TY - GEN
T1 - 3D Shape Perception from Monocular Vision, Touch, and Shape Priors
AU - Wang, Shaoxiong
AU - Wu, Jiajun
AU - Sun, Xingyuan
AU - Yuan, Wenzhen
AU - Freeman, William T.
AU - Tenenbaum, Joshua B.
AU - Adelson, Edward H.
N1 - Publisher Copyright:
© 2018 IEEE.
PY - 2018/12/27
Y1 - 2018/12/27
N2 - Perceiving accurate 3D object shape is important for robots to interact with the physical world. Current research along this direction has been primarily relying on visual observations. Vision, however useful, has inherent limitations due to occlusions and the 2D-3D ambiguities, especially for perception with a monocular camera. In contrast, touch gets precise local shape information, though its efficiency for reconstructing the entire shape could be low. In this paper, we propose a novel paradigm that efficiently perceives accurate 3D object shape by incorporating visual and tactile observations, as well as prior knowledge of common object shapes learned from large-scale shape repositories. We use vision first, applying neural networks with learned shape priors to predict an object's 3D shape from a single-view color image. We then use tactile sensing to refine the shape; the robot actively touches the object regions where the visual prediction has high uncertainty. Our method efficiently builds the 3D shape of common objects from a color image and a small number of tactile explorations (around 10). Our setup is easy to apply and has potentials to help robots better perform grasping or manipulation tasks on real-world objects.
AB - Perceiving accurate 3D object shape is important for robots to interact with the physical world. Current research along this direction has been primarily relying on visual observations. Vision, however useful, has inherent limitations due to occlusions and the 2D-3D ambiguities, especially for perception with a monocular camera. In contrast, touch gets precise local shape information, though its efficiency for reconstructing the entire shape could be low. In this paper, we propose a novel paradigm that efficiently perceives accurate 3D object shape by incorporating visual and tactile observations, as well as prior knowledge of common object shapes learned from large-scale shape repositories. We use vision first, applying neural networks with learned shape priors to predict an object's 3D shape from a single-view color image. We then use tactile sensing to refine the shape; the robot actively touches the object regions where the visual prediction has high uncertainty. Our method efficiently builds the 3D shape of common objects from a color image and a small number of tactile explorations (around 10). Our setup is easy to apply and has potentials to help robots better perform grasping or manipulation tasks on real-world objects.
UR - http://www.scopus.com/inward/record.url?scp=85062973132&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85062973132&partnerID=8YFLogxK
U2 - 10.1109/IROS.2018.8593430
DO - 10.1109/IROS.2018.8593430
M3 - Conference contribution
AN - SCOPUS:85062973132
T3 - IEEE International Conference on Intelligent Robots and Systems
SP - 1606
EP - 1613
BT - 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS 2018
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS 2018
Y2 - 1 October 2018 through 5 October 2018
ER -