THE SOLUTION
The system is made up of 3 SmartCamera NI 1722, which signal faults and errors detected to the line control PLC; communication with the line supervision software occurs via an Ethernet network and allows operators to be provided with information on inspections carried out on the production batch.
ARTICLE
Objectives
The management of the rotary labeller is completely delegated to the line operators, who are required to adjust the height of the various application groups as well as load the rolls of adhesive labels onto them. Any error in one of these manual operations can cause a prolonged stoppage of the line; the worst case is when the wrong roll is loaded: this forces operators to manually remove the wrong label from the bottles, which will then have to be reinserted into the line to apply the correct label. This eventuality is even more serious in the case of incorrect application of the bands for DOC and DOCG wines, which, in addition to the name of the wine, report a unique code that identifies the bottle to which it is applied and ensures its complete traceability.
In addition to these possible problems, without automatic control of the labels, any missing or visibly crooked applications could escape the attention of the operators.
Given the great variety of wines produced by Cantina Vinci, the system implements a recognition algorithm that is flexible enough to be applied to each label and back label; it also provides an archive in which samples of the various labels are catalogued, so that the correct recognition can be applied to each of them. This archive must also contain the information necessary for recognizing the bands and reading the unique code reported on them.
System description
The proposed vision system is composed of 3 NI 1722 smart cameras, two of which implement the same algorithm for label and back label recognition; the third camera instead inspects the band, of which it recognizes the wine denomination and reads the unique code reported. The three cameras are arranged along the CAM of the labeler, and the acquisition is triggered by presence sensors that detect the passage of the bottle; This solution ensures that the bottles are oriented in the same way at each inspection, without having to communicate with the PLC to synchronize the acquisition with the machine’s CAM.
The communication of the 3 cameras with the rest of the pre-existing system occurs in the following way:
- voltage signals to the PLC that manages the line, to indicate faults or alarms (failure to recognize, incorrect label, non-consecutive codes) that are managed by the PLC itself and lead to the possible immediate stop of production;
- communication via Ethernet network with the supervision PC.
The Program
The system includes two different applications: one used for label and back label recognition, replicated on two different cameras, the other used for the analysis of the identification band of DOC and DOCG products.
The first consists in the search, within the acquired image, of three details belonging to the label or back label associated with the batch in production; depending on the presence of such details and their relative positions, the applied label is recognized or not; with this algorithm, it is also possible to determine the position of the label on the bottle, providing information on its height and any rotation.
The second application instead provides for the search, inside the band, of the detail that reports the name of the wine and the OCR recognition of the code written on it; also in this case, in addition to detecting the presence of the correct band and verifying compliance with the code shown, the system returns information on the correct positioning of the band on the bottle.
Both algorithms have been optimized to be able to be executed at the maximum speed of the labeler, equal to 12000 bph, so the vision system has 0.3 seconds available to analyze each individual bottle.
In addition to implementing the inspection and error reporting functions, the system returns information on the batch in production, such as the number of bottles inspected and how many of them were rejected
The algorithms mentioned above can be applied to any type of label, back label and band; they require a rigid sequence of recognitions, calculations and interpretation of codes, but they have been created in order to adapt to different types of labels. This adaptability is made possible by a configuration file, present on the SmartCamera, containing a sample of the label and all the data necessary to perform the correct recognition (position of the details to be searched for, range within which the codes to be recognized must fall, etc.). Given the limited memory of the SmartCamera, the sample archive is implemented on the PC on which the supervision program resides; this program, already created using LabVIEW, is also modified in order to associate a label-back label pair to each production batch, specifying the corresponding band if necessary.
At each production launch, the supervisor sends the configuration files necessary for the correct functioning of the three cameras via FTP.
Since the variety of wines produced by the winery is constantly evolving, the customer has the possibility of creating new models of labels, back labels and bands. The application used for this purpose resides on the line manager’s laptop, connected to the same Ethernet network to which the SmartCameras belong only when necessary, and assists the user in acquiring the image and choosing the details to be recognized. The creation of the model is followed by a verification of the same and a comparison phase with each pre-existing model in the archive, to avoid that two different labels have details similar enough to be indistinguishable.
Results and conclusions
The label control system just described has improved the productivity of the line, increasing its robustness through safe and immediate verification of the bottles; in this way, prolonged production stops due to the reintroduction of a large number of bottles into the line are avoided. By checking the codes on the bands, product traceability is more secure, as the system strictly checks the correspondence between the bottle passing through the line and a specific code assigned in advance.
The operators’ work was not complicated in any way, as the only impact on the pre-existing system was the introduction of a few fields in the production batch launch interface.
The integration of the vision system into the pre-existing structure was a quick and almost painless procedure, thanks to the versatility of the LabVIEW environment, which allowed the existing system to be kept substantially unchanged before the project was created.
The choice of Studio Progetti Automation and Cantine Vinci to use the National software, made at the beginning of their collaboration, proved to be a winning one, and is an indication of their predisposition to use a high level of technology, as well as a strong desire to improve their production chain.