Managing execution variants in task coordination by exploiting design-time models at run-time

The development of service robots has gained more and more attention over the last years. Advanced robots have to cope with many different situations and contingencies while executing concurrent and interruptable complex tasks. In particular, mobile manipulation tasks increase the complexity. To manage the raising number of tasks and execution variants in complex environments there is a tremendous need for context and situation dependent composition and selection of reusable skills. This requires explicit descriptions of relevant properties and parameters of the robot, its resources and its capabilities. Different views on partial aspects of a robot system (mechanical, electrical and even software) can be provided by different models as is already common practice at design-time. However, these design-time models also need to be accessible at run-time to support run-time reasoning of the robot in order to adequately compose its skills and assign resources. That requires to extract useful information out of the design-time models and to transform it into representations which can be exploited at runtime. We present an approach to exploit information provided via design-time models (e.g. software components, simulation, planning) for run-time decision making. It allows for more informed decisions on how to compose action plots at run-time in order to manage the huge amount of different execution variants in service robotics.