PLASTIC OBJECT AND DEVICE AND METHOD FOR DETERMINING A MATERIAL TYPE OF THE PLASTIC OBJECT

Abstract

Various aspects relate to a plastic article (100) and to a device and a method for determining a material type of the plastic article (100), wherein the plastic article (100) includes: a support structure (122) consisting essentially of a plastic material and including a marking section (120), wherein the marking section (120) of the support structure (122) includes a variation in a thickness of the plastic material, wherein the variation in the thickness of the plastic material is configured to encode at least material information representing a material type of the plastic material.

Description

TECHNICAL FIELD

[0001] Various aspects of this disclosure relate to a plastic article and an apparatus and method for determining a material type of the plastic article. BACKGROUND

[0002] More than 400 million tons of plastic are produced worldwide each year. Much of this plastic is poorly managed due to the current linear "take-make-waste" economy. This overconsumption of plastics and the lack of recycling efforts have led to plastic pollution that impacts our biodiversity and also human health. In many developing countries, municipalities often lack the framework, resources, and / or capacity to implement an efficient and formal waste collection and treatment system. Therefore, it may be necessary to improve waste sorting, especially the sorting of plastics. Various sorting machines currently in use are unable to distinguish between specific post-consumer plastics (e.g., mixed plastics, blended plastics, food-grade plastics, etc.).) and / or soft packaging plastics (e.g., dark-colored plastics, black plastics, multi-layer plastics, etc.) and / or industrial plastics (e.g., acrylonitrile butadiene styrene (ABS), polyamide (PA)-based industrial plastics such as PA6-GF, PA66, and PA66-GF, etc.). For example, industrial plastics PA6 and PA66 have a higher carbon footprint than other plastic materials; therefore, their carbon dioxide equivalent can be drastically reduced through appropriate recycling. In general, it has been shown that distinguishing different types of plastic within a sorting system can increase the revenue of the recycling facility owner by approximately 20% to approximately 100%. For sorting plastic items (e.g., plastic bottles, plastic jars, plastic containers, plastic plates, plastic bowls, plastic packaging, plastic films, plastic boxes, etc.),) it may be necessary to determine the type of plastic material from which the plastic articles are made. SUMMARY

[0003] Various aspects of the disclosure relate to a plastic article comprising a marking that encodes at least material information representing the material type of the plastic material and that is detectable by terahertz transmission imaging. According to some aspects, this marking may, for example, be made of the same plastic material and encode the material information by varying the thickness of the plastic material. This has the advantage that no additional materials (e.g., chemicals) are added to encode the material information. Other methods for encoding material information, such as barcodes, quick response (QR) stickers, etc., rely on additional materials that reduce recycling efficiency. According to other aspects, this marking may include various elements made of a different material than the plastic material (e.g.,a different plastic material than the plastic material (e.g., a plastic material having a different chemical composition), a metal, or a metal alloy), and having a terahertz radiation absorption coefficient that differs from a terahertz radiation absorption coefficient of the plastic material of the plastic article, such that they are detectable via terahertz transmission imaging. The elements made of a different material may be integrated into the plastic article, molded onto the plastic article, and / or attached to the plastic article (e.g., using an adhesive). Various aspects of the disclosure relate to an apparatus and a method for detecting such a plastic article having the marking.The material type coded by the marking can be or represent a chemical composition of the plastic material, such as high density polyethylene (HDPE), polypropylene (PP), polystyrene (PS), low density polyethylene (LDPE), polyvinyl chloride (PVC), polyethylene terephthalate (PET), etc.

[0004] According to the present disclosure, a plastic article is provided including such a marking as claimed in claims 1 and 6. An apparatus for detecting such a marking on and / or in a plastic article according to the present disclosure is defined in claim 7. A method for determining a material type of a plastic material of a plastic article according to the present disclosure is defined in claim 10.

[0005] The dependent claims define some examples belonging to the plastic articles, the device and the method, respectively. BRIEF DESCRIPTION OF THE DRAWINGS

[0006] The invention will be better understood by reference to the detailed description taken in conjunction with the non-limiting examples and the accompanying drawings in which: - Fig. 1A to Fig. 1J each show a plastic article according to various embodiments; - Fig. 2A and Fig. 2B shows a system for determining a material type of plastic articles according to various embodiments; - Fig. 3A to Fig. 3C each show a corresponding plastic object and a correspondingly detected terahertz image; - Fig. 4 shows an exemplary machine learning model for recognizing a coding portion of a plastic article according to various embodiments; and - Fig. 5 shows an exemplary processing scheme for detecting a coding portion of a plastic article according to various embodiments. DETAILED DESCRIPTION

[0007] The following detailed description refers to the accompanying drawings, which, by way of illustration, show specific details and embodiments in which the disclosure may be practiced. These embodiments are described in sufficient detail to enable those skilled in the art to practice the disclosure. Other embodiments may be utilized, and structural and logical changes may be made without departing from the scope of the disclosure. The various embodiments are not necessarily mutually exclusive, as some embodiments may be combined with one or more other embodiments to form new embodiments.

[0008] Features described in connection with one embodiment may be correspondingly applicable to the same or similar features in the other embodiments. Features described in connection with one embodiment may be correspondingly applicable to the other embodiments, even if they are not explicitly described in these other embodiments. Furthermore, additions and / or combinations and / or alternatives as described for a feature in connection with one embodiment may be correspondingly applicable to the same or a similar feature in the other embodiments.

[0009] In the context of various embodiments, the articles "a," "an," and "the," as used with respect to a feature or element, include a reference to one or more of the features or elements.

[0010] The term “and / or” as used herein includes any combination of one or more of the related items listed.

[0011] In one embodiment, a "computer" may be any type of logic-implementing device, which may be hardware, software, firmware, or any combination thereof. Thus, in one embodiment, a "computer" may be a hard-wired logic circuit or a programmable logic circuit such as a programmable processor, e.g., a microprocessor (e.g., a Complex Instruction Set Computer (CISC) processor or a Reduced Instruction Set Computer (RISC) processor). A "computer" may also be software implemented or executed by a processor, e.g., any type of computer program, e.g., a computer program using virtual machine code such as Java. A "computer" may be or include one or more processors.Any other way of implementing the respective functions, which is described in more detail below, may also be understood as a “computer” according to an alternative embodiment.

[0012] A "memory" may be used in processing performed by a computer and / or may store data used by the computer. A "memory" used in embodiments may be a volatile memory, such as a DRAM (Dynamic Random Access Memory), or a non-volatile memory, such as a PROM (Programmable Read Only Memory), an EPROM (Erasable PROM), an EEPROM (Electrically Erasable PROM), or a flash memory, such as a floating-gate memory, a charge-trapping memory, an MRAM (Magnetoresistive Random Access Memory), or a PCRAM (Phase Change Random Access Memory).

[0013] The term "item" as used herein includes any item, e.g., a recyclable or reusable item that can be sorted by type and / or class. A "plastic item" can be any item that includes (e.g., is made of) a plastic material. Examples of (plastic) materials of plastic items can be high-density polyethylene (HDPE), polypropylene (PP), polystyrene (PS), low-density polyethylene (LDPE), polyvinyl chloride (PVC), polyethylene terephthalate (PET), etc. Such items can include bottles, jars, containers, plates, bowls, etc., in various shapes, sizes, and designs (e.g., partially compressed, deformed). It is understood that a plastic item can consist essentially of the plastic material or, in addition to the plastic material, can also include other materials (e.g., one or more metals and / or metal alloys, etc.).

[0014] Although the disclosure has been particularly shown and described with reference to specific embodiments, it should be understood by those skilled in the art that various changes in form and detail may be made therein without departing from the spirit and scope of the invention as defined in the appended claims. The scope of the invention is thus indicated by the appended claims, and all changes which come within the meaning and range of equivalence of the claims are therefore intended to be embraced.

[0015] To enable plastics recycling, it may be necessary to separate different types of plastics from waste, which may also contain other materials such as paper, metals, cardboard, etc. Furthermore, plastics recycling may require sorting plastic items according to material type. Various aspects relate to a plastic item having a marking portion that encodes at least material information representing the material type of the plastic item and that is detectable via terahertz transmission imaging.

[0016] Fig. 1A to Fig. 1J show a plastic article 100 according to various aspects. For illustrative purposes, the figures show an (open) plastic box as the plastic article 100. It is understood that the plastic box is an illustrative example and that the plastic article 100 may also be another type of plastic article. Thus, the plastic article 100 may be any article that encloses (e.g., consists of) a plastic material, such as a plastic bottle, a plastic jar, a plastic container, a plastic plate, a plastic bowl, a plastic package, a plastic film, a plastic box, plastic cutlery, etc. Fig. 1A shows a front view 102, a top view 104, a bottom view 106, and side views 108 and 110 for the example plastic box. The plastic article 100 may include a marking portion 120. The marking portion 120 may be configured to encode at least material information representing a material type of the plastic material that the plastic article 100 includes. Therefore, in addition to the material information, the marking portion 120 may also encode other information (e.g., a product category (this information may, for example, be important to comply with the Extended Producer Responsibility (EPR) and / or the ESG (Environmental, Social, and Governance) perspective of the producer and / or manufacturer), a manufacturer, a hazard class (e.g.,if hazardous additives are used) or other information such as a CO2 equivalent, a carbon footprint, disposal and recycling options, a lifetime based on the year of production, etc.). For example, the marking portion 120 may also encode any information that a regular barcode encodes. The marking portion 120 may also be referred to as a geometric mark or geometric identifier. The marking portion 120 may be located anywhere on or at a plastic article, such as the plastic article 100.

[0017] Fig. 1 B to Fig. 1G each show a sectional view of the marking portion 120 according to various embodiments. The plastic article 100 may include a support structure 122. The support structure 122 may consist essentially of the plastic material. The support structure 122 may include the marking portion 120.

[0018] According to some embodiments, the marking portion 120 may be part of the support structure 122 (see, for example, Fig. 1B to Fig. 1D). Therefore, the marking portion 120 may consist essentially of the (same) plastic material from which the support structure 122 is made. In this case, the marking portion 120 may include a variation in the thickness of the plastic material. This variation in thickness (referred to as thickness variation in some aspects) may be configured to encode at least the information representing the material type of the plastic material. The marking portion 120 may include a plurality of elements 124. The elements (may also be referred to as markings) of the plurality of elements 124 may have a first thickness 123 that differs from a second thickness 121 of the support structure 122. For example, the first thickness 123 of the plurality of elements 124 may be greater than the second thickness 121 (see, for example, Fig. 1B). As a further example, the second thickness 121 of the support structure 122 may be greater than the first thickness 123 of the plurality of elements 124 (see, for example, Fig. 1C). As yet another example, the thickness of the marking portion 120 may vary irregularly (see, for example, Fig. 1D). The thickness variation of the support structure 122 (and thus the thickness variation of the plurality of elements within the marking section 120) can be detected using terahertz imaging (see, for example, the description with respect to Fig. 2A and Fig. 2B). Therefore, the marking portion 120 may have a modulated thickness of the plastic material (and, for example, the edges of the plurality of elements (e.g., the perimeter of a box) may be used as a geometric marker that is only functional under THz imaging). In some aspects, the plastic article 100 may be a three-dimensional (3D) printed plastic article. For example, using the same material for the marking portion 120 allows the plastic article to be manufactured using a 3D printing device that uses (e.g., has) only one print head.

[0019] According to other embodiments, the plurality of elements 124 (within the marking portion 120) may be made substantially of a material different from the plastic material of the support structure 122 (see, for example, Fig. 1E to Fig. 1G). The material of the plurality of elements 124 may have a terahertz radiation absorption coefficient that differs significantly from a terahertz radiation absorption coefficient of the plastic material of the support structure 122. This may allow the plurality of elements 124 to be distinguished from the plastic material of the support structure 122 using terahertz transmission imaging (see, for example, description with reference to Fig. 2A and Fig. 2B). The plurality of elements 124 may be arranged over the support structure 122 (see, for example, Fig. 1E) or (e.g. completely) in the support structure 122 (see for example Fig. 1 F) or partially in the support structure 122 (see for example Fig. 1G). Examples of such distinguishable materials are metallic materials (e.g., metals or metal alloys) and plastic materials that differ from the plastic material of the support structure 122. However, it is understood that the plurality of elements may also be made of any other material, as long as the material is distinguishable from the plastic material of the support structure 122 using terahertz imaging. If the plurality of elements 124 is arranged entirely within the support structure 122 (see, for example, Fig. 1F), the plurality of elements 124 may be invisible to the human eye.

[0020] As described herein, the arrangement of the plurality of elements 124 of the marking portion 120 (referred to in some aspects as the plurality of elements 124 within the marking portion 120) may encode at least the material information of the plastic material of the plastic article 100. Therefore, the elements of the plurality of elements 124 may be arranged at varying distances from one another and / or have varying widths and / or varying thicknesses.

[0021] With reference to Fig. 1 H to Fig. 1J, the marking portion 120 may include a coding portion 126. The coding portion 126 may include the plurality of elements 124 (e.g., as a thickness variation of the plastic material of the support structure 122 and / or as elements of different (distinguishable) material(s)). The marking portion 120 may also include an indicator portion 128 (see, for example, Fig. 1H). The display portion 128 may be configured to display the coding portion 126 of the plastic article 100. For example, the display portion 128 may point to the coding portion 126. The display portion 128 may be disposed between the coding portion 126 and a surrounding portion of the support structure 122. The display portion 128 may include one or more elements that indicate a position of the plurality of elements 124. These one or more elements may have a different thickness than the surrounding portion of the support structure 122 and / or may be made of one or more materials that are different from the plastic material of the support structure 122, as described herein. Therefore, the one or more elements of the display portion 128 may also be distinguishable using terahertz imaging. Fig. 1I shows an exemplary configuration in which the display section 128 includes four elements 128(1), 128(2), 128(3), 128(4) that define a rectangle and thereby indicate that the coding section 126 is located within this rectangle. Fig. 1J shows another example configuration where the indicator portion 128 includes a rectangular element indicating that the coding portion 126 is located within this rectangular element. For example, the indicator portion 128 may be the rectangular element. It is understood that these configurations are examples and that the indicator portion 128 may have any configuration suitable for indicating a position of the coding portion 126 (such as an arrow element pointing to the coding portion 126, an ellipsoid-shaped element at least partially surrounding the coding portion 126, etc.).

[0022] The elements of the plurality of elements 124 may have any shape, pattern, and type (e.g., one or more elements comprising the plurality of elements 124 (e.g., at least one element of the plurality of elements 124 may be made of a material that is different from others)) suitable for encoding information. For example, the elements of the plurality of elements 124 may be rectangular (e.g., lines), like a barcode (e.g., the coding portion 126 may be like a three-dimensional barcode), symbols (e.g., letters), hollow (e.g., a hollow rectangle, a hollow ellipse, etc.), etc. In some aspects, the plurality of elements 124 may include at least one of the following patterns: a date stamp, a part identification (ID), a recycling symbol, a material composition of the part, information regarding the disposal of the part at the end of its life (e.g.,depending on legal requirements), a logo, etc.

[0023] Fig. 2A and Fig. 2B show a system 200 for determining a material type of the plastic article according to various embodiments.

[0024] The system 200 may include a conveyor 205. The conveyor 205 may be configured to transport items. As another example, the system 200 may be, or be part of, a waste sorting system, and the conveyor 205 may be configured to transport waste. Such waste may include items from different material categories, such as metals, cardboard, organic materials, plastics, etc. The waste may be pre-sorted or not pre-sorted. The conveyor 205 may be any type of transport device capable of transporting items (e.g., waste), such as a conveyor belt, a conveyor roller, etc. For example, the system may be or include a reverse vending machine (e.g., for simple and efficient waste separation) (in which case the conveyor 205 may be configured to transport returned empties (e.g., bottles)) or a portable device (e.g.,similar to a barcode reader / scanner). It should be understood that these are merely examples, and that the principle described here is not limited to portable THz-based scanning devices / systems with conveyors or reverse vending machines (for better waste source separation).

[0025] The system 200 may include a sensor module. The sensor module may include a terahertz source 202. The terahertz source 202 may be configured to irradiate objects, such as the plastic object 100, on the conveyor 205 with terahertz radiation 203. The terahertz source 202 may be configured to irradiate the plastic object 100 on the conveyor 205 directly or indirectly (e.g., using one or more reflectors and / or one or more lenses). The sensor module may include a terahertz camera 206. The terahertz camera 206 may be configured to detect terahertz radiation passing through the irradiated plastic object 100. The terahertz camera 206 may be configured to detect a terahertz image 204. The terahertz source 202 may be arranged above the conveyor 205. The terahertz camera 206 may be arranged below the conveyor 205 (e.g.,at least partially within a frame of the conveyor 205). The terahertz camera 206 may be arranged (e.g., directly) below the upper conveyor belt material (see, for example, . Fig. 2A). The sensor module may be a linear THz image sensing system. The sensor module may be a high-speed terahertz imaging scanner from TeraSense [1] using any source type and any camera type. According to various aspects, the terahertz source 202 and the terahertz camera 206 may continuously acquire data.

[0026] The term terahertz radiation here may refer to electromagnetic radiation in the terahertz range / spectrum (THz) and may refer to electromagnetic radiation having a frequency of 0.1 THz or more. Terahertz radiation can penetrate not only optically transparent but also opaque objects. The image data or signals detected by the terahertz camera 206 may include the penetration intensity of the electromagnetic radiation (which may be referred to as "power intensity"). It follows that image data or signals based on the "power intensity" may include heat maps that show or depict the power penetration distribution. The terahertz camera may be configured to provide a thermal image of the THz power intensity (as a terahertz image 204) using the material's transmission signals for THz radiation.Terahertz radiation can be non-destructive to the plastic object 100. Therefore, the terahertz camera 206 can capture a terahertz image 204 showing a heat map representing the distribution of the power penetration intensity of the object 100. It is conceivable that multiple (power intensity heat map) images 204 of the same object with different timestamps may be captured as multiple individual images. It is contemplated that the object 100 may have different shapes, sizes, and forms. Graphical examples of a terahertz (heat map) image 204 (also referred to as a THz output image or THz image) are shown in FIG. Fig. 3A to Fig. 3C shown.

[0027] Using terahertz radiation can be advantageous over other radiation types, such as hyperspectral optical sensor systems in the near-infrared (NIR) range. NIR cannot penetrate plastics and cannot extract color information from objects. Furthermore, NIR systems are prone to errors when the plastic objects are contaminated with organic materials, such as food waste. Systems using an NIR optical sensor system are unable to detect labels on plastic bottles, black plastics, and multilayer plastics; or an associated classification can lead to incorrect classification if the label material (e.g., PVC) is output rather than the plastic bottle material (e.g., PET).

[0028] Fig. 3A shows RGB images 300 of a top view (top left image) and a bottom view (bottom left image) of an exemplary plastic article as well as a respective detected terahertz (heat map) image 204. The support structure 122 and the marking section 120 of the in Fig. 3A are made of the same plastic material (specifically, acrylonitrile butadiene styrene (ABS)), and the marking portion 120 exhibits the thickness variation of this plastic material. As illustratively shown, the plurality of elements of the marking portion 120 can be detected by both top and bottom irradiation. Thus, in a sorting line, the plastic material can be determined even if the plastic article 100 is positioned such that the marking portion 120 is opposite a viewing direction (e.g., in contact with the conveyor).

[0029] Fig. 3B shows an RGB image 302 of an exemplary plastic object and a correspondingly detected terahertz (heat map) image 204. The support structure 122 and the marking section 120 of the Fig. 3A are made of different plastic materials (in particular, the support structure 122 is made of acrylic and the plurality of elements 124 is made of ABS).

[0030] Fig. 3C shows an RGB image 304 of an exemplary plastic object and a correspondingly detected terahertz (heat map) image 204. The support structure 122 of the Fig. 3A is made of a plastic material and the plurality of elements 124 is made of a metal (specifically, the support structure 122 is made of acrylic and the plurality of elements 124 is made of aluminum).

[0031] It should be noted that the RGB images in Fig. 3A to Fig. 3C is used to hold the plurality of elements 124 in place and has no effect on the THz image.

[0032] With reference to Fig. 2B, the system 200 may be a sorting system. The system 200 may include one or more robot arms 212. A robot arm (short robot), as used herein, may be configured to pick up an object (e.g., a waste object) from the conveyor 205, to move the picked object, and to release the object after moving it (e.g., to pick up an object and place it at specific locations (e.g., into discharge bins along the conveyor)). A robot arm may be a delta robot or an industrial robot arm. A robot arm may include a gripper configured to pick up an object by grasping the object (see, for example, the first robot arm 212(1)). A robot arm may include a suction head configured to pick up an object using negative pressure (i.e., pressure below atmospheric pressure) (see, for example, the second robot arm 212(2)).The terahertz camera may be arranged in front of the one or more robot arms 212. The system 200 may include additional components that the robot arm(s) may employ to pick up and move an object (e.g., pick up and place an object), such as a camera (e.g., a red-green-blue (RGB) camera) that tracks the objects within the waste, and any type of image recognition software. The one or more robot arms may be preconfigured depending on the configuration of the production line.

[0033] Conveyor 205 may move in direction of travel 201. System 200 may include a computer 210 or be connected to computer 210 (e.g., the computer may be a cloud computing device). Computer 210 may be configured to receive a representation of the detected terahertz radiation (e.g., an image) from terahertz camera 206. Computer 210 may be configured to receive THz image 204 from terahertz camera 206. Computer 210 may be configured to determine the material type of the plastic material of plastic article 100 using THz image 204.

[0034] The computer 210 may be configured to implement a model (e.g., a machine learning model). The model may be configured to output the material type of the plastic material of the plastic article 100 in response to the THz image 204 being input to the model. Therefore, the THz image 204 may include a representation (e.g., a heat map) of the marking portion 120, and the model may be configured to decode the material information encoded in the marking portion 120. A "model," for example, may be based on machine learning (e.g., employ a machine learning algorithm). For example, a "model" may be adapted (e.g., trained) using machine learning. A "model" may be a decision tree model, a random forest model, a gradient boosting model, a support vector machine, a k-nearest neighbor model, a neural network, etc.A "neural network" can be any type of neural network, such as an autoencoder network, a convolutional neural network, a variational autoencoder network, a sparse autoencoder network, a recurrent neural network, a deconvolutional neural network, a generative adversarial network, a forward-thinking neural network, a sum-product neural network, etc. For example, a contour-based object detection algorithm can be employed to identify one or more regions of interest (ROI) from the THz image 204, where each ROI corresponds to a respective object to be classified. Therefore, the computer 210 can perform machine learning-based decoding of marker sections (e.g.,for material classification) (using deep learning object classification in an artificial intelligence platform and a detailed material identification module using terahertz radiation). According to various aspects, the inference of the model can be performed on an edge computing device. For example, the system 200 can be configured to automatically identify the material type of a plastic material upon detecting a plastic object 100 using terahertz imaging.

[0035] In the example of system 200, which is a sorting system, computer 210 may be further configured to determine a respective waste type of one or more items within the irradiated waste using the representation of the detected terahertz radiation (e.g., THz image 204). Computer 210 may be configured (e.g., in addition to determining the material type of the plastic material based on marker portion 120 of plastic article 100) to determine a respective waste type of objects within the waste. For example, computer 210 may determine (e.g., using a model) whether the item is a metal item, a cardboard item, or a plastic item. Computer 210 may be configured to determine the waste type of an item by (e.g.,automated) object identification, feature extraction and classification to extract key features, and it can use image classification techniques to automatically distinguish between at least two classes of waste types, for example, between an HDPE bottle and a PET bottle.

[0036] The model for determining the material type of the plastic material of the plastic article may include a machine learning model. Fig. 4 shows an exemplary machine learning model 402 for recognizing a marker portion of a plastic article, such as the marker portion 120 of the plastic article 100, according to various embodiments. The machine learning model 402 may be a U-Net model (e.g., segmentation model) [2]. A training dataset may be used to train the machine learning model 402. The training dataset may include a plurality of THz images (showing a respective marker portion 120), wherein the respective marker portion 120 (e.g., geometric markers) of each THz image of the plurality of THz images is annotated. The training dataset may further include a respective mask image for each THz image of the plurality of THz images (e.g., including 500 THz images or more) (see, for example, the mask image 404 corresponding to the THz image 204).

[0037] The architecture of the U-Net model (segmentation) may include a contraction path (also called an encoder) used to capture the context of the THz image and an (e.g., symmetric) expansion path (also called a decoder) that enables precise localization using transposed convolutions. Thus, it may be an end-to-end full convolutional neural network (FCN) (i.e., it contains only convolutional layers and no dense layers). In this example, an Adam optimizer and a binary cross-entropy loss function may be used in the U-Net (e.g., since there may be only two classes, geometric marker sections and background). It is understood that the machine learning model 402 and any other suitable model may be employed.

[0038] Fig. 5 shows an example processing scheme 500 for detecting a coding portion of a plastic article using the trained machine learning model 402. An example RGB image 502 of the plastic article 100 is shown. A sensor module (e.g., of the system 200) may detect a THz image 204 showing at least the marking portion 120 of the plastic article 100. The computer 210 may implement a model (e.g., the machine learning model 402). The THz image 204 may be input to the model, and the (e.g., segmentation) model may output (e.g., predict) a mask image 404 that masks the coding portion 126. Computer 210 may be configured to overlay the THz image 204 and the mask image 404 (e.g., subtract the mask image 404 from the THz image 204) to generate an encoding image 406. The encoding image 406 may show a representation of the encoding section 126.The computer 210 may be configured to determine at least the material type of the plastic material by decoding the information in the coding image 406.

[0039] Therefore, a method for determining the material type of a plastic article may comprise determining a material type of a plastic material of a plastic article using the terahertz image. The plastic article may include a support structure enclosing (e.g., consisting essentially of) the plastic material. The plastic article may further include a marking portion in and / or on the support structure. The marking portion may be configured to encode material information representing a material type of the plastic material of the support structure. The terahertz image may represent terahertz radiation having passed through at least the marking portion in response to irradiating the plastic article with terahertz radiation.

[0040] Compared to previous solutions for determining the material type of plastic objects, the present disclosure improves robustness against dirt and / or moisture composition on the objects. The present disclosure also provides high security (e.g., the marking portion 120 may be difficult to tamper with (and thus counterfeit-proof) compared to barcode labels). Furthermore, in the case where the marking portion 120 includes the thickness variation of the plastic material, the present disclosure improves the purity of the plastic object (e.g., its material) and makes the plastic object more suitable for recycling, since no additional new material is present for encoding geometric markings.Compared to using markers for terahertz imaging according to the present disclosure, previous approaches using markers have several disadvantages, as described in detail below:. Barcodes and / or Quick Response Codes (QR Codes) must face the camera to be read. This orientation of the code is impossible, for example, in practical plastics recycling facilities (such as a plastics sorting plant). Compared to the marking section 120 described here, barcodes and / or QR Codes can be easily damaged (e.g., by dirt, scratches, moisture, etc.). Furthermore, the marking section 120 described here increases security over barcodes and QR Codes because no new label can be affixed to the marking section 120 (in the case of a different material, the marking section 120 may even be "invisible"). Furthermore, barcodes and QR Codes have poor readability due to geometric distortions. The problem of facing the camera remains even with approaches such as the Digimarc barcode in HolyGrail 2.0 [3], which propose that digital watermarks (Digimarc barcodes) can be integrated directly into a mold. This leads to manufacturing process complexity and requires time to implement them across a broad product portfolio. This line-of-sight detection is also a problem with approaches such as ArUco Markings, which are used for marker-based remote tracking (not for traceability and recycling purposes) to create digital twins of construction waste (e.g., wood scraps) [4].

[0041] The use of radio frequency identification (RFID) tags poses problems when used in sorting systems, for example, because metallic materials, moisture, and / or liquids can cause interference that prevents their detection. Furthermore, the battery life of RFID devices is limited.

[0042] Plastics may contain portions of laser-sensitive pigments, which are used to create a mark through laser irradiation. This, in turn, requires line-of-sight, and laser marking results in additional processes that require processing time and incur processing costs.

[0043] Items may be marked using fluorescent markers (e.g., ultraviolet fluorescent (UV fluorescent) tracers) that add additional (fluorescent) material (e.g., chemicals), are only visible under UV light, may not be able to store as much information as the marker portion 120, and have readability issues because readability is reduced when contaminants (e.g., dirt, moisture, etc.) are present on a surface of the marker.

[0044] The marking portion 120 for THz imaging can be attached to or integrated into the plastic article 100 as described herein and can provide robustness against dirt and moisture. Furthermore, in the case of thickness variation, no additional materials (e.g., chemicals) are added. Furthermore, the marking portion 120 increases security because it is tamper-proof. As described in Fig. As shown in Figure 2B and described with reference thereto, a sorting system enables distinguishing between waste types in general and identifying the geometric markings of plastic objects using the same method, i.e. terahertz imaging. References [1] https: / / terasense.com / products / thz-scanner [2] Olaf Ronneberger, Philipp Fischer, Thomas Brox, “U-Net: Convolutional Networks for Biomedical Image Segmentation,” Medical Image Computing and Computer-Assisted Intervention (MICCAI), Springer, LNCS, Vol. 9351: 234--241, 2015 [3] HolyGrail 2.0 Pioneering Digital Watermarks for Smart packaging recycling in the EU: https: / / www.digitalwatermarks.eu / . Last accessed on 28.03.22. [4] Caitlyn Parry and Sean Guy, “Recycling Construction Waste Material with the Use of AR.” In the book: Proceedings of the 2020 DigitalFUTURES QUOTES CONTAINED IN THE DESCRIPTION

[0000] This list of documents submitted by the applicant was generated automatically and is included solely for the convenience of the reader. This list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions. Cited non-patent literature

[0000] https: / / terasense.com / products / thz-scanner

[0044] Olaf Ronneberger, Philipp Fischer, Thomas Brox, “U-Net: Convolutional Networks for Biomedical Image Segmentation”, Medical Image Computing and Computer-Assisted Intervention (MICCAI), Springer, LNCS, Vol. 9351: 234--241, 2015

[0044] HolyGrail 2.0 Pioneering Digital Watermarks for Smart Packaging Recycling in the EU: https: / / www.digitalwatermarks.eu / . Last accessed on 28.03.2022

[0044] Caitlyn Parry and Sean Guy, “Recycling Construction Waste Material with the Use of AR.” In the book: Proceedings of the 2020 DigitalFUTURES

[0044]

Claims

[1] Plastic article (100) comprising: • a support structure (122) consisting essentially of a plastic material and comprising a marking portion (120); • wherein the marking portion (120) of the support structure (122) comprises a variation in a thickness of the plastic material, wherein the variation in the thickness of the plastic material is configured to encode at least material information representing a material type of the plastic material. [2] The plastic article (100) of claim 1, wherein the marking portion (120) of the support structure (122) comprises: • a coding section (126) comprising the variation of the thickness of the plastic material, and • a display section (128) pointing to the coding section (126). [3] The plastic article (100) of claim 1 or 2, wherein the support structure (122) comprises a surrounding portion surrounding the marking portion (120), the indicator portion (128) having a thickness different from a thickness of the surrounding portion. [4] The plastic article (100) according to any one of claims 1 to 3, wherein the marking portion (120) comprises a plurality of elements (124) having a thickness different from a thickness of the support structure (122). [5] The plastic article (100) of claim 4, wherein the plurality of elements (124) comprises at least a first element having a first width and a second element having a second width, the first width being different from the second width; and / or wherein a first pair of elements of the plurality of elements (124) are arranged at a first distance from each other and wherein a second pair of elements of the plurality of elements (124) are arranged at a second distance from each other, the second distance being different from the first distance. [6] Plastic article (100) comprising: • a support structure (122) comprising a plastic material; and • a marking section (120) comprising a plurality of elements (124) on and / or in the support structure (122), wherein the plurality of elements (124) is configured to encode material information representing a material type of the plastic material of the support structure (122), and wherein each element of the plurality of elements (124) consists essentially of one or more plastic materials that are different from the plastic material of the support structure (122), of a metal or a metal alloy. [7] The plastic article (100) according to claim 6, wherein the marking portion (120) comprises: • a coding section (126) comprising the plurality of elements (124), and • a display section (128) pointing to the coding section (126). [8] Device comprising: • a terahertz source (202) configured to irradiate a plastic article (100) according to any one of claims 1 to 7 with terahertz radiation; • a terahertz camera (206) configured to capture a terahertz image (204) representing the terahertz radiation passing through the plastic article (100) in response to the terahertz source (202) irradiating the plastic article (100); and • a computer (210) configured to determine the material type of the plastic material of the support structure of the plastic article (100) using the terahertz image (204). [9] The apparatus of claim 8, wherein the computer (210) is configured to implement a model configured to output the material type of the plastic material in response to inputting the terahertz image (204) into the model. [10] Method comprising: • Detecting a terahertz image of a plastic object, wherein the plastic object comprises a support structure comprising a plastic material and a marking portion in and / or on the support structure, wherein the marking portion is configured to encode material information representing a material type of the plastic material of the support structure, and wherein the terahertz image represents terahertz radiation passing at least through the marking portion in response to irradiating the plastic object with terahertz radiation; and • Determining the material type of the plastic material of the support structure of the plastic article using the terahertz image. [11] A computer program product comprising instructions for causing the apparatus of any one of claims 8 or 9 to perform the steps of the method of claim 10. [12] A computer-readable medium having the computer program product according to claim 11 stored thereon.

Images