Factories must be optimized faster in the future and the Institute for Integrated Production Hannover (IPH) is working hard to make this happen. In the future, a flying robot should be able to map the factory. For the research project ‘Instant Factory Maps’, researchers are still looking for industry-partners. Drones with cameras like these must simplify and speed up the mapping of factories in the future.
So far, it takes weeks or even years to capture the layout of a factory. In addition, all machines, warehouses and corridors are measured and detailed on a map. Many factory layouts have grown historically but, over time, the processes will not be more efficient. There could be a lot of optimization, but especially small and medium-sized companies are dreadful of the effort. The preparation alone is extremely time consuming. The complete factory hall and the warehouses must be measured manually and then entered into a computer model. Only then can the actual optimization begin.
Drone as a measuring instrument
In the future, a drone can determine the factory layout from above. 3D cameras or laser scanners on board measure the complete factory hall. The data is then processed and analyzed on a computer. Everything is automatically recognized. The data eventually creates a three-dimensional model that can be edited directly, for example in a CAD program.
A layout can therefore be made in a few hours, rather than a lenghty process of weeks. This is especially true for small and medium-sized companies who want to set up their production more efficiently but cannot invest much time and money. With the automated layout determination, the IPH can offer its services cheaper in the future and show its customers results in a short period of time.
But first, the researchers need to bypass two more obstacles. First of all, the drone must be able to accurately determine its position at any given moment. Outdoor measurement is GPS-driven, but that does not work in closed spaces.
Therefore, the researchers need to develop new technology. For the orientation, for example, a radio station can be used, with which the drone is connected via WLAN. If it is factory-set, the position of the drone can be calculated from the distance and the angle to the transmitter. It would also be possible to provide the drone with accelerometers and thus calculate the direction in which the vessel moves.
A third option is location determination through a so-called SLaM algorithm: Simultaneous Localization and Mapping. The drone films the factory hall, compiles a map of the images and recognizes its own position on that map. Mobile robots can already do this; the scientists only have to investigate whether the location in this application is sufficiently accurate.
Afterwards, an algorithm needs to interpret the images correctly and reliably recognize what is a machine and what is, for example, a warehouse shelf. For this purpose, the researchers are working on data processing software. To date, there are only algorithms that can create a three-dimensional image from multiple individual recordings. However, they can not interpret the images yet.
The IPH wants to develop an algorithm that learns from experience. The first factory layouts determined by the drone, will still be examined by an expert, and he will still mark machinery and warehouse positions. The algorithm recognizes these patterns and will quickly be able to automatically interpret the data.