Armin Mueller, Alexandra Kirsch, Michael Beetz
We propose an approach to transformational planning and learning of everyday activity. This approach is targeted at autonomous robots that are to perform complex activities such as household chore. Our approach operates on flexible and reliable plans suited for long-term activity and applies plan transformations that generate competent and high-performance robot behavior. We show as a proof of concept that general transformation rules can be formulated that achieve substantially and significantly improved performance using table setting as an example.
Subjects: 1.11 Planning; 17. Robotics
Submitted: Jun 27, 2007