Methods and arrangements to aid recycling

Patent metadata

Field	Value
Publication number	US-12496618-B2
Application number	US-202217721694-A
Country	US
Kind code	B2
Filing date	Apr 15, 2022
Priority date	Apr 16, 2021
Publication date	Dec 16, 2025
Grant date	Dec 16, 2025

Abstract

A waste stream is analyzed and sorted to segregate different items for recycling. Certain features of the technology improve the accuracy with which waste stream items are diverted to collection repositories. Other features concern adaptation of neural networks in accordance with context information sensed from the waste. Still other features serve to automate and simplify maintenance of machine vision systems used in waste sorting. Yet other aspects of the technology concern marking 2D machine readable code data on items having complex surfaces (e.g., food containers with integral ribbing for structural strength or juice pooling), to mitigate issues that such surfaces can introduce in code reading. Still other aspects of the technology concern prioritizing certain blocks of conveyor belt imagery for analysis. Yet other aspects of the technology concern joint use of near infrared spectroscopy, artificial intelligence, digital watermarking, and/or other techniques, for waste sorting. A variety of further features and arrangements are also detailed.

First claim

The invention claimed is:

1 . A method comprising the acts:
determining first attribute information for waste at a first location on a waste-conveying conveyor belt, in which the first attribute information comprises a digital watermark payload or a material type or a food/non-food type; providing imagery depicting said first location to a convolutional neural network, and receiving an output from the convolutional neural network indicating presence of only one waste item; controlling a diverter to act on said waste item based on the determined first attribute information and on output from the convolutional neural network indicating presence of only one waste item; determining attribute information for waste at a second location on the waste-conveying conveyor belt; providing imagery depicting said second location to the convolutional neural network, and receiving an output from the convolutional neural network indicating presence of two or more adjoining or overlapping items; and not controlling a diverter to act on waste at said second location.

2 . The method of claim 1 that includes acts of:
determining a first contiguous area around said first location that is occupied by waste;
providing imagery depicting said first contiguous area to the convolutional neural network, and receiving an output from the convolutional neural network indicating that said first contiguous area is occupied by only one waste item;
controlling the diverter to act on a diversion target within said first contiguous area, to direct said waste item to a repository associated with said determined attribute information;
determining a second contiguous area around said second location that is occupied by waste;
providing imagery depicting said second contiguous area to the convolutional neural network, and receiving an output from the convolutional neural network indicating that said second contiguous area is occupied by more than one waste item; and
not controlling a diverter to act on a diversion target within said second contiguous area.

3 . The method of claim 1 that includes determining said attribute information based on near infrared spectroscopy data.

4 . The method of claim 1 in which said imagery comprises 3D imagery or depth map information.

5 . The method of claim 1 in which said attribute information indicates for the waste at the first or second location comprises a plastic container used for food.

6 . The method of claim 1 further comprising:
determining a first contiguous area around said first location that is occupied by waste by utilizing a region growing algorithm;
providing imagery depicting said first contiguous area to the convolutional neural network, and receiving an output from the convolutional neural network indicating that said first contiguous area is occupied by only one waste item; and
controlling the diverter to act on a diversion target within said first contiguous area, to direct said waste item to a repository associated with said determined first attribute information.

7 . The method of claim 6 further comprising:
determining a second contiguous area around said second location that is occupied by waste by utilizing a region growing algorithm;
providing imagery depicting said second contiguous area to the convolutional neural network, and receiving an output from the convolutional neural network indicating that said second contiguous area is occupied by more than one waste item; and
not controlling a diverter to act on a diversion target within said second contiguous area.

8 . The method of claim 6 further comprising:
restricting said determining a first contiguous area upon receiving an output from the convolutional neural network indicating presence of only one waste item and upon receiving the determined first attribute information indicating that the waste comprises a single plastic material type.

9 . The method of claim 6 in which the region growing algorithm utilizes blob extraction, connected-component labeling, or connected component analysis.

10 . A method comprising the acts:
compiling historical conveyor belt map data derived from images depicting a conveyor belt loop at positions throughout a full cycle of conveyor belt travel without waste thereon; after compiling said historical conveyor belt map data, capturing first imagery depicting a first region of the conveyor belt with waste thereon; by comparison with the historical conveyor belt map data, identifying a first set of conveyor belt area blocks depicted in the first imagery in which the conveyor belt is visible and a second set of conveyor belt area blocks depicted in the first imagery in which the conveyor belt is not visible, said second set of area blocks including a first clump of adjoining area blocks; providing imagery depicting said first clump of adjoining conveyor belt area blocks to a convolutional neural network, and receiving an output from the convolutional neural network indicating that said first clump of adjoining area blocks is occupied by a single waste item only; controlling a diverter mechanism to act on a diversion target within said first clump of adjoining conveyor belt area blocks, to remove said single waste item to a repository; after compiling said historical conveyor belt map data, capturing second imagery depicting a second region of the conveyor belt with waste thereon; by comparison with the historical conveyor belt map data, identifying a first set of conveyor belt area blocks depicted in the second imagery in which the conveyor belt is visible and a second set of conveyor belt area blocks depicted in the second imagery in which the conveyor belt is not visible, said second set of area blocks including a second clump of adjoining area blocks; providing imagery depicting said second clump of adjoining conveyor belt area blocks to the convolutional neural network, and receiving an output from the convolutional neural network indicating that said second clump of adjoining area blocks is occupied by more than one waste item; and not controlling a diverter mechanism to act on a diversion target within said second clump of adjoining area blocks.

11 . A method comprising:
compiling historical conveyor belt map data derived from images depicting a conveyor belt loop at positions throughout a full cycle of conveyor belt travel without waste thereon; after compiling said historical conveyor belt map data, capturing first imagery depicting a first region of the conveyor belt with waste thereon; by comparison with the historical conveyor belt map data, identifying a first set of conveyor belt area blocks depicted in the first imagery in which the conveyor belt is visible and a second set of conveyor belt area blocks depicted in the first imagery in which the conveyor belt is not visible, said second set of area blocks including a first clump of adjoining area blocks; providing imagery depicting said first clump of adjoining conveyor belt area blocks to a convolutional neural network, and receiving an output from the convolutional neural network indicating that said first clump of adjoining area blocks is occupied by a single waste item only; and controlling a diverter mechanism to act on a diversion target within said first clump of adjoining conveyor belt area blocks, to remove the single waste item to a repository.

12 . The method of claim 11 further comprising:
after compiling said historical conveyor belt map data, capturing second imagery depicting a second region of the conveyor belt with waste thereon;
by comparison with the historical conveyor belt map data, identifying a first set of conveyor be

Assignees

• Digimarc Corp

Inventors

• Filler Tomas
• Sharma Ravi K
• Rodriguez Tony F
• Alattar Osama M
• Alattar Adnan M
• Brunk Hugh L

Classifications

• G06F18/214

  Generating training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• B65G2203/041

  Camera · CPC title

• B65G2203/0208

  relating to the transported articles · CPC title

• B65G47/26

  arranging the articles, e.g. varying spacing between individual articles · CPC title

• B07C2501/0054

  Sorting of waste or refuse · CPC title