Videos
Dynamic State Estimation for Cloth Folding
The video shows slowed down sequences of a robot dynamically folding a cloth. While the motion is preprogrammed, the initial position and stretching of the cloth will affect whether the robots ability to grasp the cloth at the end of the motion. Two examples are shown: one successful and one failed attemp.
Scene Understanding through Autonomous Interactive Segmentation
This video shows situations where the segmentation methods fails by grouping two objects in one segment. By letting the robot interact with the scene, the robot itself is able to disambiguate such situations. The work was presented at ICVS 2011.
Generating Object Hypotheses through Human-Robot Interaction
In this work we present how multi object segmentation and tracking can be done by giving the robot simple commands like add object, split object and detect motion. The work was presented at IROS 2011, where it received best paper award.
Integration of Visual Cues for Robotic Grasping
In this video the robot finds edges in a stereo image pair and reconstructs contours of objects in the image. By finding planar surfaces, the robot generates grasp hypotheses and grasps the objects. The work was presented at ICVS 2009.