Planning with Noisy Probabilistic Relational Rules

13 years 2 months ago

Download www.jair.org

Noisy probabilistic relational rules are a promising world model representation for several reasons. They are compact and generalize over world instantiations. They are usually interpretable and they can be learned eﬀectively from the action experiences in complex worlds. We investigate reasoning with such rules in grounded relational domains. Our algorithms exploit the compactness of rules for eﬃcient and ﬂexible decision-theoretic planning. As a ﬁrst approach, we combine these rules with the Upper Conﬁdence Bounds applied to Trees (UCT) algorithm based on look-ahead trees. Our second approach converts these rules into a structured dynamic Bayesian network representation and predicts the eﬀects of action sequences using approximate inference and beliefs over world states. We evaluate the eﬀectiveness of our approaches for planning in a simulated complex 3D robot manipulation scenario with an articulated manipulator and realistic physics and in domains of the probabilist...

Tobias Lang, Marc Toussaint

Real-time Traffic