Dynamic state charts: Composition and coordination of complex robot behavior and reuse of action plots

One of the challenges in service robotics is the integration and coordination of algorithms and abilities to make a robot perform complex tasks. Task experts of a domain should be able to compose more complex tasks (e.g. make coffee) out of basic tasks (e.g. first detect cup, grasp it,...) without the need to know about their details. State charts are adequate for robot behavior coordination in many applications but lack dynamic mechanisms to manage the complexity of real world tasks. This paper proposes Dynamic State Charts for composition and coordination of complex robot behavior. Dynamic states refine their content (sub behaviors) to execute from alternatives at runtime to manage the complexity of real world tasks. Reusable task plots are bundled with software components and are provided in a repository ("robot app store"). Domain experts can reuse these bundles off-the-shelf in a building blocks manner for their specific application. This supports the separation of\´ roles during the development process.