An official website of the United States government
Previous approaches on 3D shape segmentation mostly rely on heuristic processing and hand-tuned geometric descriptors. In this paper, we propose a novel 3D shape representation learning approach, Directionally Convolutional Network (DCN), to solve the shape segmentation problem. DCN extends convolution operations from images to the surface mesh of 3D shapes. With DCN, we learn effective shape representations from raw geometric features, i.e., face normals and distances, to achieve robust segmentation. More specifically, a two-stream segmentation framework is proposed: one stream is made up by the proposed DCN with the face normals as the input, and the other stream is implemented by a neural network with the face distance histogram as the input. The learned shape representations from the two streams are fused by an element-wise product. Finally, Conditional Random Field (CRF) is applied to optimize the segmentation. Through extensive experiments conducted on benchmark datasets, we demonstrate that our approach outperforms the current state-of-the-arts (both classic and deep learning-based) on a large variety of 3D shapes.
more »
« less
Extended Fabrication-Aware Convolution Learning Framework for Predicting 3D Shape Deformation in Additive Manufacturing
Geometric accuracy control is critical for precision additive manufacturing (AM). To learn and predict the shape deformation from a limited number of training products, a fabrication-aware convolution learning framework has been developed in our previous work to describe the layer-bylayer fabrication process. This work extends the convolution learning framework to broader categories of 3D geometries by constructively incorporating spherical and polyhedral shapes into a unified model. It is achieved by extending 2D cookiecutter modeling approach to 3D case and by modeling spatial correlations. Methodologies demonstrated with real case studies show the promise of prescriptive modeling and control of complicated shape quality in AM.
more »
« less
- Award ID(s):
- 1901514
- PAR ID:
- 10287537
- Date Published:
- Journal Name:
- IEEE International Conference on Automation Science and Engineering CASE
- ISSN:
- 2161-8089
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Digital Fringe Projection for Interlayer Print Defect Characterization in Directed Energy DepositionAbstract Directed Energy Deposition (DED) is one of the main additive manufacturing (AM) families, enabling the fabrication of multi-material parts with high material addition rates. However, the incremental nature of DED fabrication makes it prone to local defect formation due to process condition fluctuations. Known for its rapid and precise 3D surface measurement capabilities, digital fringe projection (DFP) was previously demonstrated in process monitoring for powder bed AM. This study brings DFP to the DED process through development of a custom motor stage system and validates its effectiveness in assessing surface topography and build height measurement. Measurements were taken on both correctly deposited builds and builds with off-nominal deposition conditions, where the system was able to detect pitting as small as 0.425 mm in the lateral size and 0.154 mm in depth in the case of reduced laser energy. This work paves the way for future machine learning-enabled interlayer defect identification, classification, and healing via altering subsequent processing settings.more » « less
-
Delta 3D printers can significantly increase throughput in additive manufacturing by enabling faster and more precise motion compared to conventional serial-axis 3D printers. Further improvements in motion speed and part quality can be realized through model-based feedforward vibration control, as demonstrated on serial-axis 3D printers. However, delta machines have not benefited from model-based controllers because of the difficulty in accurately modeling their position-dependent, coupled nonlinear dynamics. In this paper, we propose an efficient framework to obtain accurate linear parameter-varying models of delta 3D printers at any position within their workspace from a few frequency response measurements. We decompose the dynamics into two sub-models–(1) an experimentally-identified sub-model containing decoupled vibration dynamics; and (2) an analytically-derived sub-model containing coupled dynamics–which are combined into one using receptance coupling. We generalize the framework by extending the analytical model of (2) to account for differing mass profiles and dynamic models of the printer’s end-effector. Experiments demonstrate reasonably accurate predictions of the position-dependent dynamics of a commercial delta printer, augmented with a direct drive extruder, at various positions in its workspace. Note to Practitioners—This work aims to equip high-speed 3D printers, like delta machines, with model-based controllers to complement their speed with high-accuracy. Due to the coupled kinematic chains of the delta, complex control methodologies, some of which require real-time state measurements, are often used to achieve satisfactory control performance. Our modeling approach provides an efficient methodology for obtaining accurate linear models without the need for real-time measurements, thus enabling practitioners to design linear model-based feedforward controllers to achieve the high throughput and accuracy desired in additive manufacturing (AM). The models we develop in this paper are intended for use with feedforward vibration compensation methods, which can be beneficial for both industrial-scale AM machines that have high-powered servo motors and feedback controllers, as well as consumer-grade AM machines which use stepper motors in feedforward control.more » « less
-
K. Ellis, W. Ferrell (Ed.)Fused deposition modeling (FDM) is one of the widely used additive manufacturing (AM) processes but shares major shortcomings typical due to its layer-by-layer fabrication. These challenges (poor surface finishes, presence of pores, inconsistent mechanical properties, etc.) have been attributed to FDM input process parameters, machine parameters, and material properties. Deep learning, a type of machine learning algorithm has proven to help reveal complex and nonlinear input-output relationships without the need for the underlying physics. This research explores the power of multilayer perceptron deep learning algorithm to create a prediction model for critical input process parameters (layer thickness, extrusion temperature, build temperature, build orientation, and print speed) to predict three functional output parameters (dimension accuracy, porosity, and tensile strength) of FDM printed part. A fractional factorial design of experiment was performed and replicated three times per run (n=3). The number of neurons for the hidden layers, learning rate, and epoch were varied. The computational run time, loss function, and root mean square error (RMSE) were used to select the best prediction model for each FDM output parameter. The findings of this work are being extended to online monitoring and real-time control of the AM process enabling an AM digital twin.more » « less
-
K. Ellis, W. Ferrell (Ed.)Fused deposition modeling (FDM) is one of the widely used additive manufacturing (AM) processes but shares major shortcomings typical due to its layer-by-layer fabrication. These challenges (poor surface finishes, presence of pores, inconsistent mechanical properties, etc.) have been attributed to FDM input process parameters, machine parameters, and material properties. Deep learning, a type of machine learning algorithm has proven to help reveal complex and nonlinear input-output relationships without the need for the underlying physics. This research explores the power of multilayer perceptron deep learning algorithm to create a prediction model for critical input process parameters (layer thickness, extrusion temperature, build temperature, build orientation, and print speed) to predict three functional output parameters (dimension accuracy, porosity, and tensile strength) of FDM printed part. A fractional factorial design of experiment was performed and replicated three times per run (n=3). The number of neurons for the hidden layers, learning rate, and epoch were varied. The computational run time, loss function, and root mean square error (RMSE) were used to select the best prediction model for each FDM output parameter. The findings of this work are being extended to online monitoring and real-time control of the AM process enabling an AM digital twin.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- The DOI is not currently available.
