Deposit return machine for bulk-feeding containers

Abstract

A deposit return machine for receiving containers, the machine comprising a conveyor; a receiving station for receiving the containers in bulk, the receiving station comprising a conditioning station comprising elongated objects which distribute the received containers over time and over a width of the conveyor; and a reading station downstream of the receiving station for counting the containers on the conveyor.

Description

DEPOSIT RETURN MACHINE FOR BULK-FEEDING CONTAINERSCROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims the priority or benefit of U.S. provisional patent application 63 / 528,232, filed July 21 , 2023, the specification of which is hereby incorporated herein by reference in its entirety.BACKGROUND(a) Field

[0001] The subject matter disclosed generally relates to apparatuses for material sorting. More specifically, it relates to deposit return machines.(b) Related Prior Art

[0002] There are various ways to collect containers in a deposit return scheme. Typically, such machines are provided in grocery stores or supermarkets and allow a user to insert individual containers sequentially in the machine. A bar code is read, and then the container is dragged inside the machine and a monetary value is assigned based on the barcode that was read.

[0003] Such systems focus on individual and sequential introduction and therefore are not scalable systems which would be able to treat, for example, batches of containers provided in bulk by customers.

[0004] Batch treatment would however make the process much faster and easier for customers, especially in a context in which a deposit return scheme is more widely introduced for a wider variety of containers, for example a greater variety of container types and materials.

[0005] US Patent 11 ,042,975 (Perron et al.) attempts detecting containers but fails to properly segregate materials introduced in the system to make the individual detection of specific, individual objects more accurate.

[0006] US1 1548729 (Klemmack) attempts at improving accuracy for recyclable material treatment, but still does not prevent materials provided in bulk to be adequately spaced apart for proper identification. This issue can be minor in recyclable material sortingcenters, but this issue is particularly acute in the context of a deposit return machine or center in which individual identification of containers is contemplated.

[0007] Therefore, current deposit return machine are not efficient at all, and the solutions from industrial residual material sorting centers are not well suited to the context of deposit return machines used in a commercial setting. The prior art systems described above have various drawbacks which fail to address the streamlined efficiency and accurate individual detection of containers in the context of a commercial deposit return machine or center of containers.SUMMARY

[0008] According to a first aspect of the disclosure, there is provided a 1 . A deposit return machine for receiving containers, the machine comprising:- a conveyor for receiving the containers in bulk;- a conditioning station which distributes the received containers over time and over a width of the conveyor; and- a reading station downstream of the receiving station for counting the containers on the conveyor.

[0009] According to an embodiment, the reading station operates a neural network for identification of each container being counted for associating a monetary value thereto.

[0010] According to an embodiment, the neural network for identification of each container being counted comprises a plurality of distinct neural networks, at least one of the plurality of distinct neural networks being dedicated to detecting object presence, at least one of the plurality of distinct neural networks being dedicated to segmenting said object presence as a plurality of distinct objects, and at least another one of the plurality of distinct neural networks being dedicated to classify said plurality of distinct objects individually.

[0011] According to an embodiment, there is further provided a receiving station comprising a reception panel which is inclined downwardly toward the conveyor for receiving the containers in bulk and perform sieving before moving the containers in bulk to the conveyor .

[0012] According to an embodiment, the conditioning station comprises a set of elongated obstacles, the elongated obstacles being suspended and free at a bottom end thereof, and arranged to form rows where at least two consecutive rows are staggered to have laterally offset, for distribution of the containers introduced in the reception panel.

[0013] According to an embodiment, the set of elongated obstacles comprises at least two consecutive rows aligned together.

[0014] According to an embodiment, the set of elongated obstacles comprises elongated obstacles ending at variable heights above the conveyor within a same one of the rows.

[0015] According to an embodiment, elongated obstacles comprise chains.

[0016] According to an embodiment, there is further provided a material separation system, downstream of the conveyor, comprising a brush to propel lighter containers into an ejection box.

[0017] According to an embodiment, the material separation system further comprises a blower feeding the ejection box to propel the lighter containers thereinto.

[0018] According to an embodiment, there is further provided an apparatus downstream of the reading station and performing an individual and sequential reading of a container identification tag of the containers to provide redundancy.

[0019] According to an embodiment, there is further provided an apparatus downstream of the reading station and performing a bulk reading of a barcode of a batch of containers to provide redundancy.

[0020] According to an embodiment, the apparatus comprises lighting to expose containers to a preset illumination, and the apparatus implements an artificial-intelligence based algorithm to perform the reading of the barcode.

[0021] According to an embodiment, there is further provided a sorting station downstream of the reading station to sort the containers following said counting.

[0022] According to an embodiment, the sorting station comprises a blower for ejection.

[0023] According to an embodiment, the sorting station comprises a plurality of distinct and interspaced nozzles which can be individually controlled for individualized ejection of containers passing by a given one of the distinct and interspaced nozzles.

[0024] According to an embodiment, the sorting station comprises a plurality of distinct and sequential sorting stations each for ejecting containers of a respective type.BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Further features and advantages of the present disclosure will become apparent from the following detailed description, taken in combination with the appended drawings, in which:

[0026] Fig. 1 is a perspective view illustrating a deposit return machine, according to an embodiment of the disclosure;

[0027] Fig. 2 is a perspective view illustrating a receiving station of a deposit return machine, according to an embodiment of the disclosure;

[0028] Fig. 3 is a picture illustrating rows of elongated obstacles for more evenly distributing received containers when they reach the conveyor, according to an embodiment of the disclosure;

[0029] Figs. 4 and 5 are a perspective view and a top view illustrating a belt surface of a conveyor, according to an embodiment of the disclosure;

[0030] Fig. 6 is a perspective view illustrating a reading station or scanning head of a deposit return machine, according to an embodiment of the disclosure;

[0031] Fig. 7 is a side view illustrating a material separation system (into two types of materials: heavy and light) of a deposit return machine, according to an embodiment of the disclosure;

[0032] Figs. 8A and 8B are a perspective views illustrating a brush and its paddles for use in the material separation system of Fig. 7, according to an embodiment of the disclosure, and

[0033] Fig. 9 is a schematic diagram illustrating the material separation system in the deposit return machine, in particular with an enlarged portion showing the sorting station, according to an embodiment of the disclosure.

[0034] It will be noted that throughout the appended drawings, like features are identified by like reference numerals.DETAILED DESCRIPTION

[0035] Referring to Fig. 1 , there is described herein an equipment or more specifically a system 100 that enables the collection of returnable containers after their consumption. The system 100 addresses individual identification and counting of the collected returnable containers, in a setting in which the returnable containers are dropped in bulk. The possibility of users to drop their returnable containers in bulk makes the process much less timeconsuming for them. In order to allow for an individual identification and counting of the collected returnable containers despite them being dropped in bulk by the user, a conditioning system (in the receiving station 200) is provided by the entry point of the system 100, as described further below. The entire bag of mixed returnable materials (cans, plastic bottles, cardboard and glass containers) can be poured into the machine and processed in bulk, for example in a large batch. For example, an individual (consumer) may drop their whole bag full of returnable containers (mixed, of variable nature and size) directly into the system 100, instead of introducing them individually and sequentially, one by one in a specific orientation as is typically the case in most prior-art machines. The conditioning station such as the one disclosed herein below makes the eventual counting and material / monetary value identification / classification, which is performed downstream, more accurate thanks to a continuous spread or distribution of the received containers, both over time and over a width of the conveyor.

[0036] According to the present disclosure, the system 100 ensures that once dropped or poured in bulk, the “materials”, that is the containers, are coarsely sieved andthen distributed while being fed onto the conveyor 115 semi-automatically. They then pass under the detector, which determines the category of the container which, macroscopically speaking, is mainly the type of material and its monetary value, in the case of a deposit return scheme. Finally, the material is roughly separated into heavy and light fractions to facilitate post-processing (separation by category, compaction, transport optimization, etc.). The equipment is designed to be operated predominantly by the consumer instead of an industrial operator, requiring little action on the part of the return center operator. Nevertheless, the latter can filter the content of material inputs and intervene if necessary by stopping the machine via their operating station.

[0037] In more detail, and according to a non-limiting, exemplary embodiment of the disclosure, the consumer who operates the system 100 can either press a start button or scan a QR code via a mobile app to start the transaction and start the operation of the system 100. The transport conveyor 115 starts up, and the machine can indicate its status to the user via an on-screen user interface. For example, the start button may light up as green. The consumer can then drop the whole of these mixed containers in bulk on the material reception panel 110 acting as the system entry area. The material reception panel 110 may be laterally enclosed between two opposed side walls. The containers may be of varied nature and size and mixed together, dropped all at once together simultaneously and on the same surface area of the material reception panel 110. By having the material reception panel 110 inclined with an inclination oriented downwardly away from the consumer toward the conveyor 115, containers migrate mostly on their own to the transport conveyor 115, by gravity or by being pushed by the weight of the last containers, but may require action by the consumer to push the last containers of the batch onto the transport conveyor. The containers on the conveyor 115 then pass under the reading station or scanning head 120, downstream of the receiving station 200, which analyzes and characterizes them. The reading station or scanning head 120 should comprise a camera or the like to capture images of the conveyor for any one of image recognition, Al-based or neural-network based identification and counting of the containers thereon; it may otherwise include other features, such as barcode reading. Once the containers have passed through this detection zone having the reading station or scanning head 120, the user sees the countdown in real time,as well as the corresponding monetary amounts allocated on the user interface. Once all the containers have been processed, the user can press the end button to conclude the transaction and view a summary of his deposit (batch). The equipment then prints a coupon summarizing the transaction, or can transfer the data to a third-party electronic transaction processor. At the same time, the equipment may coarsely separate the material into heavy and light fractions. Once the stop button has been pressed, the equipment empties before stopping the conveyor and separation equipment. The machine then becomes available for the next customer.

[0038] Technical data and implementation details - The equipment of the present disclosure can be characterized by the following different sections / modules.

[0039] Material receiving module - According to a preferable embodiment, and referring to Fig. 2, the receiving station 200 is the only section accessible to the general public, the rest being behind a wall on the operator’s side. It can handle large quantities of mixed containers dropped in bulk. Advantageously, it comprises a portion which serves as a conditioning station to condition the bulk of material (the returned containers) in a form which is suitable for subsequent individual counting and identification (for individual value assignment) despite the bulk feeding, i.e., by spreading or distributing the flow of containers fed in bulk over time and over a width of the system 100 (that is, over the width of the conveyor 115 on which counting / identification eventually takes place further downstream in order to make that step more accurate). The distribution over time and over the width of the conveyor decreases the number and significance of superimpositions of containers in the feed. This means that downstream of the conditioning station, the containers are much less intertwined or superimposed and much easier to count and identify. The reading station may then count and identify the objects in batch (but more evenly spread over time and width), including a plurality of containers or objects within the same image / frame from the video captured by the camera on the conveyor. This means that the operator does not need to insert the containers one by one, sequentially, as is the case with a traditional container return machine. Instead, the material is first coarsely sieved with the sieve 112 on the material reception panel 110 to remove small debris such as broken glass, corks, a portion of liquids and other foreign bodies. This reduces the amount of breakage, maintenance andcleaning required of the machine. These foreign bodies fall into a bin beneath the receiving area such that they are filtered out of the system. The bin can later be emptied by the center operator. The system may comprise cylindrical 1 ’’-diameter plastic shafts, spaced apart from between about 2.0” to about 3.0”, preferably about 2.5” apart. These dimensions are precisely selected to allow small, low-volume containers to pass through, while ensuring that the majority of waste and other contaminants are collected. The round plastic bars limit material friction and enhance the self-feeding effect described below.

[0040] Still referring to Fig. 2, the receiving station 200 comprising the material reception panel 110, which is preferably inclined to form a downward slope toward the downstream conveyor 115, is mostly self-feeding, meaning that the material is transferred directly to the transport conveyor 115 with little human interaction, as mentioned above. This limits the flow of containers passing under the analysis module, thus increasing the accuracy of detection. The latter must see each container without overlapping, and without any movement relative to the conveyor’s belt. The advantages of this method are to increase detection efficiency and improve counting accuracy.

[0041] According to an alternative embodiment of the disclosure, the material reception panel 110 may not comprise any sieve 112.

[0042] According to another alternative embodiment of the disclosure, the receiving station 200 may not comprise any material reception panel 110 and also therefore does not comprise any sieve 112, such that the receiving station 200 is optional for such embodiments of the disclosure.

[0043] According to a preferred embodiment, to ensure that the materials (containers) are evenly distributed and fed to the conveyor 115, a regulation or distribution system 200, also referred to as a conditioning station, is provided at or by the transition between the material reception panel 110 and the conveyor 115 to condition the objects dropped in bulk into a more streamlined or distributed flow of containers, i.e., more distributed over time and also more distributed over the width of the conveyor 115. According to an embodiment, a series of obstacles with an appropriate arrangement can be provided in the receiving station 200 to perform this conditioning. According to an exemplary embodiment, this is made up ofsuspended chains, staggered and distributed in different ways, preferably, without limitation, in staggered rows (especially for the first and second rows). Containers must therefore pass through these chains 210, which offer resistance, thereby regulating the flow of containers from the bulk being dropped simultaneously on the material reception panel 110 to the conveyor 115, where the flow should be more distributed over time. The chains 210 also allow the containers to tumble, reducing the number of containers stacked on top of each other. Unlike baskets and the like which can be found in the prior art, this conditioning station can advantageously distribute the containers continuously over time and not simply in smaller batches which may still have the same issue of low counting accuracy.

[0044] Now referring to a preferred but non-limiting embodiment of Fig. 3, this system with the chains 210 may advantageously comprises 3 rows of chains with 4" and 12" spacing between each row respectively (non-limiting, exemplary embodiment). The chains may be 1 / 4"-diameter general-purpose round chains, thus increasing their linear weight. This feature helps to regulate the flow of material while offering increased strength. The first two rows taken in combination are staggered with a constant spacing of 3" between them (in other words, the first two rows are offset from each other), while the 3rd row is straight. The chains are modular and adjustable to ensure even material flow. While suspended, their bottom end may end at different heights above the conveyor. According to an exemplary, non-limiting, but preferred embodiment, every second chain on the last row is two chain links shorter to distribute the material more evenly. For the front, one chain out of three is two chain links shorter. Longest chains have their bottom end close to the conveyor’s belt surface, the height or distance in between being less than the length of a chain link. The material receiving station can handle over 300 items in less than 30 seconds.

[0045] The chains of the illustrated embodiments may be replaced by other types of elongated obstacles, including, in addition to chains: rods, bars, strings, cords, slats, blinds, rubber strips, etc. such as elongated materials made of a rigid material or a resilient material. The elongated obstacles (210) are therefore provided in rows, having for example at least a first row and a second row being offset (meaning staggered; the elongated obstacles are offset from the corresponding elongated obstacles of the second row) and a third row aligned with the second row (meaning that the elongated obstacles are aligned with (and not offsetfrom) the corresponding elongated obstacles of the second row). The elongated obstacles 210 are preferably suspended from the top and free at the bottom end thereof, and the distance between the bottom end thereof and the conveyor’s belt surface can be small but variable between different elongated obstacles of the same row. A row should be interpreted as a series of elongated obstacles laterally spaced apart in line, at a given distance in the direction of the conveyor, e.g., an object moving on the conveyor will sequentially hit consecutive rows as it moves on the conveyor.

[0046] In each row, some of the elongated obstacles may have a length, or gap between the bottom end thereof and the conveyor underneath, which is different from others of the elongated obstacles of the same row.

[0047] With these elongated obstacles 210 such as the chains as described above and arranged as described, a proper conditioning of the inputted containers dropped in bulk can be performed to distribute the containers more evenly over time and over the width of the conveyor, to facilitate their eventual individual identification (for monetary value) and counting.

[0048] Finally, the material receiving station 200 can also accommodate the user interface (III) via a screen in front of the user, for example on a dedicated base as shown in Fig. 2. Preferably, the screen is not used for user inputs, rather for display only, to simplify operation and reduce production costs. On the right-hand side of the receiving station are the stop and start buttons, the 2D code scanner and the receipt printer.

[0049] Material transport conveyor - Now referring to Figs. 4 and 5, the conveyor belt 115 can be made of stainless steel to protect it from corrosive liquids and to enable easy cleaning. The belt is of the plastic mesh type, commonly found in the food industry. This allows it to be driven by sprockets, thus eliminating false starts due to sticky liquids, a typical and common problem for rubber belt conveyors. These belts are easy to wash, to replace and to maintain. According to an exemplary embodiment of the disclosure, there is a distance of between about 40” and about 80”, preferably about 60", between the distribution system and the analysis module to allow sufficient time for the material to stabilize. The belt could reach a maximum speed of 200 FPM while still allowing proper conditioning and faithfulcounting and individual identification of monetary value. This speed ensures better distribution of the containers, while providing a certain amount of ballistic projection for the separation section. To prevent the containers from rolling or moving during transport, which can influence detection, the belt is a mix of smooth and textured slats. The textured slats are spaced 12in apart and are approximately 1 in high.

[0050] Container analysis module - The analysis module incorporates a camera coupled with lighting which could comprise LED lights. It reproduces a controlled, reproducible environment to maximize container detection. A "Pan-Tilt" system was developed to position the camera and lights in relation to the belt. Camera and lighting adjustments are independent. This system makes it possible to adjust the appearance and brightness of the objects to be recognized.

[0051] This system is typical of residual material sorting systems including recycling centers. An exemplary embodiment of such a system is shown in Fig. 6.

[0052] Detection via Al - In order to count the number of returnable containers or, more generally, objects, proper object classification needs to be implemented, especially to ensure that a monetary value or other type of individual tag can be placed on each single object being transported and scanned. According to an embodiment, the object classification as described herein can be based on deep learning, a powerful identification technique for sorting and recognizing different types of recyclable objects, from a video captured by a camera or a similar data source comprising images over time. It involves the use of artificial neural networks to learn patterns in object images and classify them into various categories based on their characteristics. In the case of deposit materials, these categories vary according to their composition (such as plastic, glass, multi-layer, metal etc.), their volume and the associated return value.

[0053] Providing an object classification system based on deep learning begins with the collection of a large dataset of images of the objects to be classified. These images must be representative of the different categories, and must be annotated with labels indicating the category to which each object belongs. The size of the dataset will depend on the complexity of the problem and the number of categories to be classified. This collection iscarried out automatically using camera capture, and the images are centralized on cloud storage.

[0054] Once the dataset has been collected, it can be used to train a deep neural network to classify objects. According to an embodiment, the neural network comprises multiple layers of interconnected neurons that process the image data and extract features relevant to classification. The first layer of the network processes raw pixel data from the images (frames) of the video captured by the camera, and subsequent layers extract increasingly abstract features.

[0055] After training, the performance of the neural network can be assessed on a separate set of images. The accuracy of the network will depend on factors such as the quality of the dataset, the network architecture and the complexity of the problem.

[0056] Material Sorting - Once the network has been trained and tested, it will be used to classify images of materials on a conveyor’s belt. The system takes an image containing several objects as input and produces a probability distribution over the different set categories. The category with the highest probability is taken as the classification for the various objects. Recyclable materials transported on the conveyor, for example the returnable containers, will be counted according to their various categories so that the user can be monetarily rewarded correctly.

[0057] Fig. 9 is a schematic diagram illustrating the material separation system 700 comprising a sorting station in the deposit return machine, in particular with an enlarged portion showing the sorting station, according to an embodiment of the disclosure. The sorting station may comprise a lamp for illuminating the contents on the conveyor properly, for a more accurate and constant evaluation of the contains. A camera takes a video of the contents and sends it to a computer for processing. The computer may segment all objects, evaluate boundaries and categorize materials, for example. The sorting station may then comprise a blower for ejecting materials corresponding to a given type (e.g., material). The sorting station may comprise a plurality of similar sequential sorting stations for ejecting a specific material from each station, the overall sequence treating all materials. Otherwise, the sorting station 700 (or one of many sequential sorting stations forming the materialseparation system 700) may comprise a plurality of nozzles which can be individually actuated by the controller in communication with the computer for ejecting specific items with a greater spatial accuracy and, if desired, to different locations for each item.

[0058] Material separation system - Now referring to Fig. 7, and according to a nonlimiting and exemplary embodiment of the system described herein, a material separation system 700 can be included to allow segregation into heavy and light fractions. Since the two types of material (heavy and light) have a large difference in surface density, aerodynamic separation is well suited to this application. The material separation 700 can comprises three sections:- A blower with its adjustable nozzle to perform selective ejection;- Ejection and separation box with recirculation system; and- An optional top brush 800, described further below.

[0059] According to a preferred embodiment, to achieve a very high separation efficiency, a continuous jet of air is blown under the conveyor’s head pulley. This causes lighter containers to float to the bottom section of the ejection box, while heavier containers such as glass fall gravitationally via a ballistic projection into the first section. The nozzle is designed to increase air velocity while minimizing the required blower output. The slot opening is about 1 / 2", and is adjustable. The rest of the nozzle is adjustable in height and rotation to maximize the material’s flotation effect. Aerodynamic pressure will be greatest closest to the material, without the material falling into it. Most of the air leaving the pulley should be vertical. Blower flow is around 3000 CFM. Power requirement is 5HP.

[0060] According to a non-limiting and exemplary embodiment of the system described herein, to limit the size of the blower, a brush 800, shown in Fig. 8A, can be provided available to increase material projection, since the conveyor is limited in speed. This brush 800 propels lighter containers, while heavier ones are less affected. It comprises a shaft 810 with relatively soft plastic finger paddles 850, shown in Fig. 8B. Brush rotation speed is between about 40 rpm and about 80 rpm, preferably about 60 RPM. A speed which would be too fast would imply undesired projection of heavy materials, and if too slow, thenthe desired effect of projection for separation would be lessened. The fingers 855 are between about 9” and about 12” long, between about 7 / 32" and about 11 / 32” wide, and between about 0.5” and about 0.625" apart and extends from the paddle base 852. As shown in Fig. 8A, taken in combination with Fig. 8B, and according to a non-limiting and exemplary embodiment of the system described herein, there are 6 paddles 850 on a shaft mounting base 815.

[0061] The main functions of the ejector box are to contain the ejected material, to provide a clearer separation of the two fractions, and to contain the blower air. The separator is adjustable in position in the direction of the conveyor and in rotation across the conveyor. This makes it possible to find the material’s natural separation zone / trajectory. The blower therefore has less work to do on light material. The ejector box can be designed to accommodate a 48” nominal conveyor separated longitudinally into two conveying zones. To reduce air emissions in the operator’s working environment, the ejector box recovers a portion of the blower air and sends it to the blower inlet. In this way, less air treatment by an external system (such as a cyclone separator or dust collector) is required. Nevertheless, these systems may be required, but will be smaller in size should they be provided.

[0062] According to an embodiment, the system 100 may additionally comprise a validation apparatus to provide redundancy. For example, a robot, such as an articulated robot, including a device to perform individual and sequential barcode reading on all or on a fraction of the objects travelling through the system, may be provided downstream of the reading station 200 in order to re-validate the identifications performed by the system 100 described above and using neural networks for classification and counting, thereby providing redundancy to assist in obtaining high-accuracy results, to identify any misclassifications and using the data to improve the training of the neural network over the duration of operation.

[0063] According to an embodiment, the system 100 may additionally comprise a sorting station downstream of the reading station 200. For example, a robot, such as an articulated robot, or a pressurized nozzle may be used to sort away specific containers following the step of counting and identification, thereby ejecting all or on a fraction of theobjects travelling through the system. For example, containers may be differentially sorted based on the identified monetary value associated thereto.

[0064] Advantages with respect to prior art - One of the main advantages of this equipment is its ability to handle in-bulk feeding of the materials. Traditionally, reverse vending machines have had to order and align individual containers and process them one at a time, sequentially, not permitting batch treatment. The system 100 as described herein permits batch processing and speeds up processing time for the consumer, while decoupling processing and payment operations. Fractional separation, compaction and post-processing are not dependent on customer transaction processing time as they can be done downstream of the conditioning system which ensures that even though containers are dropped in batch, they can be conditioned for a more even distribution over time and over width of the conveyor to allow for a more accurate while efficient individual counting and classification in real-time of the containers. Similarly, the consumer does not have to wait for the material to be post-processed before completing the transaction. For the consumer, processing time is therefore greatly reduced. Since the equipment processes containers in bulk and not one by one, the specific shapes, materials and characteristics of the containers have little influence on its processing capacity. In comparison, typical equipment such as reverse vending machine has great difficulty in processing rectangular or complex-shaped containers. Similarly, these machines can process more than 300 items per minute, but are often limited to cans to avoid any classification task (all counted containers have the same value) to be able to reach this counting frequency. The equipment proposed here can handle every type, or at least a variety of types of container, thereby providing not only individual container counting, but also classification to attribute a monetary value or other tag individually, in real time.

[0065] The equipment can be used in a stand-alone setting, without the need for large- scale infrastructure. Thanks to its integrated separation system, the equipment can receive containers, count them, pay the customer and, downstream of the counting part, it can coarse-separate material without requiring additional investment or infrastructure. Its footprint is also minimized. The machine can be used directly by the consumer, reducing the interaction required by the operator, who becomes available for other tasks in the returncenter. Screening out small pieces of waste and foreign bodies also reduces the need for cleaning, maintenance and general upkeep of the machine. The machine can be delivered semi-assembled to reduce set-up time.

[0066] Container categorization by Al is different from barcode reading. This technology can recognize containers with damaged or non-existent barcodes. This can limit “No-Reads” and unknowns, increasing overall system efficiency by reducing container postprocessing.

[0067] The equipment can operate continuously at high capacity. This limits queuing and the need to invest in more equipment to handle waiting customers. The passive material distribution system makes it possible to process large quantities of containers while minimizing investment in an active, automated system.

[0068] Software and IA detection - This provides improved accuracy. Indeed, one of the main advantages of neural networks over barcode scanning is their ability to improve accuracy. Neural networks can learn and identify complex patterns in data, enabling them to make accurate predictions and classifications. On the other hand, barcode scans are not always accurate, due to problems such as damaged or illegible barcodes and new products not in the database.

[0069] Flexibility: neural networks are highly flexible and can be trained on a wide range of data sets, making them ideal for item classification tasks. They can also be trained to identify items based on characteristics such as shape, color and texture, which can be useful when barcodes are unavailable or unreliable. In contrast, barcode scans are limited to the information encoded in the barcode, and may not capture all the relevant information about an item. This generalization will be put to good use by starting with a solid learning base when attacking a new market.

[0070] Scalability: neural networks can be scaled up to handle large volumes of data, making them ideal for item classification tasks requiring the processing of large numbers of items simultaneously. Conversely, barcode scanners may not be as scalable, and may require additional processing time to handle large volumes of items.

[0071] According to an embodiment of the disclosure, a plurality of distinct neural networks may be implemented, wherein at least one of the plurality of distinct neural networks is dedicated to identifying an object, and at least another one of the plurality of distinct neural networks is dedicated to identifying a material of said object. According to an embodiment of the disclosure, at least another one of the plurality of distinct neural networks is dedicated to read a barcode or otherwise identify any other identification tag of the container. According to an embodiment of the disclosure, a mechanical conditioning of a plurality of containers or objects can be performed to orient individually and detect the barcode (or identification tag) for a plurality of containers and maintain the orientation of a plurality of nearby containers for bulk reading of the barcodes (or identification tags) .

[0072] According to an embodiment of the disclosure, if the material separation system 700 comprises a plurality of sorting stations each dedicated to sort out one type of object, one neural network is used for each respective one of a plurality of sorting stations as described above, downstream of the conditioning station, for categorizing and ejecting a specific type of object for each respective one of the plurality of sorting stations.

[0073] Contaminant recognition - There are certain issues specific to the fact that the equipment is self-contained. Since the equipment will be usable by a customer, it is imperative that the recognition system be able to distinguish contaminants (non-returnable items) that could be inserted by uninformed or malicious users.

[0074] Application differences versus conventional sorting centers - An important point differentiating the operation of this system from that of conventional sorting centers is the material processed. In a conventional sorting center, the material may be composed of any one of: paper, commercial cardboard, plastics of all kinds (HDPE, PET, PP, film, etc.), aluminum, iron and other metals (not just cans), glass of all kinds, construction and demolition materials (some systems), organic materials (some systems) and other materials. These large-scale facilities process several dozens tons per hour, separating materials by type of material rather than by type of container. These systems are often highly automated, with industrial separation and compaction equipment such as baling presses for the differentfractions. The general public is not generally admitted to these centers, which are operated entirely by qualified personnel.

[0075] In contrast, the system 100 described herein may advantageously be provided in the contexts described below.

[0076] Deposit return center: Return systems for returnable containers deal with specific items. Conventionally, the materials processed are aluminum cans and PET bottles. They are being modernized to accommodate other materials such as glass bottles and cardboard containers. In most cases, these items are limited to beverage and ready-to-drink containers of certain volumes. Containers such as jars or general recycling are not accepted. These return centers are often smaller and less automated. They process few tons of material per hour, and their output is often measured in terms of the number of containers processed annually. Typical materials accepted and refunded are as follows: aluminum cans (various sizes); PET bottles (various sizes); cardboard containers (TetraPak, multilayer, etc.); glass containers (beer, wine, spirits, etc.), and other uncommon beverage containers. Accepted containers are generally beverages and soft drinks, refusing certain substitutes, medical drinks, etc. Generally, only containers where the deposit has been paid will be accepted and refunded. These return centers are operated by fewer employees than conventional sorting centers, and a portion is done by the consumer via return machines (reverse vending machine). The collection equipment in these centers must therefore be adapted so that a portion is used by the general public. These facilities are often more commercial than industrial.

[0077] Overview of applications and competition - The equipment in question is versatile and can be operated in many different environments.

[0078] Deposit center - Deposit centers are often made up of several pieces of equipment or methods of deposit collection, such as a reverse vending machine and a service counter. These centers can collect material from a number of retailers, kiosks and other return points. The material can be separated into heavy and light streams to optimize transport to the packaging centers where it undergoes final processing. Bulk collection equipment, as disclosed herein, could replace a few of these reverse vending machines andreduce the service counter’s workload. The deposit center would thereby save on its investment and operating costs. Drop-off or deposit centers are open to the public for collection and deposit payment. It’s a mix of operation and collection.

[0079] Kiosks - Kiosks are smaller return points attached to a retailer, reducing the floor space required for deposit return. These return points are mainly made up of semi- autonomous beakers requiring recurrent cleaning and continuous emptying of material bins. The initial investment is lower, but operating costs are often higher. Here again, a high- volume bulk solution could replace several tumblers, reducing space requirements and operating costs. There is little processing of materials at the kiosk, and everything would be transshipped to the depot or packaging center.

[0080] Conditioning centers - Conditioning centers mainly collect material from other deposit return points and post-process it. These medium-sized facilities are not generally available to the general public. Although they are similar to waste sorting centers, they handle far less volume than the latter. These centers are often equipped with equipment for purifying the material, conditioning it for recycling and compacting it (baling presses). They are, however, more specific or specialized than conventional sorting centers and not adapted for commercial reverse vending machines and the like.

[0081] The Bottle Drop - OBRC is another system from the prior art from which the system as described herein differs. This prior-art system requires the operator to be present at all times to move the material under its reading system. We can see the operator pushing the material onto the conveyor.

[0082] Corral Al - APC-1 is another system from the prior art from which the system as described herein differs. This prior-art system comprises a machine uses Al to detect returnable containers. It does not offer an in-machine separation system and requires an inclined conveyor to being upwards containers that fit the recesses of that inclined conveyor, not allowing any container to be introduced therein.

[0083] While preferred embodiments have been described above and illustrated in the accompanying drawings, it will be evident to those skilled in the art that modifications maybe made without departing from this disclosure. Such modifications are considered as possible variants comprised in the scope of the disclosure.

Claims

CLAIMS:1 . A deposit return machine for receiving containers, the machine comprising:- a conveyor for receiving the containers in bulk;- a conditioning station which distributes the received containers over time and over a width of the conveyor; and- a reading station downstream of the receiving station for counting the containers on the conveyor.

2. The deposit return machine of claim 1 , wherein the reading station operates a neural network for identification of each container being counted for associating a monetary value thereto.

3. The deposit return machine of claim 2, wherein the neural network for identification of each container being counted comprises a plurality of distinct neural networks, at least one of the plurality of distinct neural networks being dedicated to detecting object presence, at least one of the plurality of distinct neural networks being dedicated to segmenting said object presence as a plurality of distinct objects, and at least another one of the plurality of distinct neural networks being dedicated to classify said plurality of distinct objects individually.

4. The deposit return machine of claim 1 , further comprising a receiving station comprising a reception panel which is inclined downwardly toward the conveyor for receiving the containers in bulk and perform sieving before moving the containers in bulk to the conveyor .

5. The deposit return machine of claim 1 , wherein the conditioning station comprises a set of elongated obstacles, the elongated obstacles being suspended and free at a bottom end thereof, and arranged to form rows where at least two consecutive rows are staggered to have laterally offset, for distribution of the containers introduced in the reception panel.

6. The deposit return machine of claim 5, wherein the set of elongated obstacles comprises at least two consecutive rows aligned together.

7. The deposit return machine of claim 5 or 6, wherein the set of elongated obstacles comprises elongated obstacles ending at variable heights above the conveyor within a same one of the rows.

8. The deposit return machine of claim 7, wherein elongated obstacles comprise chains.

9. The deposit return machine of any one of claims 1 to 8, further comprising a material separation system, downstream of the conveyor, comprising a brush to propel lighter containers into an ejection box.

10. The deposit return machine of claim 9, wherein the material separation system further comprises a blower feeding the ejection box to propel the lighter containers thereinto.11 . The deposit return machine of any of claims 1 to 10, further comprising an apparatus downstream of the reading station and performing an individual and sequential reading of a container identification tag of the containers to provide redundancy.

12. The deposit return machine of any of claims 1 to 10, further comprising an apparatus downstream of the reading station and performing a bulk reading of a barcode of a batch of containers to provide redundancy.

13. The deposit return machine of claim 11 or 12, wherein the apparatus comprises lighting to expose containers to a preset illumination, and the apparatus implements an artificial-intelligence based algorithm to perform the reading of the barcode.

14. The deposit return machine of any of claims 1 to 13, further comprising a sorting station downstream of the reading station to sort the containers following said counting.

15. The deposit return machine of claim 14, wherein the sorting station comprises a blower for ejection.

16. The deposit return machine of claim 14, wherein the sorting station comprises a plurality of distinct and interspaced nozzles which can be individually controlled for individualized ejection of containers passing by a given one of the distinct and interspaced nozzles.

17. The deposit return machine of claim 15 or 16, wherein the sorting station comprises a plurality of distinct and sequential sorting stations each for ejecting containers of a respective type.

Images