Object Pose Estimation by Contact with Orientation-Sensing Robotic Fingertips

Pose estimation of an object is important for manipulation tasks, however the object may be partially or even fully occluded while the robotic hand is retracting upon it. To estimate the post of an object while grasping, we introduce novel, low cost surface-orientation sensors for robotic fingertips, and an algorithm that reduces the complex pose estimation problem using the information at hand, resulting in a real-time estimation of the object’s pose. 


Full paper available here