Human-to-Robot Handovers
(Essential Skills Sub-Track 4)
The real-time estimation through vision perception of the physical properties of objects manipulated by humans is important to inform the control of robots and perform accurate and safe grasps of objects handed over by humans. However, estimating the physical properties of previously unseen objects using only inexpensive cameras is challenging due to illumination variations, transparencies, reflective surfaces, and occlusions caused both by the human and the robot. Our sub-track for dynamic human-to-robot handovers is based on an affordable experimental setup that does not use a motion capture system, markers, or prior knowledge of specific object models. The sub-track focuses on food containers and drinking glasses that vary in shape and size, and may be empty or filled with an unknown amount of unknown content. The goal is to assess the generalisation capabilities of the robotic control when handing over previously unseen objects filled (or not) with unknown content, hence with a different and unknown mass and stiffness. No object properties are initially known to the robot (or the team) that must infer these properties on-the-fly, during the execution of the dynamic handover, through perception of the scene.
If you have any questions or enquiries, please contact Dr. Alessio Xompero and Prof. Andrea Cavallaro.
