News

AI-based Machine Vision: A Vision Sensor that Teaches Itself

November 18, 2022 by Damond Goodwin

SensoPart has recently developed and released a new AI-based vision system, utilizing their advanced VISOR Object vision sensor, a system with advanced vision capabilities for detection and quality control.

German-born company SensoPart creates photoelectric sensors and image processing for factory automation. As a leader in vision and image processing, they recently looked to expand sensing capabilities for their customers. With the use of artificial intelligence (AI), SensoPart created a more powerful vision system capable of operating in conditions that many vision systems have traditionally struggled with. 

 

The VISOR Object AI is SensoPart's latest sensor solution. Image used courtesy of SensoPart

 

AI Sensor Technology

SensoPart decided the use of AI could be beneficial in creating an avenue to avoid the use of switching sensors that are traditionally overwhelmed by uneven parts or discrepancies in products being viewed. The result is the new VISOR Object AI, a sensor that maintains status as a best in class AI-aided vision sensor. SensoPart hopes to help manufacturers expand the use of their vision system with new possibilities opened up as a result of the newly developed technology.  
 

VISOR Object Vision Sensor

SensoPart's earlier-released VISOR Object photoelectric sensor, the foundation for the new VISOR Object AI, is capable of detecting problems that arise from many different common scenarios. These range from parts in the wrong sequence or orientation to defective parts and parts in the wrong position, all while maintaining an overview of the area. This can all be done almost instantaneously so manufacturers don't have to worry about bottlenecks created by slow processing times. 

 

Sensor Programming

Programming and configuration of the sensor can be completed with the click of a mouse, using SensoPart’s VISOR software. In addition, programmers can look into customizing lens correction, scaling, and image correction. A wide range of built in tasks are available including:

  • pattern matching

  • contrast detection

  • brightness

  • BLOB

  • caliper

  • gray threshold

  • contour detection

 

SensoPart VISOR Object AI photoelectric sensor

SensoPart’s VISOR Object photoelectric sensor is now availible with AI. Image used courtesy of SensoPart

 

Position Tracking

The VISOR Object maintains a best in class sensor rating because of the precision it creates when looking at position and orientation data. In order to help with scenarios where parts do not come to the VISOR Object uniformly, position tracking is available to help the system adjust to the differences in part location.

Once the position is found, the sensor is able to adjust parameters to the relative location of the part and does not need data for all possible locations. There is also a color edition of the sensor available that helps in applications where differences in color are important. It is able to detect minute differences between colors whenever necessary, helping with some quality control applications.

 

Simple and Advanced Detection

The VISOR Object sensor comes in either an advanced edition or a standard edition. The standard edition of the sensor can be used for simpler tasks where extra capabilities are not needed. For example, in measurement tasks, color detection, or presence detection. The advanced version of the sensor gives customers more capabilities for more advanced applications. Some applications for the advanced include more precise tacking for position control and completeness checks. It also includes more advanced color options and result calculations.  

 

SensoPart VISOR Object AI with Classification AI detector

The classification detector with the VISOR Object AI makes correctly classifying similar-looking objects easier than ever. Image used courtesy of SensoPart

 

VISOR Object AI

Using its VISOR Object advanced sensor, SensoPart has created a new vision system capable of capitalizing on AI technology to create a more state-of-the-art vision system for its customers. With the addition of built-in AI, the VISOR Object comes with a classification detector, making image processing simpler with the sensor teaching itself, typically needing no more than five sample images to base its learning on. 

The VISOR Object AI is capable of differentiating colors, shape, size, and quality. All of this can be done without strict part orientation since the system is able to find parts without the need for exact position. The data can be monitored through a "good" or "bad" judgment or split into 200 classes within the sensor, and all pass or fail data is stored for use downstream. 

With the addition of AI in their vision sensors, SensoPart expects fewer steps and expert knowledge will be needed for customers to integrate machine vision solutions in their business.