Beside the required components for the automated flight such as flight control, navigation and sensors for obstacle perception, the auto-pilot is the central element on board an unmanned aircraft. This component has to make reliable decisions by its own, and they must be functionally safe and responsible for the operation. Further, a mission management system aims on risk and workload reduction of the human operator.
Research focus is the development and evaluation of automation techniques for motion planning and decision control. Our model-based approaches describe automatable processes which are capable of making critical decisions in real-time. Beside a formal description of the basic functionality with its test methods, this means an implementation of an extensive procedural control, higher automated behaviors (Fly home to basis station, object search and tracking, flight through dangerous areas) and monitoring functions. The concept allows to synthesize basic maneuver items (e.g. take-off, landing, hover flight) and to combine them into a more complex mission plan. Further, a monitoring component on board allows the use of planning modules which can adapt to the situation, e.g. to modify the route or the task ordering. This modern and transparent architecture allows error handling, maintenance and expandability.
Our approaches include methods of machine-based mission planning, enabling automatic decision-making for high-level task orders. A two-stage planning approach, one for path planning and the other for task planning, allows planning almost optimal missions in complex 3D environments on-line. Generated from point samples, a graph is used for path search. Additionally, the given input tasks (e.g. single way points or search areas) can be quickly optimized with respect to the overall mission (e.g. mission duration).
This whole concept for the integration of 3D obstacle mapping and the on-board decision system includes the use of a three-dimensional environment model for the automatic calculation of collision-free flight trajectories, for the optimization of the behavior in the case of detected obstacles, and for the intelligent process control and automated flight planning from take-off to landing. Our evaluation includes standardized mission scenarios that allow benchmarking with our international partners.