Pose estimation and bin picking for deformable products

Joffe, Benjamin; Walker, Tevon; Gourdon, Rémy; Ahlin, Konrad

doi:10.1016/j.ifacol.2019.12.566

Cited by 13 publications

(9 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Machine learning-based approaches have been investigated to recognize food products in the RBP scenario. Joffe et al (2019) proposed a method using a standard Faster R-CNN architecture with a Resnet 101 feature extractor to recognize chicken and evaluated two pose estimation approaches: the augmented autoencoder and direct regression approach. A suction cup gripper was used to pick the chickens.…”

Section: Recent Advancesmentioning

confidence: 99%

Challenges and Opportunities in Robotic Food Handling: A Review

2022

View full text Add to dashboard Cite

Despite developments in robotics and automation technologies, several challenges need to be addressed to fulfill the high demand for automating various manufacturing processes in the food industry. In our opinion, these challenges can be classified as: the development of robotic end-effectors to cope with large variations of food products with high practicality and low cost, recognition of food products and materials in 3D scenario, better understanding of fundamental information of food products including food categorization and physical properties from the viewpoint of robotic handling. In this review, we first introduce the challenges in robotic food handling and then highlight the advances in robotic end-effectors, food recognition, and fundamental information of food products related to robotic food handling. Finally, future research directions and opportunities are discussed based on an analysis of the challenges and state-of-the-art developments.

show abstract

Section: Recent Advancesmentioning

confidence: 99%

Challenges and Opportunities in Robotic Food Handling: A Review

2022

View full text Add to dashboard Cite

show abstract

“…Generally, a state-of-the-art object detector is first used to recognize individual objects, and the resultant cropped images are passed to the pose estimator. Following [6,7], we use Mask R-CNN [8] for object detection. As for the task of pose estimation, we consider an augmented autoencoder [5], since it has demonstrated good performance in bin picking of deformable products [7].…”

Section: Introductionmentioning

confidence: 99%

“…Following [6,7], we use Mask R-CNN [8] for object detection. As for the task of pose estimation, we consider an augmented autoencoder [5], since it has demonstrated good performance in bin picking of deformable products [7]. Sample results of our proposed method are displayed in Fig.…”

Section: Introductionmentioning

confidence: 99%

Object Detection And Autoencoder-Based 6d Pose Estimation For Highly Cluttered Bin Picking

Höfer¹,

Shamsafar

Benbarka

et al. 2021

2021 IEEE International Conference on Image Processing (ICIP)

View full text Add to dashboard Cite

Bin picking is a core problem in industrial environments and robotics, with its main module as 6D pose estimation. However, industrial depth sensors have a lack of accuracy when it comes to small objects. Therefore, we propose a framework for pose estimation in highly cluttered scenes with small objects, which mainly relies on RGB data and makes use of depth information only for pose refinement. In this work, we compare synthetic data generation approaches for object detection and pose estimation and introduce a pose filtering algorithm that determines the most accurate estimated poses. We will make our real dataset for object detection available with the paper.

show abstract

“…Generally, a state-of-the-art object detector is first used to recognize individual objects, and the resultant cropped images are passed to the pose estimator. Following [6,7], we use Mask R-CNN [8] for object detection. As for the task of pose estimation, we consider an augmented autoencoder [4], since it has demonstrated good performance in bin picking of deformable products [7].…”

Section: Introductionmentioning

confidence: 99%

“…Following [6,7], we use Mask R-CNN [8] for object detection. As for the task of pose estimation, we consider an augmented autoencoder [4], since it has demonstrated good performance in bin picking of deformable products [7]. Sample results of our proposed method are displayed in Fig.…”

Section: Introductionmentioning

confidence: 99%