Object recognition and full pose registration from a single image for robotic manipulation

By Alvaro Collet, Dmitry Berenson, Siddhartha S. Srinivasa, D. Ferguson. 2009

316 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition, Reinforcement Learning

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 Reinforcement Learning

Citations in the corpus

Domain randomization for transferring deep neural networks from simulation to the real world (2017, methodology): “There has also been some success with only monocular camera images [5].”
Deep visual foresight for planning robot motion (2016, methodology): “Learning and optimization-based methods have been applied to various parts of the state-estimation problem, such as object recognition [11], pose registration [12], and dynamics learning [13].”
Deep learning for detecting robotic grasps (2013, methodology): “sensors, etc. This makes the problem of grasping signiﬁcantly more challenging [4], as the algorithm must use more limited and potentially noisier information to detect a good grasp. While some works [10, 46] simply attempt to estimate the poses of known objects and then apply full-model grasping al”
Data-Driven Grasp Synthesis—A Survey (2013, background): “Examples for objects from an earlier version of the system [47] are shown in Fig.”
Unsupervised feature learning for 3D scene labeling (2014, background): “There is a large body of literature on 3D scene labeling in indoor environments ([1], [2], [3], [4], [5]).”
Multimodal deep learning for robust RGB-D object recognition (2015, background): “ns. Algorithms that only rely on color information typically work well for textured objects. However, they are often difﬁcult to use in scenarios were 3D pose information about the object is required [6]. On the other hand, purely shapebased approaches are often useful for large objects [20], althou”
Detection-based object labeling in 3D scenes (2012, methodology): “[4] used only visual features to build sparse 3D models of textured objects and efficiently match them.”
Close-range scene segmentation and reconstruction of 3D point cloud maps for mobile manipulation in domestic environments (2009, methodology): “We identify two main modules, namely the Scene Interpreter which extracts the supporting planes and segments the object clusters, and the Object Modeler, which creates the shape and surface models for each segmented object cluster.”
Robust 3D visual tracking using particle filtering on the special Euclidean group: A combined approach of keypoint and edge features (2011, background): “Another interest is in multiple object recognition and tracking [29] which is necessary for a realistic scenario.”
Discriminative Mixture-of-Templates for Viewpoint Classification (2010, methodology): “[19, 20]) and performed fairly well on instances of objects, preferably with texture.”
Odessa: enabling interactive perception applications on mobile devices (2011, background): “Object and Pose Recognition Figure 2(b) shows the data-flow graph for the object instance and pose recognition application [26].”
6-DoF object pose from semantic keypoints (2017, background): “To relax the assumption of known 2D landmarks, a number of approaches [5], [6], [7] have considered the detection of discriminative image keypoints, such as SIFT [8], with highly textured objects.”
Deep Object-Centric Representations for Generalizable Robot Learning (2017, methodology): “manipulation, SIFT features have also been used for 3D pose estimation and object localization, using object-specific training data gathered individually for the task [23], [24].”
The MOPED framework: Object recognition and pose estimation for manipulation (2011, methodology): “Solutions such as grouping (Lowe 1987), interpretation trees (Grimson 1991) or image space clustering (Collet et al. 2009) are often used, but false positives often arise from algorithms not being able to handle unexpected scene complexity.”