Probabilistic Pose Recovery Using Learned Hierarchical Object Models

Detry, Renaud; Pugeault, Nicolas; Piater, Justus

doi:10.1007/978-3-540-92781-5_9

“…Each trial begins with an estimate of the pose of the object relative to the robot [Detry et al, 2008] and sets its grasp location accordingly. The model's ECVD are then projected into the scene, and the robot attempts to perform the grasp and lift the object 15cm so that it is clear of the stand.…”

Section: Grasping Experiments Proceduresmentioning

confidence: 99%

“…Each descriptor is a symbolic representation for an edge in 3D. The resulting features are called early cognitive vision descriptors (ECVD) [Pugeault, 2008], and can be used in generating models of objects for pose estimation [Detry et al, 2008], and for symbolically describing 3D scenes. By using a large amount of small ECVDs, any arbitrary object can be represented.…”

Section: Appendix Dynamical Systems Motor Primitivesmentioning

confidence: 99%

Grasping with Vision Descriptors and Motor Primitives

Kroemer

¹

,

Detry

²

,

Piater

³

et al. 2011

Informatics in Control, Automation and Robotics

Self Cite

View full text Add to dashboard Cite

Abstract:Grasping is one of the most important abilities needed for future service robots. Given the task of picking up an object from betweem clutter, traditional robotics approaches would determine a suitable grasping point and then use a movement planner to reach the goal. The planner would require precise and accurate information about the environment and long computation times, both of which may not always be available. Therefore, methods for executing grasps are required, which perform well with information gathered from only standard stereo vision, and make only a few necessary assumptions about the task environment. We propose techniques that reactively modify the robot's learned motor primitives based on information derived from Early Cognitive Vision descriptors. The proposed techniques employ non-parametric potential fields centered on the Early Cognitive Vision descriptors to allow for curving hand trajectories around objects, and finger motions that adapt to the object's local geometry. The methods were tested on a real robot and found to allow for easier imitation learning of human movements and give a considerable improvement to the robot's performance in grasping tasks.

show abstract

“…The learning algorithm builds a hierarchy from a set of observations from a segmented object; the hierarchy is then used to recover the pose of the object in a cluttered scene. Preliminary results appeared in conference and workshop proceedings [20], [21], [22]. We note that even though we concentrate our evaluation on 3D data from Krüger et al, our system can in principle be applied to other 3D sources, such as dense stereo, or range data.…”

Section: Fig 2 Pose Estimation Systemmentioning

confidence: 99%

A Probabilistic Framework for 3D Visual Object Representation

Detry

¹

,

Pugeault

²

,

Piater

³

2009

IEEE Trans. Pattern Anal. Mach. Intell.

View full text Add to dashboard Cite

Abstract-We present an object representation framework that encodes probabilistic spatial relations between 3D features and organizes these features in a hierarchy. Features at the bottom of the hierarchy are bound to local 3D descriptors. Higher-level features recursively encode probabilistic spatial configurations of more elementary features. The hierarchy is implemented in a Markov network. Detection is carried out by a belief propagation algorithm, which infers the pose of high-level features from local evidence and reinforces local evidence from globally consistent knowledge, effectively producing a likelihood for the pose of the object in the detection scene. We also present a simple learning algorithm that autonomously builds hierarchies from local object descriptors. We explain how to use our framework to estimate the pose of a known object in an unknown scene. Experiments demonstrate the robustness of hierarchies to input noise, viewpoint changes and occlusions.

show abstract

“…When performing a grasping task, the robot uses a hierarchical Markov model of the object's ECVD geometry [Detry et al, 2008] to determine its pose, which can then be used to superimpose the ECVDs of the model back into the scene. The grasping techniques can therefore use geometric information of a partially occluded object.…”

Section: Appendix Dynamical Systems Motor Primitivesmentioning

confidence: 99%

Grasping With Vision Descriptors and Motor Primitives

2010

Proceedings of the 7th International Conference on Informatics in Control, Automation and Robotics

0

View full text Add to dashboard Cite

Abstract:Grasping is one of the most important abilities needed for future service robots. Given the task of picking up an object from betweem clutter, traditional robotics approaches would determine a suitable grasping point and then use a movement planner to reach the goal. The planner would require precise and accurate information about the environment and long computation times, both of which may not always be available. Therefore, methods for executing grasps are required, which perform well with information gathered from only standard stereo vision, and make only a few necessary assumptions about the task environment. We propose techniques that reactively modify the robot's learned motor primitives based on information derived from Early Cognitive Vision descriptors. The proposed techniques employ non-parametric potential fields centered on the Early Cognitive Vision descriptors to allow for curving hand trajectories around objects, and finger motions that adapt to the object's local geometry. The methods were tested on a real robot and found to allow for easier imitation learning of human movements and give a considerable improvement to the robot's performance in grasping tasks.

show abstract

Probabilistic Pose Recovery Using Learned Hierarchical Object Models

Cited by 9 publications

References 8 publications

Grasping with Vision Descriptors and Motor Primitives

Grasping with Vision Descriptors and Motor Primitives

A Probabilistic Framework for 3D Visual Object Representation

Grasping With Vision Descriptors and Motor Primitives

Contact Info

Product

Resources

About