Robotic Grasping of Novel Objects using Vision

By Ashutosh Saxena, Justin Driemeyer, Andrew Y. Ng. 2008

1,045 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition, Control and Systems Engineering, Robotics & Control

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Aerospace Engineering

In Computer Vision and Pattern Recognition

In Control and Systems Engineering

In Robotics & Control

Citations in the corpus

Learning Trajectory Preferences for Manipulators via Iterative Improvement (2013, neutral): “Our work complements previous works on grasping items [30, 21], pick and place tasks [11], and detecting bar code for grocery checkout [16].”
Learning human activities and object affordances from RGB-D videos (2012, methodology): “Saxena et al. (2008) used supervised learning to detect grasp affordances for grasping novel objects.”
Supersizing self-supervision: Learning to grasp from 50K tries and 700 robot hours (2015, background): “Apart from initialization, we have no human involvement in the process of data collection.”
Deep learning for detecting robotic grasps (2013, methodology): “These are computed after the image is aligned to the gripper so that they are always relative to the gripper plates.”
Using Simulation and Domain Adaptation to Improve Efficiency of Deep Robotic Grasping (2018, methodology): “[11] used rendered objects to learn a vision-based grasping model.”
Dex-Net 2.0: Deep Learning to Plan Robust Grasps with Synthetic Point Clouds and Analytic Grasp Metrics (2017, methodology): “an image detection problem [31, 54], where a deep CNN is used to classify image regions as graspable based on an adaptive or fixed set of candidates.”
Real-time grasp detection using convolutional neural networks (2014, background): “of finding grasps solely from RGB-D data [13].”
Data-Driven Grasp Synthesis—A Survey (2013, methodology): “Similar to [80], simple 2D ﬁlters are used that can be rapidly convolved with an image.”
Adapting Deep Visuomotor Representations with Weak Pairwise Constraints (2015, positive): “used rendered objects to learn to grasp from vision [7].”
Human Activity Detection from RGBD Images (2011, background): “…with another person in a non-contact manner, and physically assistive robots, which can physically help people (Feil-Seifer and Matari´e, 2005; Tapus, T¸˘apus¸, and Matari´e, 2008; Nguyen et al., 2008; Li et al., 2011; Saxena, Driemeyer, and Ng, 2008; Jiang et al., 2011; Saxena et al., 2001).”
An overview of 3D object grasp synthesis algorithms (2012, background): “Other techniques learn to identify grasping regions in an object image [63,64].”
Multi-Task Domain Adaptation for Deep Learning of Instance Grasping from Simulation (2017, neutral): “One alternative solution to large scale data collection in robotics is to run robot experiments in simulation [4], [5], [6], [7], [8], [9].”
Closing the Loop for Robotic Grasping: A Real-time, Generative Grasp Synthesis Approach (2018, methodology): “Some techniques work with known items, associating good grasp points with an ofﬂine database of object models or shapes [6, 8, 22], or familiar items, based on object classes [28] or object parts [7], but are unable to generalise to new objects.”
Learning spatial relationships between objects (2011, background): “We are not only interested in determining valid grasp points (Saxena et al. 2008), but are interested in the structure of a scene as a whole.”
More Than a Feeling: Learning to Grasp and Regrasp Using Vision and Touch (2018, neutral): “As an alternative, data-driven approaches have sought to predict grasp outcomes from human supervision [19, 29, 14, 22], simulation [20, 16, 25], or autonomous robotic data collection [26, 23], typically using visual or depth observations.”
50 Years of object recognition: Directions forward (2013, background): “Recent work on affordances has also focused on its relation to robotics, by building systems that use vision based systems to determine how an object should be grasped and manipulated [131, 132].”
Dex-Net 3.0: Computing Robust Robot Suction Grasp Targets in Point Clouds using a New Analytic Model and Deep Learning (2017, background): “While grasp planning searches for gripper configurations that maximize a quality metric derived from mechanical wrench space analysis [25], human labels [29], or selfsupervised labels [20], suction grasps are often planned”