How Does a Neural Processing Unit Make AI Accessible to Manufacturers?

Resource Type: Blog |

The SIMATIC S7-1500 TM NPU (neural processing unit) is the new AI (artificial intelligence) module from Siemens.  It enables you to bring AI into any SIMATIC S7-1500 PLC without a separate computer. This new technology was released in late 2019; Patti Engineering is one of the few integrators that has experience with it now.

In a recent project, we connected the Siemens NPU for a pick and place station.  Typically, the robot would know which object to pick using a vision system and creating a library of objects with exact measurements and/or other features like colors.  In this case, the NPU allowed the robot to learn to recognize anything – anything – within a defined area.  We defined a coordinate system so it would know where the bin was.

The NPU used the information from the camera pictures to determine the size, height, and location of something in a bin.  The NPU then helps the robot pick up any item within the bin and move it to the transportation device to go to the next station.  For example, in this situation, the robot could pick up boxes, tubes, and t-shirts without making a change in tooling or reprogramming robot paths.

This, obviously, has tremendous implications for bringing new products to market, and just overall flexibility for the facility.  It is also a simpler solution.  In other pre-NPU stations within the facility, 6 cameras with all the associated wiring were used to perform a similar function.  In this new station, only one camera was needed.

This new technology is big news for pick and place options, but AI can be used for many different applications.  For example, also with a camera, this could be an improvement for vision inspection stations.  It can be somewhat challenging to program in exact pass/fail parameters.  With AI in the NPU, the system could learn the many varieties of what is a pass or fail.

Another potential use case would be palletizing.  When boxes may vary in size, the boxes could be measured while on the conveyor, then an optimal layout could be decided through the NPU.

There are many other potential use cases for AI in manufacturing facilities.  This is a huge leap forward especially as large portions of the manufacturing workforce is nearing retirement and manufacturers are facing losing decades of institutional, experiential knowledge.  Making AI broadly accessible to manufacturing facilities is one way to fill that knowledge gap.

Related categories: Blog Industry 4.0 / Digitalization Siemens

Terrance Brinkley's Bio

Michigan Director of Operations

With a natural affinity for control systems integration, Terrance Brinkley has been an asset to Patti Engineering since 2004 and now leads his team as the Director of Michigan Operations. A native of Pontiac, Michigan, Terrance graduated from Michigan State University with a Bachelor of Science degree in both Electrical Engineering and Computer Engineering.