In the DLR-RMC XRotor group we work on autonomous multicopter systems like quadro- or octocopters. Such flying systems will play an important role in the future, e.g. supporting rescue teams in disaster zones or in inaccessible areas. Nowadays, they already allow the 3D reconstruction of specific areas and help during the inspection of buildings.
We especially focus on the autonomy of our flying robots. Autonomous systems need to be able to operate independently also in environments where no external position reference (like GPS) is provided. Therefore, all the sensor data has to be acquired and processed on the flying platform. This is especially problematic if we think of the limited number of sensors and processing capabilities which can be carried on such lightweight and low-power robots. The development of algorithms which run in real-time despite these resource limitations is a major interest of our group.
Furthermore, a high data rate is required to control such a high dynamic system. The pose of a flying vehicle has to be continuously known, a dropout of the pose estimation or the underlying sensors would have fatal consequences. Unlike ground vehicles, flying robots cannot just stop and stand while the system recovers.
The requirement to develop efficient and at the same time reliable and robust algorithms, make flying robots to a quite challenging platform. Algorithms which run on such systems can in general also be used on other robots or intelligent devices (like smartphones).