Non-destructive
digital

Inspection

Artificial Vision

Within the field of Artificial Perception, Vision systems are tools that permit perceiving reality by means of electronic cameras. These can be constructed based on different technologies, so, in many cases, they enable us to perceive what the human eye cannot see.

Vision technologies offer the possibility of automating a large number of industrial processes with a much greater precision than is possible with human work. These processes include quality controls, dimensional control, stock control in logistics, contactless deformation measurement, or robot guiding, among others.

The characteristics of the technology, especially its speed and the fact that no contact is required, are allowing the replacement of manual and undesired operations (or in some cases, the execution of tasks that cannot be done by other means), also contributing towards the digitalization of companies, as they incorporate relevant information about production processes.
Some of the benefits afforded by this technology include: real time quality control, early defects detection and real time monitoring of the production process.

What we are currently working on

Welding/additive quality control

The aim is to tap into technology to detect anomalies (surface defects or rejects of another type) in welded joints or additive manufactured parts.

This technology is being applied to extremely high frequency (1 second cycle time) welding processes, as well as to detect defects on variable and irregular surfaces, proving the versatility of the technology.

Process quality control

Integration of vision-based quality control system in production lines that are permitting the availability of on-site quality-related information. This technology is being applied in a way that cuts across many sectors such as capital goods, energy or transport.

3D Inspection

3D information generation and acquisition techniques are highly evolved and mature. There are different systems on the market that are based on these techniques (stereoscopy, triangular laser, etc.), which permit the robust and reliable acquisition of geometries or volumes, and with high quality data. Among the many possible applications, 3D inspection permits the external inspection of welding beads or components constructed by additive manufacturing.

Based on this technology and supported by CAD/CAM solutions, LORTEK has developed an automatic arc welding repair system for high value added parts (see success case).

Stock control in logistics

Incorporation of vision technologies to control stock in industrial warehouses, and automate orders in order to reduce the amount of company stock to a minimum.

Own algorithms

Derived from different industrial projects already executed at LORTEK, the vision department has libraries dedicated to image processing for the automatic control of quality systems. This entails a reduced ramp-up in new computer-assisted vision projects in industrial contexts.

Specific Equipment

Different camera models

  • High resolution, high speed, etc.

Optical components

  • Lenses.
  • Filters.
  • Different lighting sources including pulsed lasers for stroboscopy.

Tecnologías Deep learning

  • NVIDIA embedded systems (nano, Tx2, Xavier), Raspberry, etc.
  • Dedicated servers for modelling (more than 20,000 cores).
  • Collection of own quality control-oriented models.

Publications and downloads

Publications
2019
Ander Muniategui; Aitor García de la Yedra; Jon Ander del Barrio; Manuel Masenlle; Xabier Angulo; Ramón Moreno.
Mass production quality control of welds based on image processing and deep learning in safety components industry Proceedings Volume 11172
Fourteenth International Conference on Quality Control by Artificial Vision; 111720L

A Muniategui, JA del Barrio, XA Vinuesa, M Masenlle, AG de la Yedra, Ramón Moreno.
One dimensional Fourier Transform on Deep Learning for industrial welding quality control
International Work-Conference on Artificial Neural Networks, 174-185

R Moreno, E Gorostegui-Colinas, PL de Uralde, A Muniategui.
Towards Automatic Crack Detection by Deep Learning and Active Thermography
International Work-Conference on Artificial Neural Networks, 151-162

Success Cases

Quality control of welded joints of safety components at high production rate and low cycle time: 1 second

Challenge

Quality control of welded joints of safety components, in high takt time production line, with cycle times of close to 1 second.

Solution

An artificial vision-based system was designed, manufactured, and developed to provide a solution to this problem, consisting of introducing three cameras, each positioned at 120º, to permit the reconstruction of circular welding. This reconstruction is used to analyze the welding quality with previously trained Deep Learning-based algorithms. The result of this is a device that determines if the part is good or bad, rejecting it from the production line, preventing its delivery to the customer, and also avoiding operations on already faulty parts.

Challenges

Challenges to be faced in the coming years:

Application of technology in highly variable environments; product versatility, adverse environments, etc…

To cope with this challenge of offering reliable solutions to industrial problems, Lortek is developing robust image-based systems, with Deep Learning techniques that preserve knowledge.

Aitor García de la Yedra, PhD.

Main researcher in Photonic-based monitoring

Senior researcher in Advanced Manufacturing at LORTEK since 2013. He has proven experience in process monitoring and non-destructive testing (NDT) by means of optical techniques. He is currently leading projects in intelligent systems (based on self-learning concepts) to predict and avoid the appearance of defects in high added value processes, such as laser processes (welding and additive). He is the author of several scientific articles and multiple papers at international congresses related to structural integrity (prediction of component lifecycle, and NDT (thermography, acoustic emission, etc.). Previously, he had worked at other centres of the partnership, developing measurement and inspection processes (for machine-tools, aeronautical industry, etc.). He has also participated in numerous international research groups, noteworthy among which are, COMBILASER (H2020-FoF1), WELDMINDT (Clean-Sky) or DELASTI (Clean Sky). His participation (co-leader) in the first group is very relevant, working to introduce intelligence in complex processes such as those based on laser manufacturing.