Machine vision has long played a part in the promotion of efficiencies in the process industries, with image- based inspection and analysis ensuring greater process control.
Yet ‘traditional’ systems have come with an understandably hefty price tag plus other considerations, points out machine vision pioneer Inspekto.
Implementation has required a project approach involving the selection of diffuse components, says Inspekto VP of operations Miki Gotlieb.
“Once the integrator receives the components, they can then integrate them with one another, alongside the tedious task of preparing the software selected and then building a hard- engineered solution on the shop floor, before beginning the software training process,” explains Gotlieb.
Lack of adaptability has meant their use has been restricted for smaller enterprises that lack the budget to invest heavily
Miki Gotlieb, VP of operations, Inspekto
For a simple application implementation can take several weeks to a few months. For a more complex solution, typical time frames are several months and, in some cases, go up to an entire year.
“Lack of adaptability has meant their use has been restricted for smaller enterprises that lack the budget to invest heavily.”
The good news for process companies balancing need and investment capability, is that beneficial changes are underway in the available technologies.
SICK is making use of apps to offer its first vision camera with deep learning capabilities pre-installed. Its compact 2D Vision Camera enables quality inspections of complex or irregular-shaped goods and assemblies that have previously defied automation, allowing vision classifications using artificial intelligence for a fraction of the time and cost to program pre-set rules and patterns.
Applications include sorting fresh food, checking the orientation of timber profiles by recognising the annual ring structure and inspecting the integrity of solders in surface mount assemblies.
Neil Sandhu, SICK’s UK product manager for imaging, measurement and ranging, explains: “By embedding the Intelligent Inspection App onto SICK’s Inspector P621 deep learning camera, SICK has made it possible for users to purchase a package that uses artificial intelligence to run complex vision inspections with ease.”
Users collect images of their product in realistic production conditions with SICK Inspector 621’s in-built image capture tool. These are uploaded to the Cloud where the image training process is completed by the neural network.
The deep learning solution is downloaded to the camera where it can begin to take decisions. Results are output to the control system as sensor values and digital I/O with no need for separate training hardware or software, saving on implementation time and cost.
Comments Sandhu: “So users can automate complex vision inspections for a much lower cost of ownership. They can now consider automating quality inspections of products or goods that have just proved too difficult previously.”
While two-dimensional imaging remains common in these sectors, many users are beginning to realise the benefits of using a three-dimensional sensor, points out Clare Rathsack, business unit manager at ECCO Sensors SmartRay GmbH [pictured]. “This provides extra information about the scanned object, including depth information that, on 2D imaging, will only appear as shade. 3D scanning thus enables more precise metrology and a truer picture of the object.”
The 3D sensor measures the profile of the object. Relative movement between the object and the sensor creates a full 3D model. In addition, sensors capture the intensity and laser line thickness information from the part, for required 2D information is available, alongside quality information about each 3D point.
A laser line generator is used as the source of illumination/ lighting in 3D sensors to scan the surface of an object. Due to the properties of the laser, typical applications require very low exposure times, in the range of micro-seconds.
The laser line image can be used to judge the reflection properties of the surface being scanned and can, in turn, validate the plausibility of the data received.
Analytical software takes information from the profile image generated by the sensor and creates a point cloud that can then be saved in ASCII or CSV format.
This information can then be used for quality inspection and measurement, reverse engineering and rapid prototyping.
“By using 3D sensors in their machine vision systems, users will have access to – literally – a greater depth of data, enabling them to improve product quality, guide automation more effectively and reduce production costs,” explains Rathsack.
For Inspekto, the transformation in the scope of technologies is part of a process of industry ‘democratisation’ – providing systems that do much more and faster at a cost-effective price that puts machine vision within the grasp of SMEs or corporations.
Leading this campaign is the company’s Inspekto S70 Autonomous Machine Vision system. Launched in Stuttgart in 2018, it is designed to address those longstanding challenges facing users, says Gotlieb.
They can now consider automating quality inspections of products or goods that have just proved too difficult previously
Clare Rathsack, business unit manager at ECCO Sensors SmartRay GmbH
These include lack of adaptability, with machines designed for specific locations and accompanied with unique diagnostic tools; also, a reliance on stockpiling components to offset lack of user diagnostic knowledge when problems occur.
By contrast, the Inspekto S70 is designed for any quality assurance, gating or sorting application with an arm sufficiently flexible to be installed on any machine, at any angle and position to get the required field of view and at any location. The system’s visual sensor, comprising all camera, lenses and lighting components, has a wide range of capabilities to suit different applications.
As an off the shelf product rather than an integrator project, it requires none of the time taken for traditional machine vision solutions ordering components from different vendors.
“This is drastically different from machine vision solutions of the past, where moving a solution from location to another would be either impossible or require a very notable bill of waiting time and cost for an integrator to make the necessary change.”
Crucially it can perform self-diagnosis to identify the issue, significantly reducing the MTTR, setting up its own parameters and using its intelligent capabilities to test itself and report back to the QA manager, who requires no specific training.
Inspekto asserts the purchase price is around a tenth of a traditional solution, while its installation time is claimed to be 1,000 times faster and without the help of an integrator.
More recently the firm launched the second generation Inspekto S70 Gen.2, capable of deployment in more use cases and able to reliably inspect highly reflective parts.
Built-in 5,000K LEDs allow the system to autonomously control the direction of the illumination and take several images with varying light direction and intensities, fused to create a single reflection-less HDR image. Additionally, it can inspect moving objects running at up to 0.75m/seconds, while inspection cycle time has been shortened to less than 0.5 seconds in most cases, enabling the system to inspect up to three parts per second.
Inspekto CTO Yonatan Hyatt defines the benefits: “[It] allows small manufacturers to enjoy the benefits of automated quality inspection, without losing their agility and flexibility. Until not long ago, such technologies were reserved for large corporations.”