Thosar, Madhura and Mueller, Christian A. and Jaeger, Georg and Pfingsthorn, Max and Beetz, Michael and Zug, Sebastian and Mossakowski, Till
Proceedings of the 35th Annual ACM Symposium on Applied Computing
When performing task in open-ended environment, a robot may not have access to conventional tools required in the task. In such case, an ideal response for the robot would be to find a suitable substitute. In this paper, we present an approach to tool substitution where robot-centric knowledge is generated in an unsupervised manner about objects from multi-modal sensory data captured by a robot. This robot-centric knowledge is subsequently exploited to identify a substitute from available objects observed in the environment.