Single-Shot Learning and Scheduled Execution of Behaviors for a Robotic Manipulator

In order to introduce robots into the household domain, they must be easy to program and be able to perform various tasks without being reprogrammed every time. In this paper, we address both issues. First, we present a behavior-based architecture, which is capable of storing, executing, and switching between various manipulation tasks of a robotic manipulator. Second, we present a single-shot learning from demonstration, to allow the user to extend the set of known behaviors by guiding the robot through new tasks.