A Dataset for Improved RGBD-Based Object Detection and Pose Estimation for Warehouse Pick-and-Place

Rennie, Colin; Shome, Rahul; Bekris, Kostas E.; Souza, Alberto F. De

doi:10.1109/lra.2016.2532924

Cited by 153 publications

(85 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This can involve problems where the robot has to place objects on top of each other. Future efforts should also aim towards robust rearrangement trajectories under the presence of uncertainty, which can arise from pose estimation processes [35]. It is interesting to evaluate how different object placements may result in different probability of success during real-world execution.…”

Section: Discussionmentioning

confidence: 99%

Efficiently solving general rearrangement tasks: A fast extension primitive for an incremental sampling-based planner

Krontiris

Bekris

2016

2016 IEEE International Conference on Robotics and Automation (ICRA)

Self Cite

View full text Add to dashboard Cite

Manipulating multiple movable obstacles is a hard problem that involves searching high-dimensional C-spaces. A milestone method for this problem was able to compute solutions for monotone instances. These are problems where every object needs to be transferred at most once to achieve a desired arrangement. The method uses backtracking search to find the order with which objects should be moved. This paper first proposes an approximate but significantly faster alternative for monotone rearrangement instances. The method defines a dependency graph between objects given minimum constraint removal paths (MCR) to transfer each object to its target. From this graph, the approach discovers the order of moving objects by performing topological sorting without backtracking search. The approximation arises from the limitation to consider only MCR paths, which minimize, however, the number of conflicts between objects. To solve non-monotone instances, this primitive is incorporated in a higher-level incremental search algorithm for general rearrangement planning, which operates similar to Bi-RRT. Given a start and a goal object arrangement, tree structures of reachable new arrangements are generated by using the primitive as an expansion procedure. The integrated solution achieves probabilistic completeness for the general non-monotone case and based on simulated experiments it achieves very good success ratios, solution times and path quality relative to alternatives.

show abstract

Section: Discussionmentioning

confidence: 99%

Efficiently solving general rearrangement tasks: A fast extension primitive for an incremental sampling-based planner

Krontiris

Bekris

2016

2016 IEEE International Conference on Robotics and Automation (ICRA)

Self Cite

View full text Add to dashboard Cite

show abstract

“…LINEMOD [2], MULT-I [4], OCC [28], BIN-P [35], and T-LESS [42] are the datasets most frequently used to test the performances of full 6D pose estimators. In a recently proposed benchmark for 6D object pose estimation [5], these datasets are refined and are presented in a unified format along with three new datasets (Rutgers Amazon Picking Challenge (RU-APC) [184], TU Dresden Light (TUD-L), and Toyota Light (TYO-L)). The table which details the parameters of these datasets ( Table 1, pp.…”

Section: A Datasetsmentioning

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

“…Calculate ( N , L), the average surface normal and the tangent plane at the center of facet F ; 5 Generate F , the point cloud that is symmetrical to F with respect to plane L;…”

Section: Global Geometric Constraintsmentioning

confidence: 99%

Inferring 3D Shapes of Unknown Rigid Objects in Clutter Through Inverse Physics Reasoning

Song

Boularias

2019

IEEE Robot. Autom. Lett.

View full text Add to dashboard Cite

We present a probabilistic approach for building, on the fly, 3-D models of unknown objects while being manipulated by a robot. We specifically consider manipulation tasks in piles of clutter that contain previously unseen objects. Most manipulation algorithms for performing such tasks require known geometric models of the objects in order to grasp or rearrange them robustly. One of the novel aspects of this work is the utilization of a physics engine for verifying hypothesized geometries in simulation. The evidence provided by physics simulations is used in a probabilistic framework that accounts for the fact that mechanical properties of the objects are uncertain. We present an efficient algorithm for inferring occluded parts of objects based on their observed motions and mutual interactions. Experiments using a robot show that this approach is efficient for constructing physically realistic 3-D models, which can be useful for manipulation planning. Experiments also show that the proposed approach significantly outperforms alternative approaches in terms of shape accuracy.

show abstract

A Dataset for Improved RGBD-Based Object Detection and Pose Estimation for Warehouse Pick-and-Place

Cited by 153 publications

References 31 publications

Efficiently solving general rearrangement tasks: A fast extension primitive for an incremental sampling-based planner

Efficiently solving general rearrangement tasks: A fast extension primitive for an incremental sampling-based planner

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

Inferring 3D Shapes of Unknown Rigid Objects in Clutter Through Inverse Physics Reasoning

Contact Info

Product

Resources

About