TY - JOUR
T1 - Robust form-closure grasp planning for 4-pin gripper using learning-based Attractive Region in Environment
AU - Li, Xiaoqing
AU - Qian, Yang
AU - Li, Rui
AU - Niu, Xingyu
AU - Qiao, Hong
N1 - Publisher Copyright:
© 2019 Elsevier B.V.
PY - 2020/4/7
Y1 - 2020/4/7
N2 - In terms of the closure theory, for 3D objects, it usually requires at least 7 grasp points to ensure a form closure grasp, which is too strict for real applications. Instead, using a 4-point planar grasp is much more practical. In this paper, a robust form-closure grasping planning algorithm is proposed for a 4-pin gripper to obtain stable grasp points and improve the generalization to grasp objects that have not been seen before. Besides, a lightweight, 3-DoF (Degree of Freedom) 4-pin gripper based on our algorithm is designed for 3D object grasping. The proposed algorithm consists of two parts. First, based on Attractive Region in Environment (ARIE), the stability of the whole grasping process by obtaining form-closure grasp points is ensured. Second, considering the uncertainty of the environment, a learning grasp quality measurement is proposed to make evaluation of robustness for each group of grasp points. Our simulation and physical experiments are performed to test and verify the effectiveness of the gripper and the proposed algorithm.
AB - In terms of the closure theory, for 3D objects, it usually requires at least 7 grasp points to ensure a form closure grasp, which is too strict for real applications. Instead, using a 4-point planar grasp is much more practical. In this paper, a robust form-closure grasping planning algorithm is proposed for a 4-pin gripper to obtain stable grasp points and improve the generalization to grasp objects that have not been seen before. Besides, a lightweight, 3-DoF (Degree of Freedom) 4-pin gripper based on our algorithm is designed for 3D object grasping. The proposed algorithm consists of two parts. First, based on Attractive Region in Environment (ARIE), the stability of the whole grasping process by obtaining form-closure grasp points is ensured. Second, considering the uncertainty of the environment, a learning grasp quality measurement is proposed to make evaluation of robustness for each group of grasp points. Our simulation and physical experiments are performed to test and verify the effectiveness of the gripper and the proposed algorithm.
KW - 4-pin gripper design
KW - Attractive Region in Environment (ARIE)
KW - Generalized robotic grasping
KW - Learning-based grasping
UR - https://www.scopus.com/pages/publications/85077164577
U2 - 10.1016/j.neucom.2019.12.039
DO - 10.1016/j.neucom.2019.12.039
M3 - 文章
AN - SCOPUS:85077164577
SN - 0925-2312
VL - 384
SP - 268
EP - 281
JO - Neurocomputing
JF - Neurocomputing
ER -