.png){kind=icon}
.png){kind=icon}
Return
Cooperative Grasp Detection using Convolutional Neural Network
.png){kind=icon}
PRE
AI
DOI：10.1007/s10846-023-02028-5.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
In this work, we develop a complete robotic system to enable the robot to fulfill object-independent cooperative grasp tasks. The human subject initiates a task by either holding the object in hand or placing the object on the table. The human intention is inferred through body motions, after which the robot selects the corresponding grasp strategy. A novel real-time grasp detection model is proposed to choose the best picking locations according to the object's shape. This module enables the robot to grasp any object placed on the table. Moreover, if handover grasp task is triggered, the hand pixels are detected and filtered out from candidate grasp poses for safety purpose. The proposed grasp detection model is evaluated on two public grasping datasets and a set of casual objects. The best model variant can achieve accuracy of 97.8% and 96.6% on image-wise splitting and object-wise splitting tests on Cornell Grasp Dataset respectively. The Jacquard Dataset accuracy is 93.9%. The overall system is also evaluated on real cooperative grasp tasks. The experimental results show effectiveness of the proposed robot grasp detection and implementation system.
Keywords:
Cooperative grasp
Convolutional neural network
Human intention inference
Journal
J
IF:
2.8
Papers:
3.8K
Citations:
6.9K