An official website of the United States government
Abstract Solving problems with additive manufacturing (AM) often means fabricating geometrically complex designs, layer-by-layer, along one or multiple directions. Designers navigate this 3D spatial complexity to determine the best design and manufacturing solutions to produce functional parts, manufacturable by AM. However, to assess the manufacturability of their solutions, designers need modalities that naturally visualize AM processes and the designs enabled by them. Creating physical parts offers such visualization but becomes expensive and time-consuming over multiple design iterations. While non-immersive simulations can alleviate this cost of physical visualization, adding digital immersion further improves outcomes from the visualization experience. This research, therefore, studies how differences in immersion between computer-aided (CAx) and virtual reality (VR) environments affect: 1. determining the best solution for additively manufacturing a design and 2. the cognitive load experienced from completing the DfAM problem-solving experience. For the study, designers created a 3D manifold model and simulated manufacturing it in either CAx or VR. Analysis of the filtered data from the study shows that slicing and printing their designs in VR yields a significant change in the manufacturability outcomes of their design compared to CAx. No observable differences were found in the cognitive load experienced between the two modalities. This means that the experiences in VR may influence improvements to manufacturability outcomes without changes to the mental exertion experienced by the designers. This presents key implications for how designers are equipped to solve design problems with AM.
more »
« less
VIRL: Volume-Informed Representation Learning towards few-shot manufacturability estimation
Abstract Designing for manufacturing poses significant challenges in part due to the computation bottleneck of Computer-Aided Manufacturing (CAM) simulations. Although deep learning as an alternative offers fast inference, its performance is dependently bounded by the need for abundant training data. Representation learning, particularly through pre-training, offers promise for few-shot learning, aiding in manufacturability tasks where data can be limited. This work introduces VIRL, a Volume-Informed Representation Learning approach to pre-train a 3D geometric encoder. The pretrained model is evaluated across four manufacturability indicators obtained from CAM simulations: subtractive machining (SM) time, additive manufacturing (AM) time, residual von Mises stress, and blade collisions during Laser Power Bed Fusion process. Across all case studies, the model pre-trained by VIRL shows substantial enhancements in generalizability, as measured by R2 regression results, with improved performance on limited data and superior predictive accuracy with larger datasets. Regarding deployment strategy, case-specific phenomenon exists where finetuning VIRL-pretrained models adversely affects AM tasks with limited data but benefits SM time prediction. Moreover, the efficacy of Low-rank adaptation (LoRA), which balances between probing and finetuning, is explored. LoRA shows stable performance akin to probing with limited data, while achieving a higher upper bound than probing as data size increases, without the computational costs of finetuning. Furthermore, static normalization of manufacturing indicators consistently performs well across tasks, while dynamic normalization enhances performance when a reliable task dependent input is available.
more »
« less
- Award ID(s):
- 2113301
- PAR ID:
- 10572381
- Publisher / Repository:
- Springer Science + Business Media
- Date Published:
- Journal Name:
- Journal of Intelligent Manufacturing
- ISSN:
- 0956-5515
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Direct digital manufacturing (DDM) is the creation of a physical part directly from a computer-aided design (CAD) model with minimal process planning and is typically applied to additive manufacturing (AM) processes to fabricate complex geometry. AM is preferred for DDM because of its minimal user input requirements; as a result, users can focus on exploiting other advantages of AM, such as the creation of intricate mechanisms that require no assembly after fabrication. Such assembly free mechanisms can be created using DDM during a single build process. In contrast, subtractive manufacturing (SM) enables the creation of higher strength parts that do not suffer from the material anisotropy inherent in AM. However, process planning for SM is more difficult than it is for AM due to geometric constraints imposed by the machining process; thus, the application of SM to the fabrication of assembly free mechanisms is challenging. This research describes a voxel-based computer-aided manufacturing (CAM) system that enables direct digital subtractive manufacturing (DDSM) of an assembly free mechanism. Process planning for SM involves voxel-by-voxel removal of material in the same way that an AM process consists of layer-by-layer addition of material. The voxelized CAM system minimizes user input by automatically generating toolpaths based on an analysis of accessible material to remove for a certain clearance in the mechanism's assembled state. The DDSM process is validated and compared to AM using case studies of the manufacture of two assembly free ball-in-socket mechanisms.more » « less
-
In-context learning is the ability of a pretrained model to adapt to novel and diverse downstream tasks by conditioning on prompt examples, without optimizing any parameters. While large language models have demonstrated this ability, how in-context learning could be performed over graphs is unexplored. In this paper, we develop Pretraining Over Diverse In-Context Graph Systems (PRODIGY), the first pretraining framework that enables in-context learning over graphs. The key idea of our framework is to formulate in-context learning over graphs with a novel prompt graph representation, which connects prompt examples and queries. We then propose a graph neural network architecture over the prompt graph and a corresponding family of in-context pretraining objectives. With PRODIGY, the pre- trained model can directly perform novel downstream classification tasks on unseen graphs via in-context learning. We provide empirical evidence of the effectiveness of our framework by showcasing its strong in-context learning performance on tasks involving citation networks and knowledge graphs. Our approach outperforms the in-context learning accuracy of contrastive pretraining baselines with hard-coded adaptation by 18% on average across all setups. Moreover, it also outperforms standard finetuning with limited data by 33% on average with in-context learning.more » « less
-
In this paper, we study the role of initialization in Low Rank Adaptation (LoRA) as originally introduced in Hu et al. [19]. Essentially, to start from the pretrained model as initialization for finetuning, one can either initialize B to zero and A to random (default initialization in PEFT package), or vice-versa. In both cases, the product BA is equal to zero at initialization, which makes finetuning starts from the pretrained model. These two initialization schemes are seemingly sim- ilar. They should in-principle yield the same performance and share the same optimal learning rate. We demonstrate that this is an incorrect intuition and that the first scheme (initializing B to zero and A to random) on average yields better performance compared to the other scheme. Our theoretical analysis shows that the reason behind this might be that the first initialization allows the use of larger learning rates (without causing output instability) compared to the second initial- ization, resulting in more efficient learning of the first scheme. We validate our results with extensive experiments on LLMs.more » « less
-
Abstract Machine learning can be used to automate common or time-consuming engineering tasks for which sufficient data already exist. For instance, design repositories can be used to train deep learning algorithms to assess component manufacturability; however, methods to determine the suitability of a design repository for use with machine learning do not exist. We provide an initial investigation toward identifying such a method using “artificial” design repositories to experimentally test the extent to which altering properties of the dataset impacts the assessment precision and generalizability of neural networks trained on the data. For this experiment, we use a 3D convolutional neural network to estimate quantitative manufacturing metrics directly from voxel-based component geometries. Additive manufacturing (AM) is used as a case study because of the recent growth of AM-focused design repositories such as GrabCAD and Thingiverse that are readily accessible online. In this study, we focus only on material extrusion, the dominant consumer AM process, and investigate three AM build metrics: (1) part mass, (2) support material mass, and (3) build time. Additionally, we compare the convolutional neural network accuracy to that of a baseline multiple linear regression model. Our results suggest that training on design repositories with less standardized orientation and position resulted in more accurate trained neural networks and that orientation-dependent metrics were harder to estimate than orientation-independent metrics. Furthermore, the convolutional neural network was more accurate than the baseline linear regression model for all build metrics.more » « less
