CNNs for State Estimation of Articulated Objects
[ 1 ] Instytut Robotyki i Inteligencji Maszynowej, Wydział Automatyki, Robotyki i Elektrotechniki, Politechnika Poznańska | [ P ] employee
[2.2] Automation, electronics, electrical engineering and space technologies
2022
chapter in monograph / paper
english
- articulated objects
- robot perception
- deep learning in robotics
EN In this paper, we deal with the problem of state estimation of articulated objects during robotic interaction. The robot equipped with an RGB-D camera has to estimate the joint position and rotation of the articulated object when manipulating the object. The problem of accurate state estimation is challenging due to the properties of the RGB-D sensor. The known solutions require some assumptions about the shape of the objects. In this paper, we propose the application of Convolutional Neural Networks to the state estimation of articulated objects from two pairs of RGB-D images.
112 - 115
20