Portfolio

AUTOMATED QUALITY CONTROL SYSTEM FOR TEXTILE PRODUCTS

Project background:

Production process of textile products currently involves manual quality control, and client is interested in automating this step.

Biggest challenges: highly unpredictable environment to analyse – products overlapping each other, folded in different ways, many different potential design solutions.

Solution:

Automated textile quality control system has been developed to identify the brand, size, color, and printed design elements of textile products. It also compares this data with  specified order parameters.

Customized system for result visualization was developed and integrated with client’s business management systems and database. Computer vision and machine learning solutions were used. Visual data processing takes place in real time.

Key benefits:

• Improved production process;
• Improved quality control time and efficiency;
• Optimised workforce costs.

AUTOMATED MEASUREMENT SYSTEM FOR TIMBER LOGS

Project background:

In manual mode, a certified surveyor inspects each timber log and assigns a quality and defect category based on subjective factors, which often leads to oversight errors. Client aims to automate the quality control process by implementing a system that can analyze quality of timberlogs, detect defects, and determine the grade of logs or veneer blocks.

Biggest challenges: The high speed of conveyor line, wide visual variation, and the size of timber logs place high demands on depth of focus and light intensity. Large dimensions of logs significantly limit the feasibility of conducting tests in laboratory conditions.

Solution:

A unique system was developed and implemented, capable of performing the following functions: identifying the wood species, analyzing the wood surface both from the side and in cross-section, obtaining data on the log’s length and diameter, assessing the quality of  timberlogs, detecting defects and other parameters for sorting veneer logs by grade. The number of timber logs to be evaluated per day can reach up to 9,000.

Key benefits:

• Improved production process;
• Optimised quality data acquisition and processing;
• Reduced time needed for data extraction and improved accuracy of the data obtained;
• Optimised workforce and efficiency.

RESEARCH PROJECT IN MICROSCOPY: AUTOMATED MICROSCOPY SYSTEM FOR MICROBIOLOGY

Project background:

To develop an automated fluorescence microscopy system capable of independently and consistently quantifying and classifying bacteria in water samples on microscopy slides.

Biggest challenges: To develop a methodology for describing dataset, as every manual process involves a human subjective factor, and each laboratory technician may count a different number of bacteria in the same sample.

Solution:

As part of the research project, Apply is developing a high-performance microscope and automated microscopy system that can more accurately identify and classify bacteria. Computer vision and artificial intelligence solutions were used, and neural network models were trained for this solution.

Key benefits:

• The system reduces analysis time;
• Higher reliability and repeatability of results;
• Reduces the physical strain on laboratory workers.

DEPOSIT SYSTEM WINGO (FOR EMPTY PACKAGING AND DISPOSABLE E-CIGARETTES)

Project background:

Latvia is implementing a Deposit System and project requires the development of packaging collection and processing technology to accept and sort empty packaging, electronic cigarettes and batteries, based on artificial intelligence solutions.

Biggest challenges: Find correct parameters and develop solutions so that the system can recognize damaged, flattened empty containers.

Solution:

Within the project framework, physical devices and their software were developed for two different deposit systems:

(1) a deposit system for collecting and sorting empty containers – PET bottles, tetra packs, cans in different sizes, 0.5 beer glass bottles, finger batteries. The machine is able to accept empty containers in both intact and flattened formats.

(2) An electronic cigarette deposit system that uses artificial intelligence to accept and sort e-cigarettes and batteries.

The solution includes a mobile app, central administration system, control panel software and AI process assurance software.

In response to user interaction, technology performs real-time object detection and analysis by using computer vision, analysing both – video streaming and sensor data.

The device analyzes parameters of the deposited object – shape, symbols visible on the object, barcode, weight, raw material, the decision made as a result of computer vision analysis and size of the object.

Technologies used:

• Containerised solution architecture;
• Computer vision with classification neural networks.

QUALITY CONTROL SYSTEM FOR EMPTY GLASS BOTTLES

Project background:

The client often received defective glass containers from the supplier. The quality control person was not always able to spot the defective bottles during the production process and remove them from the conveyor line. The defective product was therefore put on the shelves of shops, which resulted in complaints from shoppers.

Biggest challenges:

Quality defects in glass bottles are difficult to detect because of material transparency, light refraction makes microcracks and bubbles hard to distinguish. In addition, optical illusions and lighting effects can mask damage. The variety of defects and the high speed of the production conveyor line make this solution even more challenging.

Solution:

Developed an automated system which detects bottle defects without human intervention, by using a computer vision solutions. System can detect following defects:

• cracks,
• scratches on the walls of the glass bottle,
• defects of bottle neck,
• rinse residue,
• and other contaminants.

Project involved developing and installing a prototype device on a filling line. The deviceconsists of the following components: control panel, handling process, administration system, physical quality inspection unit using a unique mirror system, three cameras, bottle detection sensors, and bottle separation inspection stage which physically redirects defective bottles to a separate conveyor belt.

Key benefits:

• Optimised costs and speed of production processes;
• Reduced time to extract and process quality data;
• Human error eliminated.

PLASTIC PELLET QUALITY CONTROL RESEARCH PROJECT

Project background:

The client is a recycling company that specializes in processing LDPE and PP materials into high-quality pellets. Manual quality control during the production process resulted in low workflow efficiency.

Biggest challenges: The pellets are small in size and their quality must be assessed during the pouring or falling process.

Solution:

Within the framework of project, a prototype device was developed that uses a specially designed ramp to pour and evenly distribute plastic granules in a single layer. In parallel, a custom developed computer vision system detects falling pellets using a high-speed performance camera, classifies them into 11 categories, and analyses the size and brightness of pellet colour. System provides a statistical assessment of the quality of plastic pellet batch by calculating defect rate and analysing average properties of the pellets, allowing the manufacturer automatically determine quality level and adjust production process to reduce defects.

Key benefits:

• Improved workflow efficiency;
• Optimised quality data acquisition and processing;
• Improved production process, reduced production line downtime;
• Optimised workforce and efficiency.

RESEARCH PROJECT ON SAND MOLD QUALITY CONTROL

Project background:

The client’s production process involves creation of more than 150 custom sand molds, which are then used to cast iron parts. During the mold-making process, sand grains often crumble off or, conversely, extra sand pieces/grains remain in the mold, causing inaccuracies in the cast iron part. Therefore, a quality control system is needed that visually evaluates the quality of each sand mold.

Biggest challenges: the large number of molds and variation in their shapes.

Solution:

As part of the project, a prototype algorithm was developed that could identify defects in real time and visually indicate their location within the form under production conditions and using a camera positioned above the production line. Defects down to 0.5 cm were visually identified. The mold’s alignment with the necessary standards and its proper placement was also assessed.

Visual analysis was performed by using motion tracking algorithms, computer vision algorithms, and deep learning methods.

Key benefits:

• Improved quality control time and efficiency;
• Remote monitoring;
• Increased customer satisfaction.

MANUFACTURING AUTOMATION SYSTEM FOR CHIPBOARD FURNITURE COMPONENTS

Project background:

The client manufactures chipboard furniture components. Due to increasing demand, there is a need to optimize production process and implement automated solutions.

Biggest challenges: A variety of equipment (produced by different manufacturers), robots and processes must be integrated into a unified, automated system.

Solution:

Apply develops management software that automates production processes.

Every component produced must undergo a complete production cycle. First, a large chipboard is cut to the required size. Then, the edges are glued, fixing holes are drilled, and components are shelved. Finally, the order is assembled and sent out for palletization. Customer chooses the algorithm, and in the final stage, furniture components are stacked on pallets based on that algorithm.

The control system developed ensures the complete automation of this production process. The system is active, it plans ahead, and it can be optimized on a number of parameters. It ensures automated operation of the robot arms, conveyors, various sensors and manipulators.

Integrations have been created with the client’s internal databases and various types of equipment.

Key benefits:

• Improved efficiency of the production process;
• Improved process and resource planning;
• Improved competitiveness.