TY - GEN
T1 - Maximally informative interaction learning for scene exploration
AU - Van Hoof, Herke
AU - Kroemer, Oliver
AU - Ben Amor, Heni
AU - Peters, Jan
PY - 2012
Y1 - 2012
N2 - Creating robots that can act autonomously in dynamic, unstructured environments is a major challenge. In such environments, learning to recognize and manipulate novel objects is an important capability. A truly autonomous robot acquires knowledge through interaction with its environment without using heuristics or prior information encoding human domain insights. Static images often provide insufficient information for inferring the relevant properties of the objects in a scene. Hence, a robot needs to explore these objects by interacting with them. However, there may be many exploratory actions possible, and a large portion of these actions may be non-informative. To learn quickly and efficiently, a robot must select actions that are expected to have the most informative outcomes. In the proposed bottom-up approach, the robot achieves this goal by quantifying the expected informativeness of its own actions. We use this approach to segment a scene into its constituent objects as a first step in learning the properties and affordances of objects. Evaluations showed that the proposed information-theoretic approach allows a robot to efficiently infer the composite structure of its environment.
AB - Creating robots that can act autonomously in dynamic, unstructured environments is a major challenge. In such environments, learning to recognize and manipulate novel objects is an important capability. A truly autonomous robot acquires knowledge through interaction with its environment without using heuristics or prior information encoding human domain insights. Static images often provide insufficient information for inferring the relevant properties of the objects in a scene. Hence, a robot needs to explore these objects by interacting with them. However, there may be many exploratory actions possible, and a large portion of these actions may be non-informative. To learn quickly and efficiently, a robot must select actions that are expected to have the most informative outcomes. In the proposed bottom-up approach, the robot achieves this goal by quantifying the expected informativeness of its own actions. We use this approach to segment a scene into its constituent objects as a first step in learning the properties and affordances of objects. Evaluations showed that the proposed information-theoretic approach allows a robot to efficiently infer the composite structure of its environment.
UR - http://www.scopus.com/inward/record.url?scp=84872319695&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84872319695&partnerID=8YFLogxK
U2 - 10.1109/IROS.2012.6386008
DO - 10.1109/IROS.2012.6386008
M3 - Conference contribution
AN - SCOPUS:84872319695
SN - 9781467317375
T3 - IEEE International Conference on Intelligent Robots and Systems
SP - 5152
EP - 5158
BT - 2012 IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS 2012
T2 - 25th IEEE/RSJ International Conference on Robotics and Intelligent Systems, IROS 2012
Y2 - 7 October 2012 through 12 October 2012
ER -