Machine learning-implemented inkjet processing for generation of three-dimensional relief on tiles

1 . A machine learning-based method for forming a three-dimensional (3D) relief on a substrate, the method comprising:

receiving, by a computer system, information defining a relief pattern for forming the 3D relief using a digital inkjet printer;

extracting, from the information, a feature vector comprising one or more features describing the 3D relief;

generating, using a machine learning model, control commands based on the feature vector,

wherein the machine learning model is trained to generate the control commands to:

configure the digital inkjet printer to apply binder ink to a first region of a surface of the substrate, wherein the applied binder ink is configured to form a protective layer over the first region of the surface of the substrate;

configure the digital inkjet printer to apply solvent ink to the surface of the substrate after applying the binder ink to the first region of the surface of the substrate; and

configure a mechanical brusher to brush the surface of the substrate to remove material from an unprotected second region of the surface of the substrate, wherein removing the material from the unprotected second region of the surface of the substrate forms a gradient 3D relief; and

transmitting the control commands to the digital inkjet printer to form the 3D relief on the substrate based on the relief pattern.

2 . The method of claim 1 , wherein applying the solvent ink causes at least some of the binder ink to sink into the surface of the substrate such that the protective layer formed by the binder ink is at a gradient depth relative to the surface of the substrate.

3 . The method of claim 1 , wherein the binder ink includes a resin solution.

4 . The method of claim 1 , wherein the substrate is a double-charged ceramic tile including a top layer of absorbent material upon which ink can be applied using digital inkjet printing.

5 . The method of claim 1 , wherein the machine learning model is trained to generate the control commands to:

configure a glazing apparatus to apply glaze to the surface of the substrate after the 3D relief has been formed.

6 . The method of claim 1 , wherein the machine learning model is trained to generate the control commands to:

configure the digital inkjet printer to apply color ink to the surface of the substrate after the 3D relief has been formed.

7 . The method of claim 1 , wherein the machine learning model is trained to generate the control commands to:

configure an automated kiln to fire the substrate after the 3D relief has been formed to produce a ceramic tile.

8 . A system comprising:

one or more processors; and

a non-transitory, computer-readable storage medium comprising instructions recorded thereon that, when executed by at least one processor of the system, cause the system to:

receive information defining a relief pattern for forming a three-dimensional (3D) relief on a substrate;

extract, from the information, a feature vector comprising one or more features describing the 3D relief;

generate, using a machine learning model, control commands based on the feature vector,

wherein the machine learning model is trained, based on parameters associated with forming 3D reliefs in substrates, to generate the control commands to:

configure a digital inkjet printer to apply binder ink to a first region of a surface of the substrate, wherein the applied binder ink is configured to form a protective layer over the first region of the surface of the substrate; and

configure a mechanical brusher to brush the surface of the substrate to remove material from an unprotected second region of the surface of the substrate, wherein removing the material from the unprotected second region of the surface of the substrate forms a gradient 3D relief; and

send the control commands to the digital inkjet printer to form the 3D relief on the substrate based on the relief pattern.

9 . The system of claim 8 , wherein the machine learning model is trained to generate the control commands to:

configure the digital inkjet printer to apply solvent ink to the surface of the substrate after applying the binder ink to the first region of the surface of the substrate and before brushing the surface of the substrate.

10 . The system of claim 8 , wherein the parameters include a material type of the substrate or dimensions of the substrate.

11 . The system of claim 8 , wherein the parameters include a depth of the 3D relief or a gradient of the 3D relief.

12 . The system of claim 8 , wherein the parameters describe at least one of a glazing process, a color printing process, or a firing process.

13 . The system of claim 8 , wherein the substrate includes at least one of polyvinyl chloride (PVC), polymethyl methacrylate (PMMA), acrylonitrile butadiene styrene (ABS), or luxury vinyl tile (LVT).

14 . The system of claim 8 , wherein the information includes a digital image.

15 . A non-transitory, computer-readable storage medium comprising instructions recorded thereon that, when executed by at least one processor of an automated production system, cause the automated production system to:

receive a polychrome image defining a relief pattern for forming a three-dimensional (3D) relief on a substrate;

extract, from the polychrome image, a feature vector comprising one or more features describing the 3D relief;

generate, using a machine learning model, control commands based on the feature vector,

wherein the machine learning model is trained, based on parameters describing types of 3D reliefs or types of substrates, to generate the control commands to:

cause a digital inkjet printer to apply binder ink to a first region of a surface of the substrate, wherein the applied binder ink is configured to form a protective layer over the first region of the surface of the substrate; and

cause a mechanical brusher to brush the surface of the substrate to remove material from an unprotected second region of the surface of the substrate, wherein removing the material from the unprotected second region of the surface of the substrate forms a gradient 3D relief based on the relief pattern.

16 . The non-transitory, computer-readable storage medium of claim 15 , wherein the machine learning model is trained to generate the control commands to:

cause the digital inkjet printer to apply solvent ink to the surface of the substrate after applying the binder ink to the first region of the surface of the substrate and before brushing the surface of the substrate,

wherein applying the solvent ink causes at least some of the binder ink to sink into the surface of the substrate such that the protective layer formed by the binder ink is at a gradient depth relative to the surface of the substrate.

17 . The non-transitory, computer-readable storage medium of claim 15 , wherein the parameters include a material type of the substrate or dimensions of the substrate.

18 . The non-transitory, computer-readable storage medium of claim 15 , wherein the parameters include a depth of the 3D relief or a gradient of the 3D relief.

19 . The non-transitory, computer-readable storage medium of claim 15 , wherein the machine learning model is trained to generate the control commands to:

configure the digital inkjet printer to apply color ink to the surface of the substrate after the 3D relief has been formed.

20 . The non-transitory, computer-readable storage medium of claim 15 , wherein the machine learning model is trained to generate the control commands to:

cause a glazing apparatus to apply glaze to the surface of the substrate after the 3D relief has been formed.