Affordance of Object Parts from Geometric Features

As robots begin to collaborate with humans in everyday workspaces, they will need to understand the functions of tools and their parts. To cut an apple or hammer a nail, robots need to not just know the tool’s name, but they must localize the tool’s parts and identify their functions. In this extended abstract, we give an overview of our work on localizing and identifying object part affordance. We present a framework which provides 3D predictions of functional parts that can be used by a robot, and we introduce a new RGB-D Part Affordance Dataset with 105 kitchen, workshop, and garden tools. We analyze the usefulness of different features, and show that geometry is key for this task. Finally, we demonstrate that the approach can generalize to novel object categories, so robots like PR2, Asimo, and Baxter could use tools never seen before.