The RACE Project: Robustness by Autonomous Competence Enhancement

Abstract

This paper reports on the aims, the approach, and the results of the European project RACE. The project aim was to enhance the behavior of an autonomous robot by having the robot learn from conceptualized experiences of previous performance, based on initial models of the domain and its own actions in it. This paper introduces the general system architecture; it then sketches some results in detail regarding hybrid reasoning and planning used in RACE, and instances of learning from the experiences of real robot task execution. Enhancement of robot competence is operationalized in terms of performance quality and description length of the robot instructions, and such enhancement is shown to result from the RACE system.

Metadata

Authors/Creators:

Hertzberg, J
Zhang, J
Zhang, L
Rockel, S
Neumann, B
Lehmann, J
Dubba, KSR
Cohn, AG
Saffiotti, A
Pecora, F
Mansouri, M
Konečný, S
Günther, M
Stock, S
Lopes, LS
Oliveira, M
Lim, GH
Kasaei, H
Mokhtari, V
Hotz, L
Bohlken, W

© Springer-Verlag Berlin Heidelberg 2014. This is an author produced version of a paper published in KI - Künstliche Intelligenz. The final publication is available at Springer via http://dx.doi.org/10.1007/s13218-014-0327-y. Uploaded in accordance with the publisher's self-archiving policy.

Keywords:

Task Plan; Feature Extraction; Compositional Hierarchy; Constraint Satisfaction Problem; Action Primitive; Conceptual Knowledge; Robot Behavior; Object Category; Hybrid Reasoning; Plan Generation; Ego Network; Graph Similarity; Plan Execution; Object Perception; Planning Domain

Dates: