Feb 06 2024
Flights for future - autonomous drones on a mission
by Ralf Hartmann on behalf of the AIMS5.0 consortium
With the attempt to take the European Industry to the next level of digitalisation, first fundamental research activities in AIMS5.0 prepare the way to get the 20 use cases off the ground bringing the future of a highly efficient fabrication to life. One of the striking examples is an autonomous indoor drone.
As a work result of work package 5, the drone will demonstrate an AI based improved connection between production and logistics. Developed by IPH, “Institut für integrierte Produktion Hannover” from Germany it will work as a scan device covering a complete production site of carmaker BMW to create a digital twin of it.
Daily scans for digital twins
The drone is able to navigate autonomously through a manufacturing hall catching every corner and measuring each millimetre. Other than a ground bound device, it can measure even closed robot cells or conveyor belts. Together with the metrics from a ground bound automated guided vehicle system (AGV), a robot dog, the gapless data collection enables an accurate three-dimensional reproduction for a digital twin with regular updates.
This makes it easier to restructure or rebuild production lines, for example if new products are to be manufactured. For this purpose, every machine, every robot and each workplace is recorded in the factory coordinate system. In the project, the IPH team first puts together the hardware deciding which types of cameras or sensors are suitable and how the drone must be designed in order to carry the payload. The team is also developing the software for autonomous drone flight.
Challenging flight planning
The main challenge is that the drone must determine its position at any time without using GPS, which does not work indoors. It must be able to detect obstacles and avoid collisions, not only with walls, but also with cables, glass panes or moving obstacles. The drone has to plan its flight route independently to explore every corner of unknown spaces, and in the shortest possible time so as not to disrupt production. Further, it may have to fly to a charging station independently or change the battery.
The autonomous flight planning will consider three different aspects. The speed should be optimised so that the drone can fully explore an unknown space in as short a time as possible. Then, the drone should fly in a photogrammetry-optimised manner, that is at the optimal height and at the optimal distance from objects as to to capture them completely and in three dimensions. Finally, flight planning is optimised for the laser scan. In this case, the drone must stop in the air at certain intervals and carry out the scan.
“Our aim is to create a drone where the mode of how to fly it can simply be set on the device,” IPH team member Hendrik Kumpe says. “Thus, companies can use one system for different purposes.” The plan is to fly the drone each night for a 24-7 update of the digital twin with a measurement accuracy of at least 2 centimetres with the drones / 2 mm with other equipment and measurement methods. Due to safety regulations, it can be active only when there are no persons in the hall, either during the night or during break times in a speed-optimised mode
Exploring the unknown
Besides, measuring production sites is only one of many other applications. In the future, autonomous drones may explore unknown spaces under hazardous conditions. They can search burning or collapsed buildings, inspect mines or measure the radiation in decommissioned nuclear power plants. For manufacturers the main benefit is a way easier planning, use or redesign of their production sites. They can use the existing space far more efficiently or rapidly adapt production lines to meet with the demands of changing markets.
In AIMS5.0, this research work will complete BMW’s previous efforts for a digital twin. In the end, also other manufacturers may benefit from the findings of the publically funded project in order to strengthen Europe’s competitive position.