期刊
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS
卷 69, 期 11, 页码 11611-11621出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TIE.2021.3120474
关键词
Grippers; Grasping; Adaptation models; Shape; Three-dimensional displays; Task analysis; Labeling; Adaptive grasping attribute model (AGA-model); convolutional neural network; grasp detection; oriented arrow representation (OAR)
资金
- National Key R&D Program of China [2018YFB1305300]
- National Natural Science Foundation of China [62176138, 62176136]
- Shandong Provincial Key Research and Development Program (Major Scientific and Technological Innovation Project) [2019JZZY010130, 2020CXGC010207]
This article presents a pixel-level grasp detection method based on deep neural network, which achieves high accuracy and fast speed on RGB images through novel representation models, adaptive attribute models, and feature fusion and grasp-aware network.
Machine vision-based planar grasping detection is challenging due to uncertainty about object shape, pose, size, etc. Previous methods mostly focus on predicting discrete gripper configurations, and may miss some ground-truth grasp postures. In this article, a pixel-level grasp detection method is proposed, which uses deep neural network to predict pixel-level gripper configurations on RGB images. First, a novel oriented arrow representation model (OAR-model) is introduced to represent the gripper configuration of parallel-jaw and three-fingered gripper, which can partly improve the applicability to different grippers. Then, the adaptive grasping attribute model is proposed to adaptively represent the grasping attribute of objects, for resolving angle conflicts in training and simplifying pixel-level labeling. Lastly, the adaptive feature fusion and grasp-aware network (AFFGA-Net) is proposed to predict pixel-level OAR-models on RGB images. AFFGA-Net improves the robustness in unstructured scenarios by using hybrid atrous spatial pyramid and adaptive decoder connected in sequence. On the public Cornell dataset and actual objects, our structure achieves 99.09% and 98.0% grasp detection accuracy, respectively. In over 2400 robotic grasp trials, our structure achieves an average success rate of 98.77% in single-object scenarios and 93.69% in cluttered scenarios. Moreover, AFFGA-Net completes a grasp detection pipeline within 15 ms.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据