Lightweight robotic grasping detection network based on dual attention and inverted residual

doi:10.1177/01423312241247346

	Lightweight robotic grasping detection network based on dual attention and inverted residual
	Yang, Yuequan 1; Li, Wei 1; Cao, Zhiqiang 2,3; Bao, Jiatong 4; Li, Fudong 1
刊名	TRANSACTIONS OF THE INSTITUTE OF MEASUREMENT AND CONTROL
	2024-04-22
页码	9
关键词	Robotic grasping lightweight network attention strategy inverted residual pixel-level
ISSN号	0142-3312
DOI	10.1177/01423312241247346
通讯作者	Yang, Yuequan(yang@yzu.edu.cn)
英文摘要	Grasping detection is one of the crucial capabilities for robot systems. Deep learning has achieved remarkable outcomes in robot grasping tasks; however, many deep neural networks were at the expense of high computation cost with memory requirements, which hindered their deployment on computing-constrained devices. To solve this problem, this paper proposes an end-to-end lightweight network with dual attention and inverted residual strategies (LiDAIR), which adopts a generative pixel-level prediction to achieve grasp detection. The LiDAIR is composed of the convolution modules (Conv), the inverted residual convolution module (IRCM), the convolutional block attention connection module (CBACM), and the transposed convolution modules (TConv). The Convs are utilized in downsampling processes to extract the input image features. Then, the IRCM is proposed as a bridge between the downsampling and upsampling phases. In the upsampling phase, the CBACM is designed to focus on the valuable regions from spatial and channel dimensions, where the skip connection is employed to attain multi-level feature fusion. Afterwards, the TConvs are used to restore image resolution. The LiDAIR is lightweight with 704K parameters and enjoys a good tradeoff among lightweight structure, accuracy, and speed. It was evaluated on both the Cornell data set and the Jacquard data set within 10 ms inference time, and the detection accuracy on both the data sets were 97.7% and 92.7%, respectively.
资助项目	National Natural Science Foundation of China[61973302] ; National Natural Science Foundation of China[62073322] ; National Natural Science Foundation of China[61836015]
WOS研究方向	Automation & Control Systems ; Instruments & Instrumentation
语种	英语
出版者	SAGE PUBLICATIONS LTD
WOS记录号	WOS:001206501700001
资助机构	National Natural Science Foundation of China
内容类型	期刊论文
源URL	[http://ir.ia.ac.cn/handle/173211/58265]
专题	智能机器人系统研究
通讯作者	Yang, Yuequan
作者单位	1.Yangzhou Univ, Coll Informat Engn, Coll Artificial Intelligence, Dept Automat, 196 Huayang West Rd, Yangzhou, 225127, Peoples R China 2.Chinese Acad Sci, Inst Automat, State Key Lab Multimodal Artificial Intelligence S, Beijing, Peoples R China 3.Univ Chinese Acad Sci, Sch Artificial Intelligence, Beijing, Peoples R China 4.Yangzhou Univ, Coll Elect Energy & Power Engn, Yangzhou, Peoples R China
推荐引用方式 GB/T 7714	Yang, Yuequan,Li, Wei,Cao, Zhiqiang,et al. Lightweight robotic grasping detection network based on dual attention and inverted residual[J]. TRANSACTIONS OF THE INSTITUTE OF MEASUREMENT AND CONTROL,2024:9.
APA	Yang, Yuequan,Li, Wei,Cao, Zhiqiang,Bao, Jiatong,&Li, Fudong.(2024).Lightweight robotic grasping detection network based on dual attention and inverted residual.TRANSACTIONS OF THE INSTITUTE OF MEASUREMENT AND CONTROL,9.
MLA	Yang, Yuequan,et al."Lightweight robotic grasping detection network based on dual attention and inverted residual".TRANSACTIONS OF THE INSTITUTE OF MEASUREMENT AND CONTROL (2024):9.