Abstract
We present the concept of Bayesian Tool Affordances as a solution to estimate the suitable Action and novel Tool as well as suitable Action for the given tool to realize the given effects to the robot. We define Tool affordances as the "awareness within robot about the different kind of effects it can create in the environment using a tool. For reliable prediction, inference and planning capabilities robot requires understanding the bi-directional association of executed Action, with the relevant properties of the Tool and the resulting effects. But acquiring such an understanding is difficult due to limited learning samples, uncertainty, redundancy and irrelevant information. Thus we propose Bayesian leaning of Tool Affordances to solve above problems and hence our approach is termed as Bayesian Tool Affordances. The estimation results are presented in this paper to validate the proposed concept.
