Understanding manual work from a computer's perspective

In the previous edition of newsletter, we took a brief look at importance of worker guidance, and how it can assist the worker adapt to changing conditions and new variants in manufacturing, but important question to ask yourself is how does system such as TRiMiTi understand what the worker is supposed to do, or work with different variants and such.??

The simple answer is #TRiMiTi Workorder. Here at NEXUSTEC GmbH , we have developed a ground-up concept of workorder, that is used to explain to the system the process in a fashion similar to the human understanding of the process. Currently the workorders can be used for following things:?

• The actions worker is supposed to take, such as pick up part, place part, check the sequence of placed parts etc.?
• Inline component check of the parts for variant recognition such as screw size,??
• Inline quality check of parts surface inspection of parts for scratches, or colour of the parts?
• Interfaces with external systems such as barcode readers, MES system, EC-Tool, ERP systems etc.??

All of these features make the Workorder a powerful tool as it combines actions, quality checks and system integration all in one. We have developed a custom programming language for workorders, this allows us to control the various features mentioned above and keep the API simple and clean for use. Writing of the workorders is simplified with template based modular features, where multiple functionalities are combined in a single template, so an application engineer can easily use TRiMiTi API without any background in programming. This is done so that process engineers with their expertise in process design & optimization can also program TRiMiTi to monitor and guide the workers using those processes.??

The seamless combination of systems engineering, embedded systems, distributed control architecture and software / IT know-how in one is what makes it easy to bring any process and workstation in TRiMiTi world in the matter of hours. TRiMiTi can work in already existing environments, without the necessity to change the workstations, and maintain the ergonomics of work environment that workers are well versed with. For this purpose, We have decided to choose Vision as a primary sensor, as this is what we humans use in such environments.

Using any additional sensors with TRiMiTi such as temperature sensors, proximity sensors, laser profile sensors etc are obviously possible but these are used as secondary sensors instead. Vision or camera based processes are easier to integrate as long as we have the same point of view of the worker, we can adapt & monitor the process with very little time-to-live. TRiMiTi systems for verification of Poke-Yoke tray assembly in the pre-assembly areas can be installed and commissioned successfully in under 48 hours, with this we ensure that no worker error is ever propagated down the production line ever again.??

In case of Vision based processing, we need to follow a simple guideline to solve the problems faced by machine vision experts over the last 2 decades. One of the reasons TRiMiTi is able to offer to many versatile solutions is because we have been able to manage the vision processing in real time without overloading the existing hardware systems. We have split the processing in two parts, classical machine vision and ML based machine vision, and based on the use case a specific feature from TRiMiTi may be utilized for that.

Simple verification such as checking color of the part, or the shape of the specific part can be done using this combination of ML and classical machine vision algorithms. But the most complex checks, such as surface inspection for variety of defects such as imperfections, dents and scratches can be done easier with an ML based vision system.??

With the split like this on fundamental level, we have made our system more robust and versatile to work in different working environments and address maximum number of possible scenarios. In the following article, we will take a deeper dive into the Vision based guidance for object detection & tracking.