A Concept for Autonomous Quality Control for Core Inspection in Remanufacturing

Kaiser, Jan-Philipp; Lang, Simon; Wurster, Marco; Lanza, Gisela

doi:10.1016/j.procir.2022.02.062

Cited by 11 publications

(2 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One crucial step in remanufacturing involves inspecting end-of-life (EOL) products, which is mostly carried out manually by a skilled worker [6]. However, with promising achievements in computer vision applications, scientists are now focusing on automating this remanufacturing step by adopting intelligent defect detection approaches to recognize faulty regions on EOL parts that require restoration [7][8][9].…”

Section: Introductionmentioning

confidence: 99%

A Defect Detection Method Based on YOLOv7 for Automated Remanufacturing

Satsangee,

Al-Musaibeli,

Ahmad

2024

Applied Sciences

View full text Add to dashboard Cite

Remanufacturing of mechanical parts has recently gained much attention due to the rapid development of green technologies and sustainability. Recent efforts to automate the inspection step in the remanufacturing process using artificial intelligence are noticeable. In this step, a visual inspection of the end-of-life (EOL) parts is carried out to detect defective regions for restoration. This operation relates to the object detection process, a typical computer vision task. Many researchers have adopted well-known deep-learning models for the detection of damage. A common technique in the object detection field is transfer learning, where general object detectors are adopted for specific tasks such as metal surface defect detection. One open-sourced model, YOLOv7, is known for real-time object detection, high accuracy, and optimal scaling. In this work, an investigation into the YOLOv7 behavior on various public metal surface defect datasets, including NEU-DET, NRSD, and KolektorSDD2, is conducted. A case study validation is also included to demonstrate the model’s application in an industrial setting. The tiny variant of the YOLOv7 model showed the best performance on the NEU-DET dataset with a 73.9% mAP (mean average precision) and 103 FPS (frames per second) in inference. For the NRSD dataset, the model’s base variant resulted in 88.5% for object detection and semantic segmentation inferences. In addition, the model achieved 65% accuracy when testing on the KolektorSDD2 dataset. Further, the results are studied and compared with some of the existing defect detection models. Moreover, the segmentation performance of the model was also reported.

show abstract

Section: Introductionmentioning

confidence: 99%

A Defect Detection Method Based on YOLOv7 for Automated Remanufacturing

Satsangee,

Al-Musaibeli,

Ahmad

2024

Applied Sciences

View full text Add to dashboard Cite

show abstract

“…In the pursuit of advancing circular economy, practitioners and researchers have employed the use of machine learning (ML) and artificial intelligence (AI) to enhance the reutilization of products [4,6]. The development, training, and deployment of effective ML/AI models for product identification and defect detection can have a significant financial and environmental impact on industries that deal with manufacturing, repair, and maintenance.…”

Section: Introductionmentioning

confidence: 99%

Inspection in high-mix and high-throughput handling with skeptical and incremental learning

Schimanek¹,

Bilge²,

Dietrich³

2023

Preprint

View full text Add to dashboard Cite

This preprint is based on an accepted paper for the Annals of Scientific Society for Assembly, Handling, and Industrial Robotics 2023, and may undergo further revisions before the final publication. This preprint presents a groundbreaking research study that focuses on the application of incremental learning processes to enhance high-throughput product inspection and minimize false labeling in industrial environments. The paper tackles a critical challenge faced by industries today, which is the accurate identification and labeling of products during the inspection process. Traditional methods often fall short in maintaining high accuracy rates due to evolving production lines and ever-changing product variations. However, this preprint introduces an innovative approach that leverages incremental learning techniques to adapt and improve the inspection process continuously. By employing incremental learning, the proposed system can autonomously learn from new data and make adjustments to its inspection algorithms in real-time. This adaptive capability allows the system to keep up with changing product characteristics, ensuring accurate and reliable labeling throughout high-throughput operations. The result is a significant reduction in false labeling, leading to enhanced quality control and improved efficiency in industrial environments. We invite researchers, industry professionals, and enthusiasts in the field of assembly, handling, and industrial robotics to explore this preprint. The findings presented here have the potential to revolutionize the way product inspection is conducted in industrial settings, addressing a critical aspect of quality control. We encourage readers to engage in discussions, share feedback, and contribute to the advancement of this exciting research area. Abstract: In the circular economy, remanufacturing success relies heavily on the accurate identification and classification of used products. Processes, which rely on worker experience, lack objective validation, leading to the potential for mislabeling and inaccurate damage assessment. This, in turn, results in additional manual evaluations and unproductive costs, which run counter to the principles of sustainability. To address these issues, machine learning and artificial intelligence have been applied with promising results. However, producing reliable large amounts of labeled data remains a challenge, as workers are susceptible to human error. This paper addresses process design in production. It proposes a new design to ensure that only valid labels enter the prediction models, reducing the potential for false labels in the dataset. Through this, the aim is to improve the accuracy and reliability of remanufacturing, ultimately reducing costs and mitigating the carbon footprint in the manufacturing, repair, and maintenance industries.

show abstract