Computer vision-based automated sorting machine for peruvian mangos

The automated sorting machine for Peruvian mangoes addresses the limitations of existing technologies by employing computer vision and advanced mechanical components to efficiently classify and separate high-quality from damaged mangoes, improving production speed and reducing costs through precise variety and quality assessment.

WO2024248635A1PCT designated stage expired Publication Date: 2024-12-05UNIV CONTINENTAL S A C
View PDF 7 Cites 0 Cited by

Patent Information

Application Number
PCT/PE2023/050022
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2023-06-02
Filing Date
2023-12-20
Publication Date
2024-12-05

AI Technical Summary

Technical Problem

Current automated classification machines for fruits and vegetables are limited by their small size, inability to handle large quantities, lack of variety differentiation, and failure to assess quality criteria such as condition, leading to inefficient and slow classification processes that do not effectively separate damaged products from good ones.

Method used

An automated sorting machine for Peruvian mangoes using computer vision with gray scale image processing, featuring five conveyor belts, robotic arms with two degrees of freedom, diffuse sensors, and a control system that classifies three main varieties (Haden, Kent, and Tommy Atkins) and separates damaged mangoes, utilizing pneumatic and servomotor components for precise handling and quality assessment.

Benefits of technology

The machine efficiently classifies and separates good from damaged mangoes, improving production speed and reducing costs by accurately identifying and sorting three primary varieties and distinguishing between high-quality and poor-condition fruits, enhancing the overall efficiency and quality of the classification process.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure PE2023050022_05122024_PF_FP_ABST
    Figure PE2023050022_05122024_PF_FP_ABST
Patent Text Reader

Abstract

The present invention relates to a computer vision-based automated sorting machine for mangos that is made up of five conveyor belts (29), which are formed by a vertical infeed belt (29.1) and second (29.2), third (29.3) and fourth (29.4) belts that are horizontal. Furthermore, said belts are responsible for conveying the three varieties of Peruvian mangos and the fifth belt is the outfeed belt (29.5) on which the mangos in a poor condition are conveyed. Five HP motors (24) and three gear motors (23) are coupled to operate the machine, and the five conveyor belts (29) maintain uniform movement by means of ten rollers (22). With regard to the control system, a structure for the control box (19) is assembled and this in turn is assembled with a control box cover (20) for safety reasons. Furthermore, a servomotor support (7), which contains a servomotor (14), is assembled with the control box structure (19). Moreover, the machine has two pistons (3) and two rods (4). Each piston (3) has three suction cups (2) assembled therewith. Furthermore, the movement between components is possible due to the insertion of the servomotor gear (15) and suction cup gear (16). Lastly, the entire machine is fixed to a general structure (28).
Need to check novelty before this filing date? Find Prior Art

Description

[0001] AUTOMATED SORTING MACHINE FOR PERUVIAN MANGOES BASED ON COMPUTER VISION

[0002] TECHNICAL FIELD

[0003] The following invention is related to the agroindustry in general, but specifically focuses on the technical field of automated grading machines for Peruvian mangoes based on computer vision, which perform this action by processing grayscale images. The principles of Industry 4.0 were implemented in the development of this mechanism.

[0004] BACKGROUND OF THE INVENTION

[0005] Machines are currently known that perform the sorting of various elements; for example, patent document ES2318987B1 describes a portable fruit and vegetable sorting machine. This invention is small in size and therefore easy to transport. Therefore, to use the machine, a minimum volume of products must be entered for sorting. Furthermore, the parameters used for sorting are weight, size, and color. The machine is characterized by a circular structure that moves cups around a vertical axis, onto which the products to be sorted are received. These cups show different positions along the circular path. These are conditioned by a perimeter guide on which they rest and by established sorting criteria.The products to be sorted enter the machine; it also has a feeding system placed tangentially to the ring. Furthermore, when sorting fruits and vegetables, they are diverted through the corresponding outlet when the actuator is activated, closing the slots made in the guide ring. Patent ES2136621 T3 was also found, which mentions an apparatus and method for sorting objects. This device consists of an object sorting apparatus, which is subject to preselected weights and has visually verifiable criteria. Each object is separated and transported between adjacent pairs, in which the rollers are attached to a transport chain to rotate freely on a vertically sliding axis oriented perpendicular to the transport path.It also has a control surface along the conveyor belt, which rotates the roller assemblies and the object supported between them; thus, increasing the object's surface accessible for optical scanning. Mounted between each pair of roller assemblies is a loading truss. This loading truss lifts and captures the object from its supports between the roller assemblies. Each object is unloaded at an appropriate unloading location by tilting the unloading plate upward.

[0006] The first device mentioned presents the problem of structural dimensions, since only minimal inputs can be entered and large quantities cannot be introduced. Furthermore, there is no clear limitation on the varieties of fruits and vegetables that will be entered into the machine, and only a few of these products can be incorporated, making the classification of this invention slow and inefficient. Additionally, its classification parameters only consider weight, color, and size; however, the invention does not specify that it separates fruits and vegetables that are in poor condition, which is detrimental to inputs that are in good condition; therefore, it is concluded that food quality criteria were not evaluated.Regarding the second invention presented, regarding an apparatus and method for object classification, it shows a deficiency in the area of ​​optical scanning, since it does not detail which objects can be separated, and especially their dimensions. Likewise, regarding optical scanning, there is no connection between the fact that the machine already comes with preset information or the users themselves must add information for their object to be classified.

[0007] DESCRIPTION OF THE INVENTION

[0008] As a solution to the aforementioned problems, the present invention was developed: an automated grading machine for Peruvian mangoes based on computer vision. Regarding the mango varieties to be graded by the machine, three types were selected: Haden, Kent, and Tommy Atkins, which are the ones mostly exported from Peru to the European market due to their good quality. Despite the high demand for mangoes in the international market, Peruvian farmers have problems with the mango grading system, since this process is performed manually, which generates a delay in production and increases costs. Therefore, the parameters of these three mangoes were added to the system for accurate grading. The parameters are: first, the Haden mango, which has the uniqueness of a bright red color with green and yellow hues, and also usually has a round to oval shape.In second place is the Kent mango, recognized by its dark green color and oval shape. In third place is the Tommy Atkins mango, which is characterized by its small, oval shape. Finally, images of the diseases, deformations, and enzymatic browning of each of the aforementioned varieties were added so that they can be separated by a belt and not mixed with the mangoes in good condition. Therefore, through this information entered into the control system, the classification of the three mango varieties, as well as those in poor condition, can be automated.Its execution will be carried out by a display sensor which detects the mango and activates an electric piston that moves these fruits to a separate area by means of conveyor belts where the display sensor that is in operation detects the defective and damaged mangoes that consequently activates an electric piston that directs one towards a warehouse. The objective of this process is to separate the good mangoes from the damaged ones and in turn classify the three varieties of mangoes; therefore, the machine is coupled to five conveyor belts. On the other hand, there are two robotic arms with two degrees of freedom (DOF), which are responsible for classifying the mango varieties, these work by means of the size and color that refer to the three varieties of proposed mangoes, which are: Haden, Kent and Tommy Atkins respectively. Diffuse sensors were also considered, these have the function of detecting mangoes and through programming, counting each mango.Finally, the control system is shown, which will maintain direct communication with the user. A counter system was also implemented to monitor the number of mangoes classified and ready for export. The machine is supported by a general structure made of stainless steel; in addition, a structure was assembled to hold the control system. It also has a vertical input belt through which the mangoes enter and three horizontally assembled belts for the classification of mangoes in good condition and an output belt for the disposal of mangoes in poor condition. They also have a support for conveyor belt rollers.And the components are attached to stainless steel bolts, a support was also used for the servomotor and for optimal machine operation, three gear motors and five HP motors were added which disperse the compressed air through pneumatic hoses and control the flow through two pneumatic valves and four "T" shaped pneumatic valves. BRIEF DESCRIPTION OF THE FIGURES.

[0009] Figure 1: Signposted isometric view of each component used for the automated grading machine for Peruvian mangoes based on computer vision.

[0010] Figure 2: Orderly marked view of each conveyor belt.

[0011] Figure 3: Exploded isometric view of each component of the automated grading machine for Peruvian mangoes based on computer vision.

[0012] Figure 4: Detailed view of the suction cup support structure area.

[0013] PREFERRED EMBODIMENT OF THE INVENTION

[0014] An automated sorting machine for Peruvian mangoes based on computer vision comprises five conveyor belts (29): the first conveyor belt is for the general input of mangoes (29.1); the second (29.2), third (29.3) and fourth conveyor belt (29.4) is for the classification of the three varieties of mangoes and the fifth conveyor belt (29.5) is for the output of defective mangoes. These belts are supported on the sides by a conveyor belt structure (21). The machine has five conveyor belts (29); therefore, in total ten lateral structures of conveyor belts (21) were coupled and specifically two are coupled per conveyor belt (29).The conveyor belts (29) will be driven by five HP motors (24), which conduct the compressed air through pneumatic hoses (1 1 ); in addition, they are regulated by two pneumatic valves (9) and four pneumatic valves in the form of a "T" (10). Three reduction motors (23) were also used since these allow the speed to be reduced automatically. Likewise, the motor protector (25) is assembled and welded to the structure of the conveyor belt (21 ). The conveyor belts (29) maintain a uniform movement due to the roller (22) of which ten of these were coupled for the purposes of this machine and are located inside the roller support (8). Additionally, the counting sensor support (26) was welded to the conveyor belt structure (21), which contains the diffuse counting sensor (27) inside. There are four in total and these are embedded at the beginning of each conveyor belt and only the output belt (29.5) does not have it, since it is not necessary to count the handles in poor condition, but it is assembled to an electric thrust support (13) of the piston. The control system is subject to the structure of the control box (19) which is assembled to a control box cover (20) and for the safety of the system two pneumatic adjustment bolts (12) were used. Regarding the classification piece, it will be carried out by means of two pistons (3) which will intervene in the pressure of the compressed air and generate the vertical movement of the rod (4). In the lower part of the rod (4) a support structure for the suction cup (5) is assembled; in addition, it is subject to the suction cup support (1) and internally there is the suction cup (2), it is also subject to the thread structure (6) that allow the mobilization of the aforementioned elements and what allows the fluidity of these elements is the coupling of the suction gear (16).On the other hand, the servo motor support (7) is assembled to the structure of the control box (19) and internally has a servo motor (14) that allows the precise control of parameters such as angular position, acceleration and speed. These parameters are essential for the control system of the machine, which is coupled to a gear protector (17) and this in turn internally protects the servo motor gear (15). Finally, the entire machine is subject to a general fixed structure (28). Finally, the images incorporated into the control system were loaded in grayscale; that is, the images had two colors black and white, also these images in grayscale.

Claims

CLAIMS 1. An automated computer vision mango sorting machine according to claim 1 is characterized in that it comprises five conveyor belts (29), which are made up of an input belt (29.1) which is vertical, the second (29.2), third (29.3) and fourth (29.4) belt are horizontal; in addition, they are responsible for transporting the three varieties of Peruvian mangoes and the last belt is an output belt (29.5) through which the mangoes in poor condition are transported. For its operation, five HP motors (24) and three reduction motors (23) were coupled, also by means of ten rollers (22) the conveyor belt (29) will be able to maintain a uniform movement. Specifically, two rollers (22) will be coupled for each conveyor belt (29). In order to protect the control system, a structure for the control box (19) was assembled and, for safety measures, a control box cover (20) was assembled.On the other hand, a support for the servomotor (7) was assembled to the control box structure (19), which internally has a servomotor (14). On the other hand, the machine has two pistons (3) and two rods (4). Each piston (3) is assembled to three suction cups (2). In addition, what allows movement between components is due to the insertion of the servomotor gear (15) and the suction cup gear (16). Finally, the entire machine is fixed to a general structure (28).

2. An automated computer vision mango sorting machine according to claim 1, characterized in that it comprises five conveyor belts (29) which have on each side a conveyor belt structure (19), with a total of ten supports for the five conveyor belts (29). In addition, the rollers (22) are located at the ends of each conveyor belt (29) being two for each conveyor belt (29) and a total of ten rollers (8) for this machine. Additionally, two roller supports (8) were incorporated into the machine.

3. An automated computer vision mango sorting machine according to claim 1, characterized in that five HP motors (24) were incorporated, which are attached to the conveyor belt structure (21) and for each conveyor belt (29) an HP motor (24) was added. Likewise, three reduction motors (23) were used, which are responsible for automatically controlling the speed, which is why they execute the signals received from the control system. Additionally, for the regulated passage of compressed air, two pneumatic valves and four pneumatic valves in the form of a "T" were coupled.

4. An automated computer vision mango sorting machine according to claim 1, characterized in that the control system is formed by a control box structure (19), which has a control box cover (20) assembled to protect the control system. In addition, these components are assembled to the threads on which the sucker (6) slides.

5. An automated computer vision mango sorting machine according to claim 4, characterized in that the control system has four diffuse counting sensors (27) inserted, which are located inside the counting sensor support (26). The diffuse counting sensors are responsible for counting the number of mangoes that pass through the machine and emit these signals to the control system that is located in the control box structure (20).

6. An automated computer vision mango sorting machine according to claim 1, characterized in that it is incorporated two pistons (3) each having a rod (2), making a total of two. To the rods (2) a sucker support structure (5) was assembled, in which the sucker support (1) was coupled and in turn six suckers (2) were incorporated, which are distributed three by each rod (4).

7. An automated computer vision mango sorting machine according to claim 1, characterized in that a servo motor (14) is used which is attached to the gear guard (17), said guard internally incorporating the servo motor gear (15); in addition, the gear guard (17) for safety reasons is attached to a gear guard cover (18).

8. An automated computer vision mango sorting machine according to claim 3, characterized in that the one HP motors are secured by pneumatic adjustment bolts (12), a total of ten pneumatic adjustment bolts (12) were used, since two pneumatic adjustment bolts (12) were coupled to the protective sides of the motor (25). Additionally, it should be noted that the machine used five one HP motors (24).

9. An automated computer vision mango sorting machine according to claim 1 is characterized in that it has a fixed general structure (28) which is primarily attached to the conveyor belts (29) and its conveyor belt structure (21).

Citation Information

Patent Citations

  • Automatic efficient mango sorting equipment

    CN111346834A

  • A non-destructive testing and grading device and method for pineapples

    CN113751345B

  • Dry crop sorting equipment

    CN115318661A

  • Real-time inspection and grading system device for fruit appearance quality

    CN202010662U

  • Roxburgh rose sorting system based on machine vision

    CN209255251U