Automating acceptance tests for sensor- and actuator-based systems on the example of autonomous vehicles

In projects dealing with autonomous vehicles which are driving in different contexts like highways, urban environments, and rough areas, managing the software's quality for the entire data processing chain of sensor- and actuatorbased autonomous systems is increasingly complex. One main reason is the early dependency on all sensor's raw data to setup the data processing chain and to identify subsystems. These sensors' data might be extensive, especially if laser scanners or color camera systems are continuously producing a vast number of raw data. Moreover, due to this dependency the sensors' setup including their respectively specified mounting positions and calibration information is also necessary to gather real input data from real surroundings' situations of the system. This is even more important before actually starting to integrate independently developed subsystems for carrying out tests for the entire data processing chain. To reduce this dependency and therefore to decouple tasks from the project's critical path, an approach is outlined in this thesis which was developed to support the software engineering for sensor- and actuator-based autonomous systems. This approach relies on customer's requirements and corresponding customer's acceptance criteria as well as the decoupling of the software engineering from the real hardware environment to allow appropriate system simulations.