Using AI camera vision for smart buffer management

How can twinzo digital twin be used to get rid of invetories and solve the biggest AMR/AGV problem

A lot of our customers are looking into increasing their automation footprint, especially in logistics, where this area was neglected for far too long. 

Quite often, when looking for a solution, they come to us, to see how we can help. Unfortunately, we do not manufacture AGVs or AMR. This interaction, on the other hand, provides us with a unique view of the factory and its problems when trying to implement this advanced technology.

Especially in the CEE region, the biggest decision point is ROI. We see that the required ROI for AGVs/AMRs is getting shorter to 12 - 18 months, max 24. Before the covid crisis, it was not unusual to have ROI 36 months and still go with the project.

The only way to get an ROI like that is by completely replacing existing human drivers and forklifts or tuggers with the AGV/AMR. If the driver has to remain in the factory, there is no chance to hit the ROI target.

And this fact is where the problem lies. Most of our customers work in brownfield plants, built decades ago and filled with many different production technologies. They are unable to completely change their production and logistic processes to fit the requirements of AGV and AMRs. This is true, especially in Warehouses and production buffers. 

They are expecting to utilize available space to its maximum possible capacity, and that is only available through a very dynamic setup. Dynamic positions, that are used by many processes. 

Through this experience, we've found the biggest downside of any AGV/AMR, and that is work within a hybrid warehouse with dynamic spaces, wherein one place both people and machines have to pick and drop off material. Seamlessly. AGV/AMR has no idea of reality ahead, and are not able to react on their own to changing environment. If a machine is going to pick a material, it expects that the material will be at exact position A. Same is expected with the drop-off location. The machine has predefined drop-off and pick-up locations and can move materials between those.

What happens if during the mission some human will put a box to position B? Or take material from position A ahead of the machine? The machine will not be able to handle this situation.

So the biggest problem of direct AGV/AMR implementation is, that you need to have a human at the begging of the route, to prepare the stock to be picked or to have a human driver at the warehouse to take the delivered material and stock it properly into dynamic position system. And this will destroy any ROI the customer was hoping for because they did not reduce the workforce at all.

Fortunately, we have come with a solution to this problem. One of the biggest strengths of twinzo - digital twin is the ability to recreate the reality of the space in digital based on multiple available data sources. As you can see in the video, the digital twin can combine information from RFID technology for stock identification (or can take material identification from eKanban system used to pick the material), positioning systems used to track forklifts or AGVs/AMRs, AI camera vision to detect and enhance the position of the material in a monitored area. With this combination, it holds real-time information about the area and especially what is exactly where. What position is currently occupied and with what material?


twinzo is also extended through different applications, one, in particular, is the Automatic Ordering system (AOS) that can translate sensor inputs (like the push of a button) to an active work order. When this application is connected, it can create and manage delivery orders based on a digital understanding of the space. So to create a mission for AGV and dynamically change the drop-off destination based on the actual situation ahead.

Human already stocked some material to a location where AGV/AMR was supposed to go? No problem. AOS will just change the destination to the next available one.

We at twinzo believe, that every problem has a solution. The only requirement is not to be bound to one technology and one approach, but to be willing to experiment and think outside of the box. This approach enables us to solve problems like this one efficiently and in a way, that the cost of the solution doesn't destroy customers' ROI and is payable.

TUTORIAL: Scan your model with iPhone