The Birth of the Object is a Cognitive Robotics paradigm in which a naive robot, ignorant about its surrounding, discovers the objects that populate it and the affordances thereof by interacting with its environment [1,2].
This mechanism is initiated by a visually induced grasping reflex [3], that allows the robot to probe at unknown parts of its environement for the presence of graspable objects. When such a grasp is successful, the robot knows that it has gained physical control over an object. It then starts manipulating the object to visually study it and learn more about it. This visual observation allows to build a 3D model of the object's shape and appearance [4,5,6]. The system stores this new shape, and the associated successful grasp actions as a new object class and pose estimation allows to re--apply successful grasp actions when encountering a known object [1,2]. Further attempts at grasping the object in different ways are recorded and the system progressively improves its knowledge of how the object can be manipulated.
This exploratory process that allows the robot to progressively discover objects in its environment by interacting with it, is coined the "Birth of the Object".
This research was conducted at the Cognitive Vision Lab, of the University of Southern Denmark, as part of the European project PACO-PLUS. Acknowledgements: Dirk Kraft, Mila Popovic, Emre Başeski, Justus Piater, Renaud Detry and Norbert Krüger (see references for details).
[1] D. Kraft, R. Detry, N. Pugeault, E. Başeski, F. Guerin, J. Piater, N. and Krüger (2010). Development of Object and Grasping Knowledge by Robot Exploration. IEEE Transactions on Autonomous Mental Development 2(4):368–383. (doi) (pdf)
[2] Kraft, D., Pugeault, N., Baseski, E., Popovic, M., Kragic, D., Kalkan, S., Wörgötter, F., and Krüger, N. (2008). Birth of the object: Detection of objectness and extraction of object shape through object action complexes. Special Issue on ``Cognitive Humanoid Robots'' of the International Journal of Humanoid Robotics (IJHR), 2008, 5(2):247–265. (doi) (pdf)
[3] Popovic, M., Kraft, D., Bodenhagen, L., Başeski, E., Pugeault, N., Kragic, D., Asfour, T., and Krüger, N. (2010). A strategy for grasping unknown objects based on co-planarity and colour information. Robotic and Autonomous Systems 58(5):551–565. (pdf) (doi)
[4] Pugeault, N., and Krüger, N. (2011). Temporal Accumulation of Oriented Visual Features. Journal of Visual Communication and Image Representation, 22(2):153–163. (pdf) (doi)
[5] Detry, R., Pugeault, N., and Piater, J. (2009) A Probabilistic Framework for 3D Visual Object Representation. IEEE transactions in Pattern Analysis and Machine Intelligence (PAMI) 31(10):1790–1803. (pdf) (doi)