Inspecto

Inka Krischke | Inka Krischke,

Autonomous machine image processing

Global competition and market expectations are putting increasing pressure on automotive manufacturers to improve their quality assurance procedures and switch to automated inspections. One challenge is to find the right machine vision method.

© Inspecto

The Bosch plant in Waiblingen produces plastic connectors for the automotive sector. Injection molding is traditionally a difficult application for conventional machine vision due to the lack of contrast between the background and the part being inspected - the images are 'black on black'. To solve this problem, Bosch turned to Inspekto, the founder of the autonomous machine vision category.

The problem with injection molding

A technician calibrates the 'Inspekto S70'.

© Inspecto

Plastic injection molding poses a number of problems for conventional vision systems, as it is difficult to properly illuminate the highly reflective surface of plastic. Furthermore, if the background material, mold and plastic are all a similar color, the lack of contrast can severely affect the proper functioning of the vision system. Combine this with the fact that conventional systems can only inspect one type of product at a time - while manufacturers have to produce ranges of products in different colors and sizes - and it becomes clear why a fundamental change is needed in machine vision, Inspekto believes.

This is precisely the problem that Bosch faced. The company manufactures plastic connectors for vehicles in Waiblingen, in particular for spark plugs, devices and sensors. The plant has various production lines and injection molding machines. The plant uses conventional, state-of-the-art image processing solutions, some of which were developed in-house and some of which were created and integrated by external image processing experts. However, these solutions are unsuitable for some of the company's applications because the product to be inspected consists of black plastic that appears against a black background. Under these conditions, it is practically impossible to define the parameters for the quality assurance solution to detect defects.

Consequently, the connectors had to be inspected manually at the factory. During this tedious and repetitive task, inspectors can easily become inattentive and overlook defects. In addition, the costs for inspection personnel are considerable.

After reading about autonomous machine vision in an industry report, Bosch executives decided to contact Inspekto and inquire about a possible solution that could overcome the unique challenges of this situation.

Advertisement

Plug&Inspect technology

The 'Inspekto S70' can be mounted on any Bosch profile.

© Inspecto

Autonomous machine vision (AMV) is a new category of image processing for quality assurance purposes invented by Inspekto, a German company with Israeli DNA. In 2018, the company launched its first AMV system, the 'Inspekto S70'. According to the manufacturer, Inspekto S70 is the first stand-alone product for visual quality assurance, feeding and sorting. It is self-configuring, self-learning and self-adjusting. It can therefore be used for the inspection of any parts in various manufacturing sectors. 'Inspekto S70' is ready to use in minutes and contains everything the operator needs to start inspecting products immediately. Factory staff can set up and train the system without the need for external system integrators. And last but not least, the system only needs an average of 20 to 30 good samples to learn the characteristics of a defect-free product. So there is no need to set up parameters to train the system to recognize defects, as is the case with conventional solutions.

Pilot project in Waiblingen

Ofer Nir is CEO of Inspekto in Ramat Gan, Israel.

© Inspecto

Based on these characteristics, Bosch decided to carry out a pilot project in Waiblingen. The aim was to test whether the system could help where conventional solutions had failed. Inspekto and Bosch set up an application partnership and after just a few weeks, the first system was ready for use in series production.

"In our field of business, we are experts in machine vision, so it was not difficult to integrate the system. I am therefore confident that our colleagues in other Bosch plants will also be able to use the Inspekto AMV system without any problems," reports one of the development engineers at Bosch. "The initial installation of the pilot system took about an hour and less than a day in total, considering the small adjustments we made to optimize the speed of the process. It took another 20 minutes to learn the software program, and after that we only needed to connect the camera before we could start inspecting."

During the installation, the production line continued to run normally and there was no downtime in the plant. "As Inspekto S70 does not require any input from the control system, it can be installed without disrupting production," says one of the development engineers. "Initially, we ran the S70 parallel to the production line, where it was mechanically integrated but not connected to the machine. The system then got to know the characteristics of the product very quickly. All this happened without any impact on production."

The cost of the initial system paid for itself in less than a month. Because the pilot system worked so well, Bosch has ordered a system for subsidiaries abroad and another to be installed in an older application in Waiblingen.

  • Xing Icon
  • LinkedIn Icon
Advertisement
Advertisement

You might also be interested in

Advertisement

Deevio

End of manual visual inspection?

The widespread industrial use of artificial intelligence is still a long way off. Yet self-optimizing deep learning solutions are an important milestone on the way to the smart factory, which consists of AI-based production facilities and logistics...

read more...

Spectra

Scalable for BV applications

Industrial image processing ranges from simple applications in environmental perception such as positioning, measuring, identifying and sorting objects to more complex image processing-controlled automation functions.

read more...

Optometron

Check unordered objects

Optometron has added the Deep Learning module to its VIS-Vision software. This module identifies and assesses objects that are also randomly present without structure, i.e. regardless of the position, angle, size or illumination of the objects.

read more...
Advertisement
Advertisement

Cognex

Automated inspection

Until now, optical inspection in 3D was associated with high programming effort and costs, meaning that users had to resort to 2D solutions. A new type of laser-based 3D vision system is now set to put an end to this.

read more...
Advertisement
Advertisement
Advertisement
Advertisement
Subscribe to our newsletter
Advertisement
Back to home