On Evaluation of 6D Object Pose Estimation

Hodaň, Tomáš; Matas, Jiřı́; Obdržálek, Štěpán

doi:10.1007/978-3-319-49409-8_52

Cited by 130 publications

(107 citation statements)

References 17 publications

Supporting

Mentioning

102

Contrasting

Order By: Relevance

“…In contrast to these works, our focus is not the localization of an object of interest in an image or with respect to the camera, but the relative positioning of one object with respect to another, the robot and the block. Rather than targeting full 6DOF pose (Hodaň et al, 2016), we restrict ourselves to two position and one angle parameters, since one can reasonably assume that both the robot and the block are standing on a more or less flat surface. We developed a direct prediction approach, building on the success of the CNN based-approaches and following the logic of end-to-end learning, to learn to directly predict the relative pose.…”

Section: Object Localization and Pose Estimationmentioning

confidence: 99%

Virtual Training for a Real Application: Accurate Object-Robot Relative Localization Without Calibration

2018

View full text Add to dashboard Cite

Localizing an object accurately with respect to a robot is a key step for autonomous robotic manipulation. In this work, we propose to tackle this task knowing only 3D models of the robot and object in the particular case where the scene is viewed from uncalibrated cameras -a situation which would be typical in an uncontrolled environment, e.g., on a construction site. We demonstrate that this localization can be performed very accurately, with millimetric errors, without using a single real image for training, a strong advantage since acquiring representative training data is a long and expensive process. Our approach relies on a classification Convolutional Neural Network (CNN) trained using hundreds of thousands of synthetically rendered scenes with randomized parameters. To evaluate our approach quantitatively and make it comparable to alternative approaches, we build a new rich dataset of real robot images with accurately localized blocks.

show abstract

Section: Object Localization and Pose Estimationmentioning

confidence: 99%

Virtual Training for a Real Application: Accurate Object-Robot Relative Localization Without Calibration

2018

View full text Add to dashboard Cite

show abstract

“…Figure 11 (b) shows the recall values when δ = 15, τ = 100, and t = 0.5. Note that these are also the parameters used in the experiments of the original publication of VSD [3]. Slightly decreasing the parameter δ gives rise lower recall values, however, we cannot see any significant difference from the point of characteristics of the bar charts.…”

Section: B Analyses Based On Visible Surface Discrepancymentioning

confidence: 79%

“…Sym Pose Distance. VSD [3] is presented as an ambiguityinvariant pose error function. The acceptance of a pose hypothesis as correct depends on the plausibility of the estimated pose given the available data.…”

Section: B Metrics: Correctness Of An Estimationmentioning

confidence: 99%

A review on object pose recovery: From 3D bounding box detectors to full 6D pose estimators

Şahin

Garcia-Hernando

Sock

et al. 2020

Image and Vision Computing

View full text Add to dashboard Cite

Object pose recovery has gained increasing attention in the computer vision field as it has become an important problem in rapidly evolving technological areas related to autonomous driving, robotics, and augmented reality. Existing review-related studies have addressed the problem at visual level in 2D, going through the methods which produce 2D bounding boxes of objects of interest in RGB images. The 2D search space is enlarged either using the geometry information available in the 3D space along with RGB (Mono/Stereo) images, or utilizing depth data from LIDAR sensors and/or RGB-D cameras. 3D bounding box detectors, producing category-level amodal 3D bounding boxes, are evaluated on gravity aligned images, while full 6D object pose estimators are mostly tested at instance-level on the images where the alignment constraint is removed. Recently, 6D object pose estimation is tackled at the level of categories. In this paper, we present the first comprehensive and most recent review of the methods on object pose recovery, from 3D bounding box detectors to full 6D pose estimators. The methods mathematically model the problem as a classification, regression, classification & regression, template matching, and point-pair feature matching task. Based on this, a mathematical-model-based categorization of the methods is established. Datasets used for evaluating the methods are investigated with respect to the challenges, and evaluation metrics are studied. Quantitative results of experiments in the literature are analysed to show which category of methods best performs across what types of challenges. The analyses are further extended comparing two methods, which are our own implementations, so that the outcomes from the public results are further solidified. Current position of the field is summarized regarding object pose recovery, and possible research directions are identified.

show abstract

“…In robotics, fiducial markers are commonly used for detecting the objects and predicting their pose relative to the camera [17] but their use limits the type of environments the robot can operate in. This constraint can be removed by using a trainable object detector architecture [18]- [20,27]- [29]. However, these methods often require gathering training data for the target objects at hand, which is often time consuming and requires the knowledge of the object (e.g.…”

Section: Related Workmentioning

confidence: 99%

Monte-Carlo Tree Search for Efficient Visually Guided Rearrangement Planning

Labbé

Zagoruyko

Kalevatykh

et al. 2020

IEEE Robot. Autom. Lett.

View full text Add to dashboard Cite

We address the problem of visually guided rearrangement planning with many movable objects, i.e., finding a sequence of actions to move a set of objects from an initial arrangement to a desired one, while relying on visual inputs coming from an RGB camera. To do so, we introduce a complete pipeline relying on two key contributions. First, we introduce an efficient and scalable rearrangement planning method, based on a Monte-Carlo Tree Search exploration strategy. We demonstrate that because of its good trade-off between exploration and exploitation our method (i) scales well with the number of objects while (ii) finding solutions which require a smaller number of moves compared to the other state-of-the-art approaches. Note that on the contrary to many approaches, we do not require any buffer space to be available. Second, to precisely localize movable objects in the scene, we develop an integrated approach for robust multi-object workspace state estimation from a single uncalibrated RGB camera using a deep neural network trained only with synthetic data. We validate our multi-object visually guided manipulation pipeline with several experiments on a real UR-5 robotic arm by solving various rearrangement planning instances, requiring only 60 ms to compute the plan to rearrange 25 objects. In addition, we show that our system is insensitive to camera movements and can successfully recover from external perturbations. Supplementary video, source code and pre-trained models are available at https://ylabbe.github.io/rearrangement-planning.

show abstract

On Evaluation of 6D Object Pose Estimation

Cited by 130 publications

References 17 publications

Virtual Training for a Real Application: Accurate Object-Robot Relative Localization Without Calibration

Virtual Training for a Real Application: Accurate Object-Robot Relative Localization Without Calibration

A review on object pose recovery: From 3D bounding box detectors to full 6D pose estimators

Monte-Carlo Tree Search for Efficient Visually Guided Rearrangement Planning

Contact Info

Product

Resources

About