Programming by configuration in service robotics

Robotic systems operate in an increasing amount of domains. They perform tasks in work processes and assist in daily life activities. The target scenarios for these systems range from simple toys to complex manufacturing processes and dynamic interactions with humans. All these systems have to be programmed to properly perform their operations. Due to their different nature, also the complexity of the desired behaviour changes. During the development in a team the different levels of programming knowledge and experience of the target scenarios of the members have to be regarded. Different tools were created to aid this development process. One popular programming paradigm is to graphically model algorithms. Graphical designers hide most of the internal complexity from the modeller developing the algorithm, however still no tool exists in the current state of art to model arbitrary algorithms for any defined data structures. A new graphical modelling language model to overcome these limitations is developed in this thesis. It bases on the concept of function-block-networks. Function blocks are parametrizable black boxes that can perform defined functionality. The utilized function blocks of the algorithm and how they are interconnected in the network define the performed task. Algorithms may work on arbitrary data structures. Parameters of any function block can be automatically set by defined rules, using an integrated general feedback system. A graphical designer permits easy modelling and testing of the function-block-networks. Developed algorithms can be directly embedded into program code. They are interpreted and can thereby be processed inside the program. An implementation demonstrates the feasibility and benefits of the proposed language model.