate inkjet parameter constraints corresponding to the inkjet content based on the area range corresponding to the target inkjet area, the semantic element weights, and the semantic information. The specific process is as follows:

[0100] A semantic association graph corresponding to the semantic information is extracted based on a preset LSTM model. If the inkjet content consists of a combination of text and patterns, the text is first segmented, breaking the continuous text into independent words. Part-of-speech tagging is then used to determine the grammatical properties of each word. The pattern is then contour-extracted to identify the main components of the pattern and the connections between them. The LSTM model inputs are a preprocessed sequence of text segmentations and feature vectors of pattern elements. Through recurrent connections across multiple hidden layers, the model captures dependencies between preceding and following elements in the sequence, such as the relationship between modifiers and the central word in the text, and the spatial relationship between the main and auxiliary elements in the pattern. The semantic association graph output by the model uses nodes to represent semantic elements. Node attributes include element type (text / pattern) and feature values (such as the part of speech of the text or the shape of the pattern element). Directed edges represent the type of association between elements (such as modification or inclusion), and edge weights reflect the strength of the association.

[0101] The importance score of each semantic node in the semantic association graph is calculated based on the preset graph neural network as the semantic element weight. The input of the graph neural network is the adjacency matrix and node feature matrix of the semantic association graph. The adjacency matrix records the connection relationship between nodes, and the node feature matrix contains the attribute information of each node. The network updates the node features through the graph convolution layer. The new features of each node are obtained by the weighted sum of its own features and the features of adjacent nodes. The weights are dynamically adjusted according to the correlation strength of the edges. After multi-layer graph convolution operations, the features of all nodes are aggregated into graph-level features through the global pooling layer, and then the importance score of each node is output through the fully connected layer. During the score calculation process, for text elements, the focus is on their core degree in semantic expression; for pattern elements, the focus is on their prominence in visual presentation. The final importance score is the weight of each semantic element.

[0102] The area parameters corresponding to the area range are converted into nozzle arrangement constraints, and the semantic element weights are mapped into weighted coefficients for information hierarchy presentation rules. The geometric dimensions within the area range parameters include the length, width, area, and proportional relationships of the target inkjet area. These parameters determine the number and arrangement of nozzles. For example, when the area width is large, multiple nozzles are arranged in parallel to cover the entire width. For irregularly shaped display areas, the nozzle arrangement constraints must include the nozzle steering angle range to ensure that the nozzles are always aligned toward the center of the area. For edge smoothness, the edge feature parameters determine the required inkjet accuracy when the nozzle approaches the area edge; smoother edges require higher accuracy requirements. For material properties, the surface hardness affects the distance constraint between the nozzle and the object surface. For materials with lower hardness, the nozzle distance must be increased to avoid contact damage. The spatial position and orientation parameters determine the initial positioning coordinates and working plane of the nozzle. The topological parameters, including the connection relationship with other areas, determine the transition method when the nozzle switches between areas. For stability parameters, if there is a risk of deformation in the display area, the nozzle arrangement constraints must include the pressure adjustment range. The higher the semantic element weight, the greater the corresponding weighting coefficient. In the information hierarchy presentation rule, the element with a large weighting coefficient will adopt a higher inkjet concentration or a more eye-catching color.

[0103] Based on a dynamic programming approach, the inkjet parameter constraints corresponding to the inkjet content are derived from the nozzle arrangement constraints and weighting coefficients. The dynamic programming method decomposes the inkjet process into multiple continuous inkjet segments, each corresponding to a sub-region within the target area. For each sub-region, the nozzle arrangement constraints determine the nozzle combination and inkjet range for that segment. The weighting coefficients are then used to allocate inkjet resources, such as the number of ink drops and jetting time, to each semantic element within that segment. The droplet size threshold is determined by comprehensively considering the sub-region's material properties and the weights of the semantic elements. Larger droplet sizes are used for areas with high-weight elements and good material adsorption, while smaller droplet sizes are used for areas with low-weight elements and poor material adsorption. The information hierarchy presentation rule sorts the elements according to the weighting coefficients and determines the inkjet order for each element, prioritizing high-weight elements to ensure integrity. The inkjet pattern size parameters are determined based on the proportional relationship between the region's geometric dimensions and the semantic elements, ensuring that the pattern maintains an appropriate scale within the region. The material adhesion strength parameter references the surface roughness parameter within the material properties. Areas with higher roughness require a higher ink droplet adhesion ratio to enhance adhesion. Through multi-stage decision-making of dynamic programming, the inkjet parameter constraints containing all the above parameters are finally generated to ensure that the inkjet process is compatible with the characteristics of the target area and the semantic information requirements.

[0104] Example 5:

[0105] In step 4, an intelligent inkjet solution corresponding to the inkjet content is generated according to the inkjet parameter constraints. The specific process is as follows:

[0106] An intelligent inkjet architecture based on a reinforcement learning policy network is constructed. The policy network of the reinforcement learning policy network utilizes a multi-scale feature fusion network with a multi-head attention mechanism. This multi-head attention mechanism consists of multiple parallel attention heads, each focusing on a different feature dimension within the inkjet content and inkjet parameter constraints. For example, some attention heads focus on the matching relationship between ink droplet size thresholds and material properties, while others focus on the adaptability of information hierarchy presentation rules and regional scope. The multi-scale feature fusion network extracts feature information at different scales using convolution kernels of different sizes. Small convolution kernels capture local details, while large convolution kernels extract global structural features. These features are fused at multiple levels to form a comprehensive feature representation that fully reflects the inkjet scene. The value network of the reinforcement learning policy network incorporates a simulation rendering module. This module uses a physics engine to simulate the diffusion and adhesion process of ink droplets after injection. Based on the inkjet parameter constraints, it generates a virtual inkjet effect image. The image includes the distribution density of the ink droplets, the color overlay effect, and the adhesion state of the ink droplets to the object surface. The policy network parameters of the reinforcement learning policy network are optimized based on policy gradients. During the optimization process, a corresponding reward function for reinforcement learning is generated based on information entropy, visual coordination, and material matching. Information entropy is calculated by calculating the uncertainty of the ink droplet size distribution in the inkjet solution; the more uniform the distribution, the greater the information entropy. Visual coordination is determined by analyzing the color harmony and pattern integrity of the virtual inkjet effect image. Material matching is determined by comparing the degree of fit between the ink droplet adhesion parameters and the material characteristic parameters.

[0107] A style parameter vector is generated based on preset visual specifications and inkjet parameter constraints. The visual specifications cover color matching ranges, font edge processing methods, pattern line thickness standards, and other aspects. These are then mapped to ink droplet size thresholds and inkjet pattern size parameters in the inkjet parameter constraints to convert them into numerical style parameters. For example, the color matching range is converted to a numerical interval in the RGB color space, and the font edge processing method is converted to a numerical indicator of edge blur. Once the style parameter vector is formed, it is input into the feature space corresponding to the intelligent inkjet architecture, enabling the architecture to understand and comply with the preset visual requirements.

[0108] The inkjet content is input into the intelligent inkjet architecture, which uses a policy network to comprehensively analyze the inkjet content, style parameter vector, and inkjet parameter constraints to generate a preliminary inkjet solution. This solution includes parameters such as the nozzle opening sequence, ink droplet ejection frequency, and movement speed. The value network evaluates this preliminary solution through a simulation rendering module and outputs a benefit function value. The policy network adjusts parameters based on this value, and iterates repeatedly until a stable benefit function value is generated.

[0109] In step 5, based on the cubic spline interpolation algorithm, a preset moving path corresponding to the mobile robot arm is generated according to the first position information corresponding to the object to be inkjetted, the preset second position information of the inkjet device, and the area range corresponding to the target area. The specific process is as follows:

[0110] A Cartesian kinematic model is constructed. This model, based on the Cartesian coordinate system, incorporates parameters such as the length, connection relationships, and range of motion of each joint of the mobile robot. The first position information of the object to be inkjetted is converted into the object's coordinate origin and attitude matrix in Cartesian space. The second position information of the inkjet device is converted into the initial coordinates and attitude matrix of the end effector. The kinematic model is used to calculate the position transformation matrix between the two. The matrix contains translational and rotational components, representing the changing relationship between position and attitude, respectively.

[0111] Trajectory planning is performed based on the kinematic model and the pose transformation matrix to generate an initial movement path. The initial path consists of multiple path points, and the trajectories between adjacent path points are connected by a cubic spline curve. The first and second derivatives of the curve are continuous at the path points, ensuring the smoothness of the mobile manipulator's motion. Velocity smoothing constraints using cubic spline interpolation are introduced into the configuration space corresponding to the initial movement path. These constraints include a maximum velocity threshold, an acceleration threshold, and a jerk threshold. By limiting the rate of change of velocity at the path points, abrupt stops and starts of the manipulator during movement are avoided.

[0112] Collision detection information for the initial movement path is calculated based on a preset dynamic simulation. This dynamic simulation simulates the forces acting on each joint of the mobile manipulator as it moves along the initial path, as well as the relative position of the movement trajectory and the surrounding environment. The collision probability is determined by calculating the spatial overlap between the manipulator model and the obstacle model; the higher the overlap, the greater the collision probability. The obstacle position is determined using environmental point cloud data acquired in real time by a laser rangefinder. The path energy consumption is calculated by multiplying the driving torque and the movement angle of each joint during movement. The initial movement path is adjusted based on the collision detection information. If the collision probability exceeds the set value, obstacles are avoided by adding path points or changing the path direction. If the path energy consumption is too high, the path point distribution is optimized to shorten the movement distance to reduce energy consumption, ultimately generating an adjusted movement path.

[0113] According to the moving path, the mobile robot is controlled to apply the target inkjet pattern to the target area. The specific process is as follows:

[0114] The pressure feedback data of the end effector corresponding to the mobile robotic arm is obtained. The pressure sensor installed on the end effector collects the pressure value when in contact with the object surface in real time. The motion parameters of the mobile robotic arm are adjusted according to the change of the pressure value. When the pressure value exceeds the set range, the feed speed of the end effector is reduced; when the pressure value is lower than the set range, the feed speed is increased to keep the pressure in the appropriate range.

[0115] The inkjet process of the mobile robot is monitored in real time by a preset real-time image acquisition device. This device uses a high-speed industrial camera to capture images of the inkjet area at a fixed frame rate. Image recognition algorithms analyze the deviation between the position of the object in the image and the preset first position information, as well as any surface fluctuations. When position deviations are detected, the pose transformation matrix is recalculated, and the movement path is dynamically corrected. When surface fluctuations are detected, the height of the end effector is adjusted based on the amplitude of the fluctuations to ensure a stable inkjet distance. The mobile robot is then controlled to complete the inkjet process based on the dynamically adjusted motion parameters and movement path.

[0116] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0117] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent inkjet control method, characterized in that: include: The surface profile data and color information of the object to be inkjetted are obtained according to a preset laser distance measuring device, and used to construct a surface feature vector of the object including surface morphological features and material properties; Performing a regional inkjet suitability evaluation on the surface feature vector of the object to obtain a suitable inkjet area distribution map including a probability distribution of inkjettable areas, and determining a target inkjet area according to a highest confidence value corresponding to the suitable inkjet area distribution map; Acquiring semantic information corresponding to preset inkjet content, calculating semantic element weights corresponding to the semantic information, and generating inkjet parameter constraints corresponding to the inkjet content based on the area range corresponding to the target inkjet area, the semantic element weights, and the semantic information; the inkjet parameter constraints at least include an ink droplet size threshold, information hierarchy presentation rules, inkjet pattern size parameters, and material adhesion firmness parameters; generating an intelligent inkjet scheme corresponding to the inkjet content according to the inkjet parameter constraints, so as to control a preset inkjet device to perform inkjet according to the intelligent inkjet scheme to form a target inkjet pattern; Based on the cubic spline interpolation algorithm, a preset moving path corresponding to the mobile robotic arm is generated according to the first position information corresponding to the object to be inkjetted, the second position information of the preset inkjet device, and the area range corresponding to the target area, so as to control the mobile robotic arm according to the moving path to apply the target inkjet pattern to the target area.

2. The intelligent inkjet control method according to claim 1, characterized in that: The surface profile data and color information of the object to be inkjetted are obtained according to the preset laser distance measuring device, and are used to construct the object surface feature vector containing surface morphological features and material properties, including: Inputting the color information into a preset recurrent neural network to extract color feature maps of multiple color channels; Inputting the surface contour data into a preset point cloud processing network to perform surface morphology modeling on the contour data and outputting surface morphology features including spectral reflectance and material properties; The color feature map and the surface morphology feature are fused to form a feature vector, and the fused feature vector is dimensionally compressed through normalization processing to generate a surface feature vector of the object with local morphology preservation.

3. The intelligent inkjet control method according to claim 1, characterized in that: The performing regional inkjet suitability evaluation on the surface feature vector of the object to obtain a distribution map of inkjet-suitable areas including a probability distribution of inkjet-suitable areas includes: Constructing a deep region segmentation network; the deep region segmentation network adopts the encoding and decoding structure of the ResNet architecture, introduces a spatial attention module in the encoding stage to weight the importance of the object surface feature vector; embeds a hollow convolution layer in the decoding stage, and dynamically adjusts the receptive field to adapt to the contour characteristics of different morphological regions; Inputting the surface feature vector of the object into a deep region segmentation network, and outputting a spraying probability map including the spraying suitability of each pixel point; The spraying-suitable probability map is processed according to median filtering to eliminate discrete noise points, thereby generating a spraying-suitable area distribution map with regional connectivity.

4. The intelligent inkjet control method according to claim 1, characterized in that: The step of determining the target inkjet area according to the highest confidence value corresponding to the suitable inkjet area distribution map includes: An evaluation index system is constructed based on the maximum coverage area, minimum surface roughness variance, and maximum edge clarity index corresponding to the target inkjet area; Optimizing the inkjet-suitable area distribution map according to a preset evaluation index system and a particle swarm optimization algorithm to obtain a plurality of candidate inkjet areas; The MOEA / D algorithm is used to perform non-dominated sorting on multiple candidate inkjet areas, and the analytic hierarchy process is used to calculate the comprehensive score of each sorted candidate inkjet area. Among the multiple candidate inkjet areas, a continuous area with a comprehensive score greater than a critical value is obtained as the target inkjet area.

5. The intelligent inkjet control method according to claim 1, characterized in that: The calculating of the semantic element weight corresponding to the semantic information to generate the inkjet parameter constraint corresponding to the inkjet content according to the area range corresponding to the target inkjet area, the semantic element weight, and the semantic information includes: Extract the semantic association graph corresponding to the semantic information based on the preset LSTM model; Calculating the importance score of each semantic node in the semantic association graph as the semantic element weight based on a preset graph neural network; The regional range parameters corresponding to the regional range are converted into nozzle arrangement constraints, and the semantic element weights are mapped into weighted coefficients of information hierarchy presentation rules; the regional range parameters include one or more of geometric size parameters, morphological feature parameters, edge feature parameters, material characteristic parameters, spatial position and direction parameters, topological parameters, and stability parameters; The inkjet parameter constraints corresponding to the inkjet content are obtained according to the nozzle arrangement constraints and weighting coefficients based on a dynamic programming method.

6. The intelligent inkjet control method according to claim 1, characterized in that: Generating the intelligent inkjet solution corresponding to the inkjet content according to the inkjet parameter constraints includes: Construct an intelligent inkjet architecture based on a reinforcement learning policy network; the policy network of the reinforcement learning policy network adopts a multi-scale feature fusion network with a multi-head attention mechanism, and the value network of the reinforcement learning policy network introduces a simulation rendering module to simulate the actual inkjet effect; the policy network parameters of the reinforcement learning policy network are optimized based on policy gradient, and the corresponding reward function of the reinforcement learning is generated according to information entropy, visual coordination and material matching; Generating a style parameter vector according to preset visual specifications and inkjet parameter constraints, and inputting the style parameter vector into a feature space corresponding to the intelligent inkjet architecture; The inkjet content is input into the smart inkjet architecture to generate the smart inkjet solution.

7. The intelligent inkjet control method according to claim 1, characterized in that: Based on the cubic spline interpolation algorithm, a preset moving path corresponding to the mobile robot arm is generated according to the first position information corresponding to the object to be inkjetted, the second position information of the preset inkjet device, and the area range corresponding to the target area, including: Constructing a kinematic model based on Cartesian space for converting the first position information and the second position information into a posture transformation matrix in a preset space; Performing trajectory planning based on the kinematic model and the posture transformation matrix to generate an initial movement path; wherein a velocity smoothing constraint of cubic spline interpolation is introduced into the configuration space corresponding to the initial movement path; Calculating collision detection information of the initial movement path according to a preset dynamic simulation, adjusting the initial movement path according to the collision detection information, and generating the movement path; wherein the collision detection information is calculated based on the collision probability, obstacle position, and path energy consumption corresponding to the initial movement path; The controlling the mobile mechanical arm to apply the target inkjet pattern to the target area according to the moving path includes: Acquiring pressure feedback data of an end effector corresponding to the mobile robotic arm, and adjusting motion parameters of the mobile robotic arm according to the pressure feedback data; The inkjet process of the mobile robotic arm is monitored in real time according to a preset real-time image acquisition device. When a position deviation and / or surface undulation change of the object to be inkjetted is detected, the moving path is dynamically adjusted to control the mobile robotic arm to complete the inkjet process according to the dynamically adjusted motion parameters and moving path.

8. An intelligent inkjet control system, used to implement the intelligent inkjet control method according to any one of claims 1 to 7, characterized in that: include: A feature construction module is used to obtain surface profile data and color information of the object to be inkjetted based on a preset laser ranging device, and to construct a surface feature vector of the object containing surface morphological features and material properties; a region selection module, configured to perform a regional inkjet suitability assessment on the object surface feature vector, obtain a suitable inkjet region distribution map including a probability distribution of inkjettable regions, and determine a target inkjet region based on a highest confidence value corresponding to the suitable inkjet region distribution map; a parameter generation module, configured to obtain semantic information corresponding to a preset inkjet content, calculate semantic element weights corresponding to the semantic information, and generate inkjet parameter constraints corresponding to the inkjet content based on the area range corresponding to the target inkjet area, the semantic element weights, and the semantic information; the inkjet parameter constraints at least include an ink droplet size threshold, information hierarchy presentation rules, inkjet pattern size parameters, and material adhesion strength parameters; a scheme formulation module, configured to generate an intelligent inkjet scheme corresponding to the inkjet content according to the inkjet parameter constraints, so as to control a preset inkjet device to perform inkjet according to the intelligent inkjet scheme to form a target inkjet pattern; The inkjet execution module is used to generate a preset moving path corresponding to the mobile robotic arm based on the cubic spline interpolation algorithm according to the first position information corresponding to the object to be inkjetted, the second position information of the preset inkjet device, and the area range corresponding to the target area, so as to control the mobile robotic arm according to the moving path to apply the target inkjet pattern to the target area.

9. An intelligent inkjet control robot, used to execute the intelligent inkjet control method according to any one of claims 1 to 7, characterized in that: include: A laser distance measuring device for obtaining surface profile data and color information of the object to be inkjetted; An inkjet device for performing an inkjet operation; A mobile robotic arm is used to apply the target inkjet pattern to an object to be inkjetted; A control device is configured to construct a surface feature vector of an object including surface morphological features and material properties based on the surface profile data and color information; perform a regional inkjet suitability assessment on the surface feature vector of the object to obtain a suitable inkjet area distribution map including a probability distribution of inkjettable areas, and determine a target inkjet area based on a highest confidence value corresponding to the suitable inkjet area distribution map; Acquiring semantic information corresponding to a preset inkjet content, calculating semantic element weights corresponding to the semantic information, and generating inkjet parameter constraints corresponding to the inkjet content based on the area range corresponding to the target inkjet area, the semantic element weights, and the semantic information; the inkjet parameter constraints at least include an ink droplet size threshold, information hierarchy presentation rules, inkjet pattern size parameters, and material adhesion firmness parameters; generating an intelligent inkjet solution corresponding to the inkjet content based on the inkjet parameter constraints, and controlling a preset inkjet device to perform inkjet according to the intelligent inkjet solution to form a target inkjet pattern; Based on the cubic spline interpolation algorithm, a preset moving path corresponding to the mobile robotic arm is generated according to the first position information corresponding to the object to be inkjetted, the second position information of the preset inkjet device, and the area range corresponding to the target area, so as to control the mobile robotic arm according to the moving path to apply the target inkjet pattern to the target area.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a stored computer program, wherein when the device where the computer-readable storage medium is located executes the computer program, the intelligent inkjet control method according to any one of claims 1 to 7 is implemented.

Images