WO2023126654A1

WO2023126654A1 - Automatic optical inspection system to verify the outer packaging conformity specifications of palletized packages, according to customer requirements

Info

Publication number: WO2023126654A1
Application number: PCT/IB2021/062451
Authority: WO
Inventors: Toni Daniel NETO MACHADO; Paul RODRIGUES VIEIRA; António José DE FREITAS MAIO; Edgar Emanuel LEITÃO TAVEIRA; Luís Gonzaga MENDES MAGALHÃES; João Eduardo QUINTELA ALVES DE SOUSA VARAJÃO; Miguel FREIXO GONÇALVES; Gustavo FRAGA DE CARVALHO NOGUEIRA ROSA; André Filipe ALVES FREITAS; Samuel COELHO MACHADO; Alexandra Filipa ARAÚJO GOMES
Original assignee: Bosch Car Multimedia Portugal, S.A.; Universidade Do Minho
Priority date: 2021-12-28
Filing date: 2021-12-30
Publication date: 2023-07-06

Abstract

The present application describes an automatic optical inspection system to verify the outer packaging conformity specification of palletized packaging according to customer requirements. The system is composed by hardware and software and its design was considering the possibility to extend the digitalization coverage (in terms of hardware and software modules). This document describes the system to be applied to Logistics shipping areas, but it is also valid for Logistics reception areas, allowing its integration in logistics processes related with reception and/or shipping of boxed palletized goods. In terms of packaging items, the system is prepared to deal with pallets, boxes, labels, sticker dots, pallet lids and straps, according known packaging rules.

Description

DESCRIPTION

"Automatic optical inspection system to verify the outer packaging conformity specifications of palletized packages , according to customer requirements"

Technical Field

The present application describes an automatic optical inspection system to veri fy the outer packaging conformity speci fication of palleti zed packaging according to customer requirements .

Background art

Document US2020202506A1 describes a method for inspecting pallets to ensure that the overall structure and strapping are correct . The cameras are placed on the top and can also be placed on the sides of the pallet , which itsel f can move on a conveyor . It does not perform an inventory like the other existing systems but determines i f the captured images are within the tolerances of the manufacturing speci fications . It only allows for specific manufacturing conformity inspections , namely the alignment of the pallet on its support and the arrangement of the obj ects ( straps , metal bands and plastic film) used to hold together the packages making up the pallet . The manufacturing tolerance may be to determine i f a plurality of items ( i . e . , packages ) are properly secured to a pallet with straps and i f the straps are properly arranged . The manufacturing tolerance may also address the alignment of the loaded pallet on the support surface ( e . g . , a conveyor ) by means of a target area visibly marked on the support surface or defined by crosshairs generated by or on the image recorder . The software routine uses a pattern recognition subroutine or a pattern matching logic subroutine to determine whether loaded pallet is within a tolerance of the manufacturing speci fication .

Summary

The present invention describes an automatic optical inspection system for outer packaging conformity speci fication veri fication comprising a gate station connected to a web application ; the gate station comprising a computing device provided with a web browser ; a camera system, a trigger system and a measurement system, all configured to gather and provide data to an inspection software ; and the web application comprising a backend application and a frontend application; wherein the inspection software is configured with inspection rules through the backend application to inspect images from the camera system upon request determined by the trigger system .

In a proposed embodiment of present invention, the computing device is connected to the frontend application, which is further connected to the backend application for proper definition of the inspection rules .

Yet in another proposed embodiment of present invention, the camera system comprises at least a set of four cameras over a reception area in a structure that allows the capture of at least four obliquely upper view and side view images surrounding a pallet wherein stacked packages will be analyzed . Yet in another proposed embodiment of present invention, the structure further comprises a set of lights .

Yet in another proposed embodiment of present invention, the inspection software is configured to operate in accordance with a workflow comprising at least four main core modules comprised of a Capture , Pre-proces sing, Classi fy and Locate and Rules Checker .

Yet in another proposed embodiment of present invention, the measurement system comprises at least one of a height or width or length measuring sensor .

The present invention further discloses the operating method of an automatic optical inspection system for outer packaging conformity specification veri fication as previously described, characteri zed by comprising the steps of a user actuates the trigger, setting of a request station data in the inspection software workflow from a station ID; capture one 2D image per camera containing perspective up-down of the stacked packages located in the reception area, and simultaneously determine the volume height of the stacked packages ; preprocess the captured 2D images , one per each volume lateral side face of the stacked packages , and simultaneously request the recording of the captured 2D images ; classi fy and locate the stacked packages , one per each volume lateral side face of the stacked packages ; merge side face results of each volume lateral side face of the stacked packages ; check packet rules which comprise request conformity rules and package rules checker, one per package ; providing and record inspection results . In a proposed embodiment of the operating method of the automatic optical inspection system for outer packaging conformity speci f ication veri fication, the preprocessing the captured 2D images comprises at least the steps of process lens distortion correction of the captured 2D images containing up-down perspectives of the stacked packages ; stitch the captured 2D images i f more than one camera per lateral side face of the inspection volume is installed; correct the perspective , split packages , and simultaneously calculate scale factor of each captured 2D images , providing a bottom to top 2D image array corresponding to the perpendicular view of the lateral side face of packages and corresponding pixel to millimeter conversion value .

Yet in another proposed embodiment of the operating method of the automatic optical inspection system, the classi fy and locate the stacked packages compri ses at least the steps of detect and classi fy packaging items ; segment packaging items ; extract item features and read package ID; output relevant extracted data of packages lateral side face to check .

Yet in another proposed embodiment of the operating method of the automatic optical inspection system, the package rules checker comprises at least the steps of validate packaging item types ; validate packaging item quantities ; and validate packaging item poses .

General Description

The present invention describes a system capable of performing a quick 360-degree optical inspection of pallets . Optical inspection systems are already widely used in industry, including logistics . Indeed, the development of machine learning applied to computer vision has increased the number of uses of cameras in industry, given the increasing amount of data and algorithms available for image processing . Cameras became very useful in monitoring automated production processes and are one of the key elements in the transition to Industry 4 . 0 . In the field of logistics , cameras and optical inspection systems are mainly used for quality and compliance control . Indeed, the results of human visual inspection depend largely on the operator, who is obviously not infallible . These controls are repetitive and careless mistakes are natural . They lead to errors that af fect the rest of the supply chain . It is therefore important for companies to have automatic means of veri fication to assist the human and to ensure transparency of the process by ensuring that the results of inspections are recorded for later analysis . This is of particular importance for companies involved in sending and/or receiving bulky parcels , for which the existing industrial optical inspection systems are applied to the veri fication of palleti zation . The palleti zing check is typically carried out when the pal let has j ust arrived at the warehouse or when the pallet of stacked packages is assembled and ready to be loaded into its carrier . Checking a pallet on arrival allows the inventory to be made and its condition to be veri fied . In the case of a shipment , a f inal check, often carried out at the loading gate of the warehouse , must be done by the shipper to ensure that the pallet is made up of the correct packages and not those of another customer . I f a mistake is made , this leads to logistical problems and additional costs as the parcel has to be returned to the warehouse and reshipped to the correct receiver. Being this a well-known problem in logistics, new needs arise in order to mitigate these setbacks.

On the whole, factories usually deliver their goods to a large number of customers, and some of the active customers require their suppliers to comply with specific packaging rules. These packing rules define how the packages are assembled on the pallet (in the case of bulky shipments) and how certain elements are positioned on the packages such as labels, straps, barcodes, etc. Based on customer's orders, finished goods are properly packed and labeled according to those rules before being sent to them. When the supply chain within the factory is not fully automated, most of these operations involve the execution of numerous manual tasks, such as packaging the finished goods in the appropriate way, attaching labels to palletized finished goods (according to customer's packaging instructions) and loading pallets to trucks. The more customers there are, the more different packaging rules need to be respected. Even for the same customer, depending on the order destination, the package rule may be different. If the steps of labelling and assembling the packages into pallets are done manually, the risk of error becomes then significant.

All known existing industrial optical pallet verification industrial products only focus on checking whether the packages are on the right pallet before they are sent. They perform an inventory, but they do not verify whether customer-imposed packaging rules have been respected when assembling the pallet. They are mainly developed to use cameras optimized to detect and read barcodes only. The information on the barcode provides an inventory of the packages constituting the pallet but does not allow further analysis of the constitution and structure of the packages . In some particular cases the developed systems use the cameras not only to detect the barcodes , but also to define the visual content . However, even i f the cameras are used to recogni ze predetermined obj ects in a crate or pallet , it only performs an inventory of the shipment by comparing the quantity of detected obj ects with the expected one according to their database . Apart from counting the number of items , it does not check any other package composition rules .

In addition, as the palleti zation optical veri fication systems are usual ly installed at the receiving or dispatching gate of the warehouse , the structure on which the cameras are installed must not interfere with the forkli ft trucks operating there . Some known systems include portal-type structures placed on both sides of the gate . However, this type of layout only captures the two sides of the pallet visible to the cameras . In order to veri fy the packaging rules , each side must be captured . On portal-type structures , the front and rear faces are not captured .

All the known previously described products perform pallet inventory through an optical system, failing to address the checking of the pallet construction conformity and the packages composition according to customer rules . Present invention aims therefore to ful fill these flaws , providing a solution for this problem .

Therefore , and to overcome known technologies , the present invention allows to determine whether the customer-imposed packaging rules have been respected when assembling the pallet , ensure that the structure on which the cameras are installed does not interfere with nearby operating forkli ft trucks , and also perform the complete inventory of the packages comprised on the pallet

The present invention allows therefore to perform package conformity veri fication, using a set of rules speci fied by the customer and information acquired through several computer vision processes , this application can then determine i f the packages are compliant to the customer speci fications . During this process , the application also saves some key photos as a proof of package condition . The developed system comprises a simple and user- friendly interface that can be used to define all the package conformity requirements set by the customer . Through this interface , the user can define all the packaging items needed for the package construction and the speci fic requirements for label positioning, packaging item type and container disposition, according to the customer speci fications .

Within the scope of the packaging items available for configuration it is possible to define :

- Pallet support : the base of the package that bears and supports all the boxes ;

- Boxes : the containers/packages that make up the package ;

- Straps : belts that hold together the pallet , boxes and lid;

- Sub labels : each box may have a label containing information relative to its content ;

- Stickers : used to hold labels in place , when requested;

- Master label : contains information relative to the whole package (pallet ID) . Based on customer requirements , the set of rules is created (using the above stated packaging items ) and then used in the conformity validation of each package .

The system also grants a full package coverage using arrays of cameras capable o f capturing the four lateral side faces of stacked packages . This induces the speeding of the image acquisition process minimi zing the impact of the shipping process .

Brief description of the drawings

For better understanding of the present application, figures representing preferred embodiments are herein attached which, however, are not intended to limit the technique disclosed herein .

Fig . 1 - depicts the developed automatic optical inspection system architecture that is compri sed of two main components , where the reference numbers relate to :

100 -architecture of the automatic optical inspection system;

10 - web application;

11 - backend application;

12 - frontend application ;

20 - gate station;

21 - inspection software ;

22 - camera system;

23 - trigger system;

24 - computing device with web browser ;

25 - measurement system;

30 - wired or wireless telecom connection; 40 user .

Fig. 2 - depicts a preferred embodiment of the physical structure of the developed automatic optical inspection system, wherein the reference numbers relate to:

200 - automatic optical inspection system;

210 - trigger;

220 - camera;

230 - light;

240 - reception and/or dispatch area;

250 - pallet of stacked packages;

260 - height sensor.

Fig. 3 - depicts the workflow of the inspection software (21) of the gate station (20) , wherein the reference numbers relate to:

210 - trigger;

1000 - inspection software workflow from a station ID;

1001 - request station parameters;

1002 - Capture, one 2D image per camera containing perspective up-down of the configured area;

1003 - get volume height;

1004 - preprocessing, one per volume's lateral side face ;

1005 - request record captures;

1006 - classify and locate, one per package's lateral side face;

1007 - join face results about each package;

1008 - packet rules checker;

1009 - request conformity rules;

1010 - package rules checker, one per package;

1011 - request record inspection results;

2101 - station ID; 10011 - station parameters ;

10021 - 2D image array containing the perspective up- down of the configured volume ;

10031 - total height of the packages in the inspection allowed volume ;

10041 - 2D image array ( from bottom to top ) corresponding to the perpendicular view of the lateral side face of packages and corresponding pixel to millimeter conversion value ;

10042 - lens distortion correction;

10043 - stitch;

10044 - perspective correction;

10045 - detect split packages references ;

10046 - calculate scale factor (pixel to millimeters conversion value ) ;

10047 - split packages ;

10061 - relevant extracted data of packages lateral side face to check ;

10062 - detect and classi fy packaging items ;

10063 - segment packaging items ;

10064 - extract item features ;

10065 - read package ID;

10071 - relevant extracted data of packages lateral side face to check;

10081 - package rules checker ;

10091 - customer package rules for the package under inspection;

10101 - validate packaging item types ;

10102 - validate packaging item quantities ;

10103 - validate packaging item poses ;

100421 - 2D image array containing the perspectives up- down of the configured volume + volume height + calibration parameters ; 100441 - 2D image corresponding to the perpendicular view of the lateral side face;

100451 - array of lines (from bottom to top) slip packages ;

100461 - pixel to millimeter conversion value;

100471 - 2D image array [from bottom to top] corresponding to the perpendicular view of the lateral side face of packages;

100621 - objects detection and classification array;

100631 - segment packaging items;

100641 - packaging item coordinates and features;

100651 - package ID;

Description of Embodiments

With reference to the figures, some embodiments are now described in more detail, which are however not intended to limit the scope of the present application.

A particular embodiment of the presently disclosed invention describes the system architecture based on components dependencies where the automatic optical inspection system (100) comprises a gate station (20) and a web application (10) .

The gate station (20) is comprised of a trigger system (23) , a camera system (22) , an inspection software (21) , a computing device (e.g., computer, smartphone) with web browser (24) and a measurement system (25) . The web application (10) is comprised of a backend application (11) and a frontend application (12) . Both gate station (20) and a web application (10) are connected and communicate through a wired or wireless telecom connection (30) . The gate station (20) is operated by a user (40) which is responsible for its operability (i.e., triggering and accepting results) . The inspection software (21) uses the inspection rules accessible via backend application (11) to inspect images from the camera system (22) upon request by the trigger system (23) . The web browser (24) , which ends up being the system user interface to the user (40) , is connected to the frontend application (12) , that, in turn is further connected to the backend application (11) to satisfy user's interaction .

The camera system (22) is comprised of a set of at least four cameras (220) installed and arranged in particular positions over a reception and/or dispatch area (240) wherein stacked packages (250) will be subject to inspection. In a preferable embodiment of the present invention, each of the cameras (220) is positioned in a fixed structure in order to be able to capture at least an obliquely upper view and side view imaged of the pallet of the stacked packages (250) . Therefore, the camera system (22) will be responsible for collecting at least four oblique upper view images surrounding the pallet of the stacked packages (250) .

The trigger system (23) is comprised of at least one (physical or digital) pushbutton trigger (210) , available for user (40) to request the inspection software (21) to iterate its conformity inspection workflow as illustrated in Figure 3.

The gate station (20) is therefore responsible for the conformity inspection of the stacked packages (250) through the camera system (22) configured to capture at least four independent lateral side faces of stacked packages, and the inspection software (21) verifies the package conformity with the predetermined set of rules by the user (40) through the web browser terminal (24) and stored in the backend application (11) .

The Web Application (10) represents a web-based software application, comprising both backend application (11) and frontend application (12) , that will be made available via a web browser installed and accessed through a computer (24) or adequate device for the purpose. The Web Application (10) includes features to specify packaging items, packaging rules and to consult data recorded about conformity verifications .

On the reception and dispatch area (240) , the pallet status must be automatically registered based with a snapshot of the four lateral side faces of the packages. To mitigate the risks regarding physical limitations and operation restrictions, the cameras (220) and lights (230) are placed in a supporting structure that in a preferred embodiment can be suspended on the ceiling of the of the warehouse where the physical system (200) is installed. The cameras (220) must be positioned to ensure that at least one set of cameras is faced to each of the lateral side face of the allowed volume, in order to capture pallets stacked. Figure 2 discloses one possible arrangement of the cameras (220) in the structure, that in some cases could comprise only one unit, but the system is prepared and can be configured to have one or more cameras (220) capturing each lateral side face of the allowed volume located in the dispatch and/or shipping area (240) . Said reception and dispatch area (240) comprises a trigger (210) that will signal and activate the inspection software (21) to promote the capture of images when package (s) is (are) placed in the capturing zone (240) . The inspection software (21) has access to the cameras (220) through high-speed data connections .

The inspection software (21) workflow (1000) is illustrated in Figure 3. The workflow (1000) represents the orchestration of the four main core modules to constitute the conformity validation of outer packaging. The core modules are: Capture (1002) ; Pre-processing (1004) ; Classify and Locate (1006) ; and Rules Checker (1010) . Essentially, the workflow (1000) orchestration consists in calling core modules in a manner to make possible check the conformity of outer packaging (250) according customer specifications (10091) . Process computing about the same core module (1002, 1004, 1006 or 1010) can be parallelized in the running inspection context. As example of orchestrating those calls, see Workflow main module :

- request station data (1001) through trigger (210) actuation, wherein the data comprises all station' s specific parameters (2101) ;

- capture data (1002) , calling the core module Capture in a loop and calling a module to Get Volume Height (1003) ;

- while processing captured images, those can be stored in the web application (10) , calling the module re quest Re cordCapture ;

- processing of captured images by: o preparing each set of captured images (note that each lateral side face can be obtained by more than one camera) to insulate each package present in the whole volume face in order to feed the core module Classi fy and Locate ; o once having each package face insulated in distinct images , those images are ready to pass through the core module Classi fy and Locate in order to obtain all the relevant data of each lateral side face ; o after these data extraction is completed, in module Join Faces Results , they are organi zed/grouped as package data to pass to be passed to Rules Checker core module ;

- next , the conformity validation is performed by looping each package data extracted against the packaging rules required by the customer of each package . This is achieved by requesting to the backend application the applicable package rule for the package ID through the request conformity rules module ( 1009 ) and passing package data extracted and applicable package rules to the core module Rule Checker to confirm is every applicable package rule is or is not ful filled; finally, the inspection result is sent to the backend application via request recording inspection results module .

This capture module ( 1002 ) is responsible for collecting images from the camera system ( 22 ) . This module is called at least four times , one per each camera ( 220 ) . Said images will be channel led for further processing in the next modules . Said data requires at least an array of images ( 10021 ) corresponding to the images to cover the lateral side face of the volume allowed volume , and the physical height of the volume to inspect ( 10031 ) and the corresponding calibration parameters (get from station data requested previously)

The input of the pre-processing module (1004) comprises an array of images (100421) containing the perspectives up-down of the configured volume, the volume height of the volume in the configured capturing area and cameras' calibration parameters. Once gathered raw images, the first step of the pre-processing, the lens distortion correction (10042) , (if needed) comprises correcting lens distortion in the captured images (due light rays bending passing through lens) . To be noted that, in case of more than one camera (220) per lateral side face of the configured volume (250) , a stitching sub module (10043) is needed (right after this first step) to create a single image for each lateral side face of the captured volume. The perspective correction (1044) for each lateral side face of the volume under capture is obtained using calibration parameters about the volume allowed to capture (virtual points of the corners in the captured image that delimits the allowed area for the lateral side face in capture) . Knowing the width and height of the rectangular allowed area, based on distances between those virtual points, using the physical height (10031) (e.g., via distance sensor (260) from top to bottom) makes possible to calculate the scale factor (10046) , which allow the conversion from pixel units to metric units. To detect split references (10045) , a pretrained model to classify and locate pallets (250) is used. Since a package always contains a pallet (250) on its base, the split package sub module (10047) is able to split the already pre-processed image into multiple ones containing single lateral side faces of packages (one input image may result in multiple output images, according the number of pallets detected - remember that one detected pallet corresponds to the base of one package) . The output of this core module (10041) is one image containing a single lateral side face per package and its scale factor value (pixel to measure unit conversion value) which is a fundamental value to compute measurement values for the different packaging items.

The input for the locate and classify (1006) module is an image containing perpendicular view of a single pallet side face; and conversion factor from pixels to world measures (e.g., millimetres) (10041) .

With the undistorted insulated images containing single lateral side face of package, the software now uses a pretrained object detection model (10062) to locate and segment all the packaging items (10063) , classifying them according the classes trained, distinguishing sub types (also called as references) of each packaging item type (e.g. for a packaging item type of "pallet" will be classified as "pallet_dummy_l" to "pallet_dummy_N", where "pallet_*" represents a reference a packaging item of the type "pallet") . Depending on the object type some features are extracted (10064) to be used in the next core module, such as the dominant colour, position, orientation and dimensions. The previously calculated scale factor is used to convert pixels to a unit of measurement allowing the system to convert the detected and/or segmented objects' bounding boxes into real measurements. In one of the images there is a Main Label. When the software finds this label, it starts a segmentation process that tries to isolate the Package ID so that the Read Package ID module (10065) can read that same ID using OCR or barcode decoding algorithms. The output of this core module is all the relevant data of package's lateral side face (10061) that will be used in the Rules Checker module.

The module Rules Checker (1010) requires extracted package data (10061) and customer's package rules to fulfil (10091) . Using these inputs, a three-phase validation process is initiated (it is not mandatory to have those validation steps nor respect this sequence) .

1. Packaging item type validation (10101) - the system compares the information from the packaging rules with the package data extracted in the Classify and Locate core module. If all types match, the system validates the packaging item type and goes to the next validation phase .

2. Packaging item quantity validation (10102) - to determine the expected number of labels, first the system needs to determine how many individual boxes compose the package. From the package data extracted in the core module Classify and Locate, the total boxes height in the inspecting package can be obtained by subtracting the heights of pallet and pallet lid from the package height. The number of layers can be obtained by dividing this total boxes height by the customer specification box's height. The total number of boxes for an inspecting package can be obtained by multiplying the amount of "box distribution per layer" (found in the packaging rules) by the number of layers of the inspecting package. For each box may be expected a single sub-label. Using the packaging rules and the expected number of labels, the system knows the number of stickers that should be in each label. After all the calculation are complete, the system only needs to compare the newly acquired information with the obj ects detected in the previous process .

3 . Packaging item pose validation ( 10103 ) - In this phase , the determined pose (position and orientation) of the packaging items is veri fied against the expected one , di f ferent techniques are used according to the item type . The items expected pose are obtained via packaging rules speci fication .

After these three phases are concluded, the validation is recorded with images and illustrations of the validation as evidence of package state right before it leaves the gate station .

The package and labelling rules are stored in a relational database . The inspection results and all inspection evidence ( snapshots of the packages ' status ) are also stored in a database . A Web Application, containing both backend and frontend, is used to speci fy the packaging items characteristics , the packaging and labelling rules ( feed the rules database with information) and to consult recorded data about inspections that were carried out by the system . The backend contains all the Application Program Interfaces (API ) , the business logic and its data stored in a relational database . On the other hand, the frontend is responsible for providing web access to the web application functionalities in a user- friendly way .

The suggested embodiments of the disclosed invention constitute a complete automatic optical inspection system ( 200 ) of palleti zed packages that veri fies the outer packaging conformity to the customers ' requirements , which can be used to inspect packages on the reception or dispatch area . With the developed system, the palleti zed packages conformity is ensured, leading to a reduced number of customers complains . This overcomes existing inspection systems since they do not address nor comprise the ability of checking the conformity of the pallet construction and the packages composing is according to customer rules .

Claims

22 CLAIMS

1. Automatic optical inspection system (100) for outer packaging conformity specification verification comprising a gate station (20) connected to a web application (10) ; the gate station (20) comprising a computing device provided with a web browser (24) ; a camera system (22) , a trigger system (23) and a measurement system (25) , all configured to gather and provide data to an inspection software (21) ; and the web application (10) comprising a backend application (11) and a frontend application (12) ; wherein the inspection software (21) is configured with inspection rules through the backend application (11) to inspect images from the camera system (22) upon request determined by the trigger system (23) .

2. Automatic optical inspection system (100) according to the previous claim, wherein the computing device (24) is connected to the frontend application (12) , which is further connected to the backend application (11) for proper definition of the inspection rules.

3. Automatic optical inspection system (100) according to any of the previous claims, wherein the camera system (22) comprises at least a set of four cameras (220) over a reception area (240) in a structure that allows the capture of at least four obliquely upper view and side view images surrounding a pallet wherein stacked packages (250) will be analyzed .

4. Automatic optical inspection system (100) according to any of the previous claims, wherein the structure further comprises a set of lights (230) .

5. Automatic optical inspection system (100) according to any of the previous claims, wherein the inspection software (21) is configured to operate in accordance with a workflow (1000) comprising at least four main core modules comprised of a Capture (1002) , Pre-processing (1004) , Classify and Locate (1006) and Rules Checker (1010) .

6. Automatic optical inspection system (100) according to any of the previous claims, wherein the measurement system (25) comprises at least one of a height (260) or width or length measuring sensor.

7. Operating method of an automatic optical inspection system (100) for outer packaging conformity specification verification described in any of the previous claims 1 to 5, characterized by comprising the steps of a user (40) actuates the trigger (210) , setting of a request station data (1001) in the inspection software workflow from a station ID (100) ; capture one 2D image per camera (1002) containing perspective up-down of the stacked packages (250) located in the reception area, and simultaneously determine the volume height of the stacked packages (250) ; preprocess the captured 2D images (1004) , one per each volume lateral side face of the stacked packages (250) , and simultaneously request the recording of the captured 2D images (1005) ; classify and locate (1006) the stacked packages, one per each volume lateral side face of the stacked packages (250) ; merge side face results of each volume lateral side face of the stacked packages (250) ; check packet rules (1008) which comprise request conformity rules (1009) and package rules checker (1010) , one per package; providing and record inspection results (1011) .

8. Operating method of the automatic optical inspection system (100) for outer packaging conformity specification verification according to the previous claim, wherein preprocess the captured 2D images (1004) comprises at least the steps of process lens distortion correction (10042) of the captured 2D images containing up-down perspectives of the stacked packages (250) ; stitch (10043) the captured 2D images if more than one camera per lateral side face of the inspection volume is installed; correct the perspective (10044) , split packages (10047) , and simultaneously calculate scale factor of each captured 2D images, providing a bottom to top 2D image array corresponding to the perpendicular view of the lateral side face of packages and corresponding pixel to millimeter conversion value (10041) .

9. Operating method of the automatic optical inspection system (100) for outer packaging conformity specification verification according to any of the previous claims, wherein the classify and locate (1006) the stacked packages comprises at least the steps of 25 detect and classify packaging items (10062) ; segment packaging items (10063) ; extract item features (10064) and read package ID (10065) ; output relevant extracted data of packages lateral side face to check (10061) .

10. Operating method of the automatic optical inspection system (100) for outer packaging conformity specification verification according to any of the previous claims, wherein the package rules checker (1010) comprises at least the steps of validate packaging item types (10101) ; validate packaging item quantities (10102) ; and validate packaging item poses (10103) .