Point Based Deep Learning to Automate Automotive Assembly Simulation Model Generation with Respect to the Digital Factory

Petschnigg, Christina; Bartscher, Stefan; Pilz, Jürgen

doi:10.1109/icitm48982.2020.9080347

Cited by 7 publications

(8 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A first methodology for systematic data collection and processing in a large-scale industrial environment was presented in [2]. On the process side, all of the non-highlighted steps that are illustrated in Figure 1 are described in more detail in future work, including technology specifications and a mathematical concept for the placement of the segmented objects in an environment model.…”

Section: Discussionmentioning

confidence: 99%

“…Finally, to generate an environment model, the poses of the relevant objects need to be estimated and the point sets have to be replaced by computer-aided design (CAD) models. Figure 1 summarizes the process for generating an environment model on the basis of a point cloud, which was introduced in our previous work [2]. Each of the process steps can be subdivided into further building blocks.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Uncertainty Estimation in Deep Neural Networks for Point Cloud Segmentation in Factory Planning

Petschnigg

Pilz

2021

Modelling

Self Cite

View full text Add to dashboard Cite

The digital factory provides undoubtedly great potential for future production systems in terms of efficiency and effectivity. A key aspect on the way to realize the digital copy of a real factory is the understanding of complex indoor environments on the basis of three-dimensional (3D) data. In order to generate an accurate factory model including the major components, i.e., building parts, product assets, and process details, the 3D data that are collected during digitalization can be processed with advanced methods of deep learning. For instance, the semantic segmentation of a point cloud enables the identification of relevant objects within the environment. In this work, we propose a fully Bayesian and an approximate Bayesian neural network for point cloud segmentation. Both of the networks are used within a workflow in order to generate an environment model on the basis of raw point clouds. The Bayesian and approximate Bayesian networks allow us to analyse how different ways of estimating uncertainty in these networks improve segmentation results on raw point clouds. We achieve superior model performance for both, the Bayesian and the approximate Bayesian model compared to the frequentist one. This performance difference becomes even more striking when incorporating the networks’ uncertainty in their predictions. For evaluation, we use the scientific data set S3DIS as well as a data set, which was collected by the authors at a German automotive production plant. The methods proposed in this work lead to more accurate segmentation results and the incorporation of uncertainty information also makes this approach especially applicable to safety critical applications aside from our factory planning use case.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Uncertainty Estimation in Deep Neural Networks for Point Cloud Segmentation in Factory Planning

Petschnigg

Pilz

2021

Modelling

Self Cite

View full text Add to dashboard Cite

show abstract

“…Benefit from the development of highprecision laser 3D surface scanning technology, online sensing devices are capable of generating large-scale point-based surface model of the input mechanical part [3]- [5]. Resolution of current modern 3D laser scanning systems can reach less than 0.0001mm [6], [7], and the scale of surface model is consequently more than 1000k points.…”

Section: Fig1 Application Framework Of Automatic Assembly Line Basedmentioning

confidence: 99%

Fast Assembly Tolerance Inspection Method Using Feature-Based Adaptive Scale Reduction in Automatic Assembly Line

Qiu

Zhang

et al. 2020

IEEE Access

View full text Add to dashboard Cite

Enormous computation and long data processing time of the online tolerance inspection procedure has been a bottleneck in reducing the production cycle time of automatic assembly line. To reduce time consumption while maintaining accuracy, a fast assembly tolerance inspection method is proposed in this paper using scale-reduced models. First, the general part surface, input by a laser-scanned dense point cloud, is segmented through feature-based clustering. Shape variation estimation are designed to determine reduction ratios between clusters adaptively. Then, assembly simulation method is further advanced to utilize coarse point models with imbalanced density. Finally, numerical experiments are designed to estimate both static and dynamic assembly tolerances as results of tolerance inspection. A case study on a motion component of a rail material transporter shows that the proposed method could reduce 81% of the data processing time compared to traditional practice. The precision loss has minor influence on the inspection results because spatial positioning error is no more than 6% of tolerance interval and 18% of scanning resolution. Moreover, the proposed method has a comprehensive advantage of time and accuracy than two conventional point cloud simplification techniques based on iterative and clustering methods.

show abstract

“…Currently, this is still an open research question with hardly any literature focusing on the requirements of industrial applications. We focus on the brownfield planning case and propose a holistic and methodical modelling approach based on [ 9 ] for the automated generation of an environment model in a simulation engine starting from the collection of a raw point cloud of a manufacturing plant. Initially, a data collection methodology based on 3D laser scanning and photogrammetry is proposed, which enables us to collect a data set at a German automotive OEM in order to thoroughly test our remaining workflow steps.…”

Section: Introductionmentioning

confidence: 99%

From a Point Cloud to a Simulation Model—Bayesian Segmentation and Entropy Based Uncertainty Estimation for 3D Modelling

Petschnigg

Spitzner

Weitzendorf

et al. 2021

Entropy

Self Cite

View full text Add to dashboard Cite

The 3D modelling of indoor environments and the generation of process simulations play an important role in factory and assembly planning. In brownfield planning cases, existing data are often outdated and incomplete especially for older plants, which were mostly planned in 2D. Thus, current environment models cannot be generated directly on the basis of existing data and a holistic approach on how to build such a factory model in a highly automated fashion is mostly non-existent. Major steps in generating an environment model of a production plant include data collection, data pre-processing and object identification as well as pose estimation. In this work, we elaborate on a methodical modelling approach, which starts with the digitalization of large-scale indoor environments and ends with the generation of a static environment or simulation model. The object identification step is realized using a Bayesian neural network capable of point cloud segmentation. We elaborate on the impact of the uncertainty information estimated by a Bayesian segmentation framework on the accuracy of the generated environment model. The steps of data collection and point cloud segmentation as well as the resulting model accuracy are evaluated on a real-world data set collected at the assembly line of a large-scale automotive production plant. The Bayesian segmentation network clearly surpasses the performance of the frequentist baseline and allows us to considerably increase the accuracy of the model placement in a simulation scene.

show abstract

Point Based Deep Learning to Automate Automotive Assembly Simulation Model Generation with Respect to the Digital Factory

Cited by 7 publications

References 10 publications

Uncertainty Estimation in Deep Neural Networks for Point Cloud Segmentation in Factory Planning

Uncertainty Estimation in Deep Neural Networks for Point Cloud Segmentation in Factory Planning

Fast Assembly Tolerance Inspection Method Using Feature-Based Adaptive Scale Reduction in Automatic Assembly Line

From a Point Cloud to a Simulation Model—Bayesian Segmentation and Entropy Based Uncertainty Estimation for 3D Modelling

Contact Info

Product

Resources

About