RU2796050C1 - Automatic device for receiving and sorting containers on the basis of pattern recognition system - Google Patents

Abstract

FIELD: sorting.

SUBSTANCE: systems for receiving and automatic sorting of containers for drinking beverages and can be used in places where bottles and cans are sold, as well as in places where drinks are consumed. The machine for receiving and sorting containers based on the image recognition system contains a body with an opening for placing containers, a receiving cell, two additional cells installed symmetrically relative to the axis of the body under the first receiving cell, four containers located in the lower part of the body, four ultrasonic filling sensors containers, a video camera, a light source, a microcontroller with a control program, which has a built-in neural network trained to recognize at least three types of containers and a GSM module.

EFFECT: sorting of containers into three fractions, availability of a container for unrecognized containers, availability of container capacity control, possibility of reconfiguring the machine to separate other fractions, simplicity of design.

2 cl, 3 dwg

Description

The device belongs to systems for receiving and automatic sorting of containers from drinking beverages and can be used in places where bottles and cans are sold, as well as in places where drinks are consumed.

A device for receiving recyclables is known (RU 183256, 03/27/2018, G07F 7/06). The device is designed for receiving and sorting packages and containers made of polymer, aluminum, glass, paper materials.

The device contains a receiving window, a movable damper with a drive, a recyclable weight sensor. The reception window is additionally equipped with a barcode scanner. The recyclable weight sensor is made in the form of a strain gauge mounted on a movable damper. The device for receiving recyclables is additionally equipped with a sorting device made in the form of containers and a movable guide equipped with a drive. Each container is mounted on a cantilever element equipped with a strain gauge weight sensor. The device is equipped with a printer and a computer. The computer is connected to the scanner, weight sensors, drives and printer.

Disadvantages of this utility model: only a container with a bar code is recognized, there is no signaling about the filled volume of the container.

A hardware-software complex for receiving polyethylene bottles and aluminum cans is known (RU 2727549, 12/11/2019, G07F 11/00). The invention relates to automated equipment for the collection of certain solid containers and is intended primarily for the collection of polyethylene bottles and aluminum cans. The hardware-software complex for receiving polyethylene bottles and aluminum cans contains at least one housing, in which a container recognition compartment is located, containing at least one video camera, a microcontroller configured to process primary data for container recognition and issuing commands to actuators on acceptance of containers and payment of remuneration or refusal in these operations, a compartment for storing containers and a means for issuing rewards, while a manual compactor is installed on the body, and the microcontroller is configured to recognize a compressed container moving along the tray, or is configured to communicate with a remote server, by at least for transmitting information to the server that allows recognizing the container, and receiving commands from the server to accept the container, or pay a reward, or to refuse these operations. The microcontroller may be configured to communicate with a remote server containing the databases and program code of the trained neural network responsible for analyzing images and metadata received from the microcontroller. The hardware-software complex may contain a user identification tool that allows non-cash payment for the accepted container through a remote server.

The disadvantages of this invention: the ability to sort only into two fractions, the need to communicate with a remote server, the absence of a container for unrecognized containers, the complexity of the design containing the conveyor belt, the lack of signaling about the filled volume of the container.

The closest analogue of this machine is a container receiving machine (RU 188755, 12/21/2018, G07F 11/00), containing a housing, a receiving device with a device placed in it that collects information about the recognized container, and a container recognition device - a microcontroller, an accumulation means containers and a means of issuing a reward, a video camera is used as a device that collects information about a recognizable object, and the microcontroller contains in memory a neural network algorithm that allows recognizing a container.

The disadvantages of the prototype are the ability to sort only into two specific fractions, the absence of a container for unrecognized containers, the inability to reconfigure the machine to recognize other fractions, and the lack of signaling about the filled volume of the container.

The objective of this invention was the development of an automatic machine for receiving and sorting containers into three fractions, the presence of a container for unrecognized containers, the presence of control over the filling of the volume of containers, the possibility of reconfiguring the machine to separate other fractions, and simplicity of design.

The task of the invention with the achievement of the technical result was solved due to the fact that in the well-known machine for receiving containers, containing a body, a receiving cell, a video camera, a microcontroller with a neural network algorithm, a light source, two additional receiving cells are located in the body, each cell has movable opening flaps , two cells are located symmetrically under the first cell, the shutters of the two lower cells serve as lids for four containers located at the bottom of the body, and one container is intended for unrecognized containers, each container has a volume occupancy sensor, the GSM module transmits a container full signal to the operating personnel, the neural network is trained to recognize three fractions and can be reconfigured to recognize three other selected fractions.

In FIG. 1 shows a diagram of the proposed machine for receiving containers, where: 1 - housing, 2 - receiving opening for placing containers, 3 - containers, 4 - receiving cell, 5 - additional receiving cells located symmetrically under the first receiving cell, 6 - cell shutters, 7 - containers for containers.

The machine for receiving and sorting containers based on the pattern recognition system contains a body with an opening for placing containers, a receiving cell, two additional cells installed symmetrically relative to the axis of the body under the first receiving cell, four containers located in the lower part of the body, four ultrasonic filling sensors containers, video camera, light source, microcontroller. The microcontroller contains a control program with a built-in neural network trained to recognize at least three types of containers and a GSM module. Each of the three cells consists of two flaps. Each leaf is located at an angle to the vertical walls of the body. The sashes are attached to the vertical wall of the housing with one side, the other side is in contact with each other, forming an angle. Each leaf is equipped with an electric drive for opening and closing.

The machine works as follows.

The recognizable object is placed through the receiving opening into the receiving cell, the recess formed by the flaps. The video camera photographs the object, the image is transmitted to the microcontroller. The neural network, using the pattern recognition algorithm, recognizes the object as belonging to one of the three predetermined fractions or as an unrecognized container. The microcontroller configures the path of the object to get it into the specified container. The microcontroller sends signals to the door leaf actuators to open the specified doors. An object under the influence of gravity falls into one of four containers.

In FIG. 2 and FIG. 3 shows the options for the operation of the machine in the direction of the object to a specific container.

Cell configuration can be changed at the control program level. To set the fractions to be determined, a trained neural network is loaded into the microcontroller.

Advantages: there is a sorting into four predefined any fractions, the possibility of increasing the number of containers and recognizable fractions, a container for unrecognized GSM garbage, a container full module.

Claims (2)

1. An automatic machine for receiving and sorting containers based on image recognition, containing a body, a receiving cell, a video camera, a microcontroller with a neural network algorithm, characterized in that a light source is placed in the body, two additional receiving cells located symmetrically relative to the axis of the body under the first receiving cell , four containers, one of which is intended for unrecognized containers, with volume fullness sensors located in the lower part of the body, and the flaps of two additional receiving cells serve as lids for four containers, and the microcontroller contains a control program in which a neural network is embedded, trained to recognize as at least three types of containers, and a GSM module for transmitting a signal about the fullness of the container. 2. The machine according to claim 1, characterized in that the neural network is reconfigurable to recognize other fractions.

Images