Real Time Digital Twin
Our W2MO Digital Twin provides a real time location and recognition system that can be built without using technologies like UWB or RFID or similar. Adding recognition units to the ceiling of your warehouse allows the system to detect and localize all objects in the camera view with artificial intelligence.
The recognition units consist of 4, 8 or 12 parallel camera inputs connected to a powerful embedded AI computer. Each camera has a resolution of 5 MP and a 120-degree viewing angle for wide-angle view of the warehouse.
The operations area is stored as a true-to-scale Digital Twin in W2MO. Based on this, a camera plan is generated showing where the cameras as well as the AI computers should be placed on the warehouse ceiling.
When objects move into the view of one of the cameras, the detections are processed immediately directly on the GPU of the AI computer. This means that only position data will be transferred to the Digital Twin.
Knowing the location of each camera in reality and in the Digital Twin allows us to transfer each object position within an image to an absolute position on the Digital Twin.
Object Localization with AI
Our pre-trained neuronal networks can recognize standard logistical objects such as forklifts, persons and pallets etc. in real time. If special objects are present in your operation, we can easily add them to the neuronal networks. We can also tune them to cope with special light conditions or other custom requirements.
This means that the AI can detect and localize all relevant objects in an image. If you have similar looking objects, such as forklifts of the same brand, an ID marker allows for unique differentiation. These markers can be quickly mounted on any kind of forklift.
The use of AI also enables us to distinguish between different states of the same object, such as whether a forklift is loaded or unloaded, or if a person is wearing a security vest or not.
The application of advanced technologies allows us to not only localize the object on an image but even estimate its pose. In this way, the driving direction of the forklift and the lifting height of the fork can be determined, even if only one camera is used.
Data protection through person pixelation
The recognition unit’s AI computer handles all image data processing, with only position data transmitted to the Digital Twin.
In support mode the cameras can be configured to allow viewing of the video stream, whereupon all detected persons are pixelated on the recognition unit. This avoids any personal data processing.