OPTRObot: A Synthetic Training Paradigm for Robotic Grasping of Specular Objects in Cluttered Environments

Result code in IS VaVaI
<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21730%2F25%3A00385527" target="_blank" >RIV/68407700:21730/25:00385527 - isvavai.cz</a>
Result on the web
<a href="https://doi.org/10.1109/CASE58245.2025.11164038" target="_blank" >https://doi.org/10.1109/CASE58245.2025.11164038</a>
DOI - Digital Object Identifier
<a href="http://dx.doi.org/10.1109/CASE58245.2025.11164038" target="_blank" >10.1109/CASE58245.2025.11164038</a>

Result language
angličtina
Original language name
OPTRObot: A Synthetic Training Paradigm for Robotic Grasping of Specular Objects in Cluttered Environments
Original language description
We present a novel and complete vision-based pipeline for 6DoF object pose estimation in challenging industrial bin-picking scenarios, characterized by significant clutter, occlusions, and reflective surfaces. Our approach addresses the limitations of both computationally expensive fine-tuning methods and the current immaturity of foundation models in handling such complex environments. The key contribution lies in a balanced approach leveraging synthetic data augmentation and a streamlined architecture to achieve robust performance without extensive per-object optimization. The pipeline integrates existing state-of-the-art object detection, coarse pose estimation, and a render-and-compare refinement strategy, enabling accurate pose estimation from monocular images. We introduce a new benchmark on the recently released dataset, establishing a baseline for future research. Unlike existing industrial approaches, our system minimizes reliance on multi-sensor configurations, offering a cost-effective and easily deployable solution. We demonstrate the impact of error propagation across a complete pipeline using datasets that mirror real-world industrial conditions, in contrast to commonly used but less representative datasets.
Czech name
—
Czech description
—

Project
<a href="/en/project/EF16_026%2F0008432" target="_blank" >EF16_026/0008432: Cluster 4.0 - Methodology of System Integration</a><br>
Continuities
P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Publication year
2025
Confidentiality
S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Article name in the collection
2025 IEEE 21st International Conference on Automation Science and Engineering (CASE)
ISBN
979-8-3315-2246-9
ISSN
2161-8070
e-ISSN
2161-8089
Number of pages
7
Pages from-to
146-152
Publisher name
IEEE Industrial Electronic Society
Place of publication
Vienna
Event location
Los Angeles
Event date
Aug 17, 2025
Type of event by nationality
WRD - Celosvětová akce
UT code for WoS article
001701272600013

Similar results(10)