A vision-based system for monitoring block assembly in shipbuilding

Kim, Min Soo; Choi, Woosung; Kim, Byung-Chul; Kim, Hokyeong; Seol, Jae Hun; Woo, Jong-Hun; Ko, Kwang Hee

doi:10.1016/j.cad.2014.09.001

Cited by 21 publications

(13 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Common action recognition models based on deep learning include the LSTM-based LRCN model [ 17 ], the two-stream convolutional model [ 10 ], and the C3D model [ 14 ]. The recognition accuracy of these three models on the public data set UCF-101 [ 19 ] is similar, but the C3D model is fastest, reaching 313 fps [ 14 ], while the two-stream convolutional model is 1.2 fps [ 12 ].…”

Section: Related Workmentioning

confidence: 99%

“…Kim et al proposed a vision-based system for monitoring the block assembly in ship building. Their system can extract areas of blocks; the extracted blocks are then identified and compared with CAD data in an effort to estimate the assembly progress [ 10 ]. Although the abovementioned research realizes the monitoring of the assembly progress, it does not provide part recognition, part positioning and assembly recognition in the overall process.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Monitoring of Assembly Process Using Deep Learning Technology

Chen

Zhang

Wang

et al. 2020

Sensors

View full text Add to dashboard Cite

Monitoring the assembly process is a challenge in the manual assembly of mass customization production, in which the operator needs to change the assembly process according to different products. If an assembly error is not immediately detected during the assembly process of a product, it may lead to errors and loss of time and money in the subsequent assembly process, and will affect product quality. To monitor assembly process, this paper explored two methods: recognizing assembly action and recognizing parts from complicated assembled products. In assembly action recognition, an improved three-dimensional convolutional neural network (3D CNN) model with batch normalization is proposed to detect a missing assembly action. In parts recognition, a fully convolutional network (FCN) is employed to segment, recognize different parts from complicated assembled products to check the assembly sequence for missing or misaligned parts. An assembly actions data set and an assembly segmentation data set are created. The experimental results of assembly action recognition show that the 3D CNN model with batch normalization reduces computational complexity, improves training speed and speeds up the convergence of the model, while maintaining accuracy. Experimental results of FCN show that FCN-2S provides a higher pixel recognition accuracy than other FCNs.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Monitoring of Assembly Process Using Deep Learning Technology

Chen

Zhang

Wang

et al. 2020

Sensors

View full text Add to dashboard Cite

show abstract

“…Large blocks (i.e., erection blocks) are made by joining several blocks together. Finally, large blocks are moved to the dock and welded together to form an entire ship (Kim et al 2015). As noted by Roh and Lee (2007b) in their overview of shipbuilding process, "essentially, the manufacturing process of ship is similar to that of a large product by use of Lego blocks".…”

Section: Introductionmentioning

confidence: 99%

Comparing BIM in Construction with 3D Modeling in Shipbuilding Industries: Is the Grass Greener on the Other Side?

Ran¹,

Singh²

2016

Product Lifecycle Management in the Era of Internet of Things

View full text Add to dashboard Cite

Part 4: Building Information Modeling (BIM)International audienceBuilding Information Modelling (BIM), as an object-oriented tool, has been the buzzword in Architecture Engineering and Construction (AEC) sector in recent years. The buzz has created a lot of promise of an imminent paradigm shift and productivity and lifecycle improvements in the AEC sector, and plenty of benefits are cited in the literature. Consequently, the word has reached the shipbuilding industry as well, which faces many similar productivity and lifecycle challenges as the AEC industry, and which is seeking similar advancements in digital tools that can bring around the change. As a result, the Finnish shipbuilding industry expressed interest in exploring what are the BIM-enabled best practices in the AEC sector, and which of these can potentially be transferred to the shipbuilding industry. This research explored these issues. Findings suggest that due to lack of mutual communication, the professionals across each industry believe the other to be doing better

show abstract

“…Kim, M. et al [ 20 ] proposed a vision-based monitoring system for the ship-block assembly process. Ship block images acquired from the camera are processed to extract the areas of the ship blocks.…”

Section: Introductionmentioning

confidence: 99%

A Real-Time Physical Progress Measurement Method for Schedule Performance Control Using Vision, an AR Marker and Machine Learning in a Ship Block Assembly Process

Choi

Seo

2020

Sensors

View full text Add to dashboard Cite

Progress control is a key technology for successfully carrying out a project by predicting possible problems, particularly production delays, and establishing measures to avoid them (decision-making). However, shipyard progress management is still dependent on the empirical judgment of the manager, and this has led to delays in delivery, which raises ship production costs. Therefore, this paper proposes a methodology for shipyard ship block assembly plants that enables objective process progress measurement based on real-time work performance data, rather than the empirical judgment of a site manager. In particular, an IoT-based physical progress measurement method that can automatically measure work performance without human intervention is presented for the mounting and welding activities of ship block assembly work. Both an augmented reality (AR) marker-based image analysis system and a welding machine time-series data-based machine learning model are presented for measuring the performances of the mounting and welding activities. In addition, the physical progress measurement method proposed in this study was applied to the ship block assembly plant of shipyard H to verify its validity.

show abstract

A vision-based system for monitoring block assembly in shipbuilding

Cited by 21 publications

References 21 publications

Monitoring of Assembly Process Using Deep Learning Technology

Monitoring of Assembly Process Using Deep Learning Technology

Comparing BIM in Construction with 3D Modeling in Shipbuilding Industries: Is the Grass Greener on the Other Side?

A Real-Time Physical Progress Measurement Method for Schedule Performance Control Using Vision, an AR Marker and Machine Learning in a Ship Block Assembly Process

Contact Info

Product

Resources

About