Synthetic Training Data Generation for Visual Object Identification on Load Carriers

Schoepflin, Daniel; Holst, Dirk; Gomse, Martin; Schüppstuhl, Thorsten

doi:10.1016/j.procir.2021.11.211

Cited by 16 publications

(14 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Due to the randomly chosen simulation parameters and in order to create the possibility to compare grasps with each other and to enable statistical analyses, like distributions of succesfull grasp parameters, the contact points of a these grasp including their simulation parameters are saved. 4) Transfer to Synthetic Data Generation Toolbox [5]: A structured set of rules is used to generate packed transport boxes, ensuring that logical consistencies are maintained within the training data domain. These include lighting conditions, camera positions, positioning of objects, and packaging material used.…”

Section: Methodology -Synthetic Grasp Data Generationmentioning

confidence: 99%

See 1 more Smart Citation

Generation of Synthetic AI Training Data for Robotic Grasp-Candidate Identification and Evaluation in Intralogistics Bin-Picking Scenarios

Holst

Schoepflin

Schüppstuhl

2022

Lecture Notes in Mechanical Engineering

Self Cite

View full text Add to dashboard Cite

Robotic bin picking remains a main challenge for the wide enablement of industrial robotic tasks. While AI-enabled picking approaches are encouraging they repeatedly face the problem of data availability. The scope of this paper is to present a method that combines analytical grasp research with the field of synthetic data creation to generate individual training data for use-cases in intralogistics transportation scenarios. Special attention is given to systematic grasp finding for new objects and unknown geometries in transportation bins and to match the generated data to a real two-finger parallel gripper. The presented approach includes a grasping simulation in Pybullet to investigate the general tangibility of objects under uncertainty and combines these findings with a previously reported virtual scene generator in Blender, which generates AI-images of fully packed transport boxes, including depth maps and necessary annotations. This paper, therefore, contributes a synthesizing and cross-topic approach that combines different facets of bin-picking research such as geometric analysis, determination of tangibility of objects, grasping under uncertainty, finding grasps in dynamic and restricted bin-environments, and automation of synthetic data generation. The approach is utilized to generate synthetic grasp training data and to train a grasp-generating convolutional neural network (GG-CNN) and demonstrated on real-world objects.

show abstract

Section: Methodology -Synthetic Grasp Data Generationmentioning

confidence: 99%

“…Following previous work [5], an existing synthetic data generator is used for the creation of training images, to identify objects on load carriers in intralogistics settings. This tool will be extended with functionalities to annotate grasps for a two-finger parallel gripper.…”

Section: Introductionmentioning

confidence: 99%

Generation of Synthetic AI Training Data for Robotic Grasp-Candidate Identification and Evaluation in Intralogistics Bin-Picking Scenarios

Holst

Schoepflin

Schüppstuhl

2022

Lecture Notes in Mechanical Engineering

Self Cite

View full text Add to dashboard Cite

show abstract

“…Focusing on the ability to detect aircraft components in production supplying logistic operations with delivery units, [2,3] provide the capability to enable AI-based visual sensor applications with the help of synthetic training data. Such an approach can be incorporated into the design flow of this paper, however, is limited to components that can be differentiated through means of object detection and a top-view sensor configuration.…”

Section: Related Work and State Of The Artmentioning

confidence: 99%

“…Due to the necessary waiver of markings, visual and markerless identification have to be employed. Such approaches are feasible for distinctive and featurerich aviation components [3]. Within this work, we focus on components that wear little to no features, and have high similarities.…”

Section: Introductionmentioning

confidence: 99%

Assessing Visual Identification Challenges for Unmarked and Similar Aircraft Components

Schoepflin

Gierecker

Schüppstuhl

2022

Lecture Notes in Mechanical Engineering

Self Cite

View full text Add to dashboard Cite

Highest demands for complete traceability and quality control of each component, require thorough identification of each produced, replaced, and (dis-)assembled aircraft component. As many production and MRO-processes for modern aircraft remain to be carried out manually, this poses a great challenge. Many small components either do not feature a Part Number or in MRO-processes their Part Number is occluded or not readable due to dirt and wear. Considering unmarked components with a high resemblance to one another and few characteristics, e.g. standard parts such as bushings and pipes, manual identification is an error-prone task. Avoiding errors through digitalized procedures has the potential to significantly reduce error rates and costs for a typical manual dual control. However, automated identification of components has to overcome the high classification complexity that originates in the manifold of aircraft components and is additionally increased by individualistic MRO modifications for specific aircraft. This work presents a methodological approach to reveal possible challenges for identification procedures and gives special focus to the assessment of similarities between components. Two similarity metrics are introduced that are calculated either through feature-based analysis or through 3D-shape similarity assessment. The methodology is demonstrated with two to this date unsolved Use-Cases that represent different challenges of visual identification systems for similar and unmarked components.

show abstract

“…Synthetic training data for deep learning can be particularly suitable in the industrial context for several reasons [ 3 ]. On the one hand, few industrial datasets are publicly available because data is rarely shared due to confidentiality.…”

Section: Introductionmentioning

confidence: 99%

Automatic Image Generation Pipeline for Instance Segmentation of Deformable Linear Objects

Dirr

Gebauer

Yao

et al. 2023

Sensors

View full text Add to dashboard Cite

Robust detection of deformable linear objects (DLOs) is a crucial challenge for the automation of handling and assembly of cables and hoses. The lack of training data is a limiting factor for deep-learning-based detection of DLOs. In this context, we propose an automatic image generation pipeline for instance segmentation of DLOs. In this pipeline, a user can set boundary conditions to generate training data for industrial applications automatically. A comparison of different replication types of DLOs shows that modeling DLOs as rigid bodies with versatile deformations is most effective. Further, reference scenarios for the arrangement of DLOs are defined to generate scenes in a simulation automatically. This allows the pipelines to be quickly transferred to new applications. The validation of models trained with synthetic images and tested on real-world images shows the feasibility of the proposed data generation approach for segmentation of DLOs. Finally, we show that the pipeline yields results comparable to the state of the art but has advantages in reduced manual effort and transferability to new use cases.

show abstract

Synthetic Training Data Generation for Visual Object Identification on Load Carriers

Cited by 16 publications

References 21 publications

Generation of Synthetic AI Training Data for Robotic Grasp-Candidate Identification and Evaluation in Intralogistics Bin-Picking Scenarios

Generation of Synthetic AI Training Data for Robotic Grasp-Candidate Identification and Evaluation in Intralogistics Bin-Picking Scenarios

Assessing Visual Identification Challenges for Unmarked and Similar Aircraft Components

Automatic Image Generation Pipeline for Instance Segmentation of Deformable Linear Objects

Contact Info

Product

Resources

About