Automatic grasp planning using shape primitives

By Andrew T. Miller, S. Knoop, Henrik Iskov Christensen, Peter K. Allen. 2003

795 citations recorded by the source corpus.

Subfields: Biomedical Engineering, Control and Systems Engineering

How researchers see this paper

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

In Biomedical Engineering

In Control and Systems Engineering

Citations in the corpus

Supersizing self-supervision: Learning to grasp from 50K tries and 700 robot hours (2015, methodology): “Other approaches to predict grasping includes using simulators such as Graspit [17], [18].”
Deep learning for detecting robotic grasps (2013, methodology): “From this 24x24 pixel image, seven channels’ worth of features are extracted, giving 24x24x7 = 4032 input features.”
Imitation learning for locomotion and manipulation (2007, methodology): “The exponentiated gradient variant was used in the grasp prediction experiments described below in Section 3.”
Real-time grasp detection using convolutional neural networks (2014, background): “Significant past work uses 3-D simulations to find good grasps [3] [4] [5] [6] [7].”
Data-Driven Grasp Synthesis—A Survey (2013, background): “4a) Primitive Shape Decomposition [35].”
Learning Visual Affordances for Robotic Manipulation (2019, methodology): “More recent data-driven approaches ( [40, 123, 104, 117]) use ConvNets to provide bounding box proposals or segmentations, followed by geometric registration to estimate object poses, which ultimately guide handcrafted picking heuristics ([13, 75]).”
Learning to place new objects in a scene (2012, background): “While there has been significant previous work on grasping objects (e.g., Bicchi and Kumar, 2000; Ciocarlie and Allen, 2008; Miller et al., 2003; Saxena et al., 2006, 2008b,c; Berenson et al., 2009; Rao et al., 2010; Hsiao et al., 2009; Brook et al., 2011; Le et al., 2010; Jiang et al., 2011a; Dogar”
Data-driven grasping with partial sensor data (2009, background): “An important body of recent work in dexterous grasping has focused on empirical approaches, such as simulation [23], [24], [12], demonstration [1] and stochastic optimization [5], rather than anayltical solutions.”
Learning robust, real-time, reactive robotic grasping (2019, background): “…with known items, associating good grasp points with an offline database of object models or shapes (Detry et al., 2009; Goldfeder et al., 2007; Miller et al., 2003), or familiar items, based on object classes (Saxena et al., 2008) or object parts (El-Khoury and Sahbani, 2008), but are unable…”
Hand Posture Subspaces for Dexterous Robotic Grasping (2009, methodology): “Miller et al. (2003) used Cutkosky’s grasp taxonomy concept to define a number of starting positions, or pre-grasps, when searching for good grasps of a given object using a robotic hand.”
Learning to search: Functional gradient techniques for imitation learning (2009, methodology): “To produce grasp configurations, we use a control strategy similar to that used in the GraspIt! system (Miller et al., 2003), although in this case, we constrain the wrist axis to align with a single approach direction (the palm always faces the object).”
AnyGrasp: Robust and Efficient Grasp Perception in Spatial and Temporal Domains (2022, background): “D knowledge of the objects [2], [12], or approximate the objects into a set of primitive shapes [13], [14].”
Efficient grasping from RGBD images: Learning using a new rectangle representation (2011, methodology): “[12] used heuristic rules to generate and evaluate grasps for three-fingered hands by assuming that the objects are made of basic shapes such as spheres, boxes, cones and cylinders, each with pre-computed grasp primitives.”
Pick and Place Without Geometric Object Models (2017, background): “explored this in the context of grasp synthesis [6].”
Learning grasp affordance densities (2011, background): “The most popular 3D model for grasping is the 3D mesh, which allows for studying contact forces, and for finding a force-closure optimal grasp therefrom [1, 17, 21, 30].”
Active learning of visual descriptors for grasping using non-parametric smoothed beta distributions (2012, background): “After this initial research, the focus shifted to the relation between perception and action and several approaches tried to extract visual information to plan grasping actions: based on 2D information [14,15], approximating to known geometric shapes [16] or using range information [17].”
Real-World Robot Learning with Masked Visual Pre-training (2022, background): “, joint velocities, end49 effector poses, etc) directly from observations [17, 18, 19], without the need to perform explicit 3D 50 scene understanding [20], grasp planning [21], and motion planning [22, 23].”
Graspit! A versatile simulator for robotic grasping (2004, methodology): “We have implemented a grasp planner for the Barrett hand that uses a simplified version of an object built out of shape primitives, such as cubes, spheres, and cylinders, and uses heuristic grasp strategies defined for these shapes to generate a set of possible grasping configurations [19].”
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.”