ABCDEFGHIJKLMNOPQRSVWXYZAAABACADAEAFAGAHAIAJAK
1
Latest update:
🤖 = latest additions		July 31, 2025
2
Note: drop-down filters only work when open in Google SpreadsheetsNumber of entries: 48
3
Data for many entries sourced from Newbury et al. [2023]Grasp DatasetsSimulators
4
NameDescription
Planning Method
Training Data
End-effector Hardware
Object Configuration
Input Data
Output Pose
Corresponding Dataset
(see repository linked above)
Simulator
(see repository linked above)
Backbone
Metric(s) Used
Camera Position(s)
Language
Link(s)
License
Maintainer(s)Citation
Year (Initial Release)
5
🤖 DexGrasp AnythingDexGrasp Anything is a method that effectively integrates physical constraints into both the training and sampling phases of a diffusion-based generative model for multi-finger dexterous hand grasp generation, achieving state-of-the-art performance across nearly all open datasets.GenerativeSim, RealMulti-fingerSingulatedPoint cloudDexterous graspDexGrasp-Anything​PointTransformerGrasp success rate, Maximum Penetration, Diversity​PyTorch, Pythonhttps://github.com/4DVLab/DexGrasp-Anything

https://dexgraspanything.github.io/ 		MIT4DVLab, ShanghaiTech UniversityZhong, Yiming, Qi Jiang, Jingyi Yu, and Yuexin Ma. "Dexgrasp anything: Towards universal robotic dexterous grasping with physics awareness." In Proceedings of the Computer Vision and Pattern Recognition Conference, pp. 22584-22594. 2025.2025
6
🤖 FoundationGraspA foundation model-based TOG (Task oriented grasping) framework that leverages the open-ended knowledge from foundation models to learn generalizable TOG skills. Built on top of the public TOG benchmark TaskGrasp dataset, we contribute a multi-modal TOG dataset, Language and Vision Augmented TaskGrasp (LaViATaskGrasp) dataset, to support the training and evaluation of generalizable TOG methods.Grasp evaluation + SamplerSimTwo-fingerSingulatedRGBD image, Point cloud, Natural language, GraspGrasp success probabilityLanguage and Vision Augmented TaskGrasp (LaViA-TaskGrasp)​​Grasp success rate​Python, PyTorchhttps://github.com/mkt1412/GraspGPT_public

https://sites.google.com/view/foundationgrasp 		​Tang, Chao, Dehao Huang, Wenlong Dong, Ruinian Xu, and Hong Zhang. "Foundationgrasp: Generalizable task-oriented grasping with foundation models." IEEE Transactions on Automation Science and Engineering (2025).2025
7
🤖 GraspGenWe build upon the recent success on modeling the object-centric grasp generation process as an iterative diffusion process. Our proposed framework - GraspGen - consists of a Diffusion-Transformer architecture that enhances grasp generation, paired with an efficient discriminator to score and filter sampled grasps. We introduce a novel and performant on-generator training recipe for the discriminator.GenerativeSimTwo-finger, SuctionSingulatedPoint cloud, GraspGrasp success probabilityGraspGenIssac-SimPointTransformerGrasp success rate​Python, PyTorchhttps://graspgen.github.io/

https://github.com/NVlabs/GraspGen 		NVIDIANVIDIAMurali, Adithyavairavan, Balakumar Sundaralingam, Yu-Wei Chao, Wentao Yuan, Jun Yamada, Mark Carlson, Fabio Ramos, Stan Birchfield, Dieter Fox, and Clemens Eppner. "GraspGen: A Diffusion-based Framework for 6-DOF Grasping with On-Generator Training." arXiv preprint arXiv:2507.13097 (2025).2025
8
🤖 GraspGPTWe propose GraspGPT, a large language model (LLM) based TOG framework that leverages the open-end semantic knowledge from an LLM to achieve zero-shot generalization to novel concepts. We conduct experiments on Language Augmented TaskGrasp (LA-TaskGrasp) dataset and demonstrate that GraspGPT outperforms existing TOG methods on different held-out settings when generalizing to novel concepts out of the training set. Grasp evaluation + SamplerSimTwo-fingerSingulatedPoint cloud, Natural language, GraspGrasp success probabilityLanguage and Vision Augmented TaskGrasp (LaViA-TaskGrasp)​​Grasp success rate​Python, PyTorchhttps://github.com/mkt1412/GraspGPT_public

https://sites.google.com/view/graspgpt/ 		​Tang, Chao, Dehao Huang, Wenqi Ge, Weiyu Liu, and Hong Zhang. "Graspgpt: Leveraging semantic knowledge from a large language model for task-oriented grasping." IEEE Robotics and Automation Letters 8, no. 11 (2023): 7551-7558.2023
9
🤖 GraspSAMIn this paper, we introduce GraspSAM, an innovative extension of the Segment Anything Model (SAM) designed for prompt-driven and category-agnostic grasp detection. Unlike previous methods, which are often limited by small-scale training data, GraspSAM leverages SAM’s large-scale training and prompt-based segmentation capabilities to efficiently support both targetobject and category-agnostic grasping. Direct regressionSimTwo-fingerSingulatedRGB image"2D grasp rectangle (x, y, width, height, angle)"Jacquard, Grasp-Anything, Grasp-Anything++​ViTGrasp success rateOverheadPython, PyTorchhttps://github.com/gist-ailab/GraspSAM

https://gistailab.github.io/graspsam/ 		​Noh, Sangjun, Jongwon Kim, Dongwoo Nam, Seunghyeok Back, Raeyoung Kang, and Kyoobin Lee. "Graspsam: When segment anything model meets grasp detection." arXiv preprint arXiv:2409.12521 (2024).2025
10
🤖 NeuGraspNetNeuGraspNet is a novel, fully implicit 6DoF grasp detection method that re-interprets robotic grasping as surface rendering and predicts high-fidelity grasps from any random single viewpoint of a scene. Our method exploits a learned implicit geometric scene representation to perform global and local surface rendering. This enables effective grasp candidate generation (using global features) and grasp quality prediction (using local features from a shared feature space).Grasp evaluation + SamplerSimTwo-fingerPiled, StackedDepth imageGrasp success probability​PyBulletPointNetGrasp success rate, Declutter rate​Python, PyTorchhttps://github.com/pearl-robot-lab/neugraspnet

https://sites.google.com/view/neugraspnet 		​Jauhri, Snehal, Ishikaa Lunawat, and Georgia Chalvatzaki. "Learning Any-View 6DoF Robotic Grasping in Cluttered Scenes via Neural Surface Rendering." In RoboNerF: 1st Workshop On Neural Fields In Robotics at ICRA 2024.2024
11
🤖 ShapeGraspTask-oriented grasping of unfamiliar objects is a necessary skill for robots in dynamic in-home environments. Inspired by the human capability to grasp such objects through intuition about their shape and structure, we present a novel zero-shot task-oriented grasping method leveraging a geometric decomposition of the target object into simple, convex shapes that we represent in a graph structure, including geometric attributes and spatial relationships. Our approach employs minimal essential information—the object's name and the intended task—to facilitate zero-shot task-oriented grasping. Grasp evaluation + SamplerTraining-lessThree-fingerSingulatedRGBD image3D position and angle in table plane​​​Grasp success rate, Part selection acc.OverheadPythonhttps://github.com/samwli/ShapeGrasp

https://shapegrasp.github.io/ 		MIT​Li, Samuel, Sarthak Bhagat, Joseph Campbell, Yaqi Xie, Woojun Kim, Katia Sycara, and Simon Stepputtis. "Shapegrasp: Zero-shot task-oriented grasping with large language models through geometric decomposition." In 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 10527-10534. IEEE, 2024.2024
12
🤖 UniGraspTransformerWe introduce UniGraspTransformer, a universal Transformer-based network for dexterous robotic grasping that simplifies training while enhancing scalability and performance. Unlike prior methods such as UniDexGrasp++, which require complex, multi-step training pipelines, UniGraspTransformer follows a streamlined process: first, dedicated policy networks are trained for individual objects using reinforcement learning to generate successful grasp trajectories; then, these trajectories are distilled into a single, universal network. Reinforcement learning, Direct regressionSimMulti-fingerSingulatedPoint cloudDexterous grasp​Issac-Gym​Grasp success rateMulti-viewPyTorch, Pythonhttps://github.com/microsoft/UniGraspTransformer

https://dexhand.github.io/UniGraspTransformer/ 		MITMicrosoftWang, Wenbo, Fangyun Wei, Lei Zhou, Xi Chen, Lin Luo, Xiaohan Yi, Yizhong Zhang et al. "Unigrasptransformer: Simplified policy distillation for scalable dexterous robotic grasping." In Proceedings of the Computer Vision and Pattern Recognition Conference, pp. 12199-12208. 2025.2025
13
🤖 ZeroGraspZeroGrasp is a framework for simultaneous reconstruction and grasp pose prediction of multiple unseen objects from a single RGB-D image in near real-time. ZeroGrasp achieves the state-of-the-art results across all splits on GraspNet-1B benchmark and demonstrates strong generalization and robustoness in real-world scenarios.Grasp evaluation + SamplerSimTwo-fingerClutteredRGBD imageGrasp success probabilityZeroGrasp-11BIssac-Gym​Average PrecisionOverheadPython, PyTorchhttps://github.com/sh8/ZeroGrasp

https://sh8.io/#/zerograsp 		CC BY-NC 4.0Iwase, Shun, Muhammad Zubair Irshad, Katherine Liu, Vitor Guizilini, Robert Lee, Takuya Ikeda, Ayako Amma et al. "ZeroGrasp: Zero-Shot Shape Reconstruction Enabled Robotic Grasping." In Proceedings of the Computer Vision and Pattern Recognition Conference, pp. 17405-17415. 2025.2025
14
6-DoF GraspNet6-DoF GraspNet formulates the problem of grasp generation as sampling a set of grasps using a variational autoencoder and assess and refine the sampled grasps using a grasp evaluator model. Both Grasp Sampler and Grasp Refinement networks take 3D point clouds observed by a depth camera as input. Our model is trained purely in simulation and works in the real world without any extra steps.SamplingSimTwo-fingerSingulatedPoint cloud"6-DoF grasp pose (x, y, z, r, p, y)"ACRONYMPyRenderPointNet++Image-based Grasp Quality (IGQ), Grasp success rate, Precision, Robust grasp rate, Planning timeEye-in-handPython, TensorFlowhttps://github.com/NVlabs/6dof-graspnet MITNVIDIAMousavian, Arsalan, Clemens Eppner, and Dieter Fox. "6-dof graspnet: Variational grasp generation for object manipulation." In Proceedings of the IEEE/CVF international conference on computer vision, pp. 2901-2910. 2019.2019
15
AnyGraspUtilization of grasp correspondence across observations enables dynamic grasp tracking. Our model, AnyGrasp, can generate accurate, full-DoF, dense and temporally-smooth grasp poses efficiently, and works robustly against large depth sensing noise.
Direct regressionReal, SimTwo-fingerPiledPoint cloud"7-DoF grasp pose (x, y, z, r, p, y, gripper width)" GraspNet-1Billion PyRenderPointNet++Grasp success rate, Completion rateEye-in-hand, OverheadPython, PyTorchhttps://github.com/graspnet/anygrasp_sdk

https://graspnet.net/anygrasp 		CC BY-NC-SAMachine Vision and Intelligence Group, Shanghai Jiao Tong UniversityFang, Hao-Shu, Chenxi Wang, Hongjie Fang, Minghao Gou, Jirong Liu, Hengxu Yan, Wenhai Liu, Yichen Xie, and Cewu Lu. "Anygrasp: Robust and efficient grasp perception in spatial and temporal domains." IEEE Transactions on Robotics 39, no. 5 (2023): 3929-3945.2023
16
CAPGraspCAPGrasp is an R3 × SO(2)-equivariant 6-Degrees of Freedom (DoF) continuous approach-constrained generative grasp sampler. It includes a novel learning strategy for training CAPGrasp that eliminates the need to curate massive conditionally labeled datasets and a constrained grasp refinement technique that improves grasp poses while respecting the grasp approach directional constraints.SamplingSim, RealTwo-fingerSingulatedPoint cloud"6-DoF grasp pose (x, y, z, r, p, y)"ACRONYM, CustomIssac-GymVAEGrasp success rateSide viewPythonhttps://github.com/wengzehang/CAPGrasp

https://wengzehang.github.io/CAPGrasp/ 		​Weng, Zehang, Haofei Lu, Jens Lundell, and Danica Kragic. "CAPGrasp: An $\mathbb {R}^{3}\times\rm {SO (2)-equivariant} $ Continuous Approach-Constrained Generative Grasp Sampler." IEEE Robotics and Automation Letters (2024).2024
17
CaTGraspThis work proposes a framework to learn task-relevant grasping for industrial objects without the need of time-consuming real-world data collection or manual annotation. To achieve this, the entire framework is trained solely in simulation, including supervised training with synthetic label generation and self-supervised, hand-object interaction. In the context of this framework, CaTGrasp uses a novel, object-centric canonical representation at the category level, which allows establishing dense correspondence across object instances and transferring task-relevant grasps to novel instances.SamplingSimTwo-fingerPiledPoint cloud"6-DoF grasp pose (x, y, z, r, p, y)"CustomPyBulletPointNetGrasp success rateOverheadPython, PyTorchhttps://github.com/wenbowen123/catgrasp

https://sites.google.com/view/catgrasp 		MIT, Apache 2.0Bowen WenWen, Bowen, Wenzhao Lian, Kostas Bekris, and Stefan Schaal. "Catgrasp: Learning category-level task-relevant grasping in clutter from simulation." In 2022 International Conference on Robotics and Automation (ICRA), pp. 6401-6408. IEEE, 2022.2022
18
Contact-GraspNetContact-GraspNet treats 3D points of the recorded point cloud as potential grasp contacts. By rooting the full 6-DoF grasp pose and width in the observed point cloud, we can reduce the dimensionality of our grasp representation to 4-DoF which greatly facilitates the learning process. Our class-agnostic approach is trained on 17 million simulated grasps and generalizes well to real world sensor data.Direct regressionSimTwo-fingerStructuredPoint cloud"6-DoF grasp pose (x, y, z, r, p, y)"ACRONYMNVIDIA FleXPointNet++Grasp success rate, Grasp success at first trial, Number of re-graspsOblique viewPython, TensorFlowhttps://github.com/NVlabs/contact_graspnet ​NVIDIASundermeyer, Martin, Arsalan Mousavian, Rudolph Triebel, and Dieter Fox. "Contact-graspnet: Efficient 6-dof grasp generation in cluttered scenes." In 2021 IEEE International Conference on Robotics and Automation (ICRA), pp. 13438-13444. IEEE, 2021.2021
19
Deep Dexterous Grasping (DDG)Deep Dexterous Grasping (DDG) uses deep neural networks to improve the state-of-the-art in dexterous grasping of previously unseen objects using a single view camera.
SamplingSimMulti-fingerSingulatedDepth image"6-DoF grasp pose (x, y, z, r, p, y)"Deep Dexterous Grasping (DDG)MuJoCoVGG-16, ResNet-50Grasp success rate (top ranked)Oblique viewC++https://github.com/rusen/Grasp/

https://rusen.github.io/DDG/ 		GPL-3.0Umit Rusen AktasAktaş, Ümit Ruşen, Chao Zhao, Marek Kopicki, and Jeremy L. Wyatt. "Deep dexterous grasping of novel objects from a single view." International Journal of Humanoid Robotics 19, no. 02 (2022): 2250011.2019
20
Deep Dexterous Grasping in Clutter (DDGC)Deep Dexterous Grasping in Clutter (DDGC) takes an RGB image of the complete scene and the mask of the object to grasp and predicts a multi-fingered robotic grasp on the target objects. First, the objects' shapes are completed. Next, grasps are generated on the target object by first feeding the RGB image and the mask through an image encoder, and based on the image encoding 6D multi-finger grasp poses and configurations are generated. Finally, the hand is refined to be close to the surface of the target object but not in collision with it by using our parameter-free fully differentiable barrett kinematics layer.SamplingSimThree-fingerStructuredRGBD image"6-DoF grasp pose (x, y, z, r, p, y)"CustomPyBullet, GraspIt!​ε-quality grasps (10 top-scoring), Average clearance rate, Average grasp success rate, Grasp sampling time (10 grasps)Oblique view Python, PyTorchhttps://github.com/aalto-intelligent-robotics/DDGC

https://irobotics.aalto.fi/ddgc/ 		MITAalto University Intelligent RoboticsLundell, Jens, Francesco Verdoja, and Ville Kyrki. "Ddgc: Generative deep dexterous grasping in clutter." IEEE Robotics and Automation Letters 6, no. 4 (2021): 6899-6906.2021
21
DeepRLManipOur approach is to formulate the problem as a Markov decision process (MDP) with abstract yet generally applicable state and action representations. Finding a good solution to the MDP requires adding constraints on the allowed actions. We develop a specific set of constraints called hierarchical SE(3) sampling (HSE3S) which causes the robot to learn a sequence of gazes to focus attention on the task-relevant parts of the scene.Reinforcement learningSimTwo-fingerStructuredDepth imageGrasping PolicyCustomOpenRAVEDQN with gradient Monte CarloAverage grasp success rateEye-in-handPython, MATLABhttps://github.com/mgualti/DeepRLManip MITMarcus GualtieriGualtieri, Marcus, and Robert Platt. "Learning 6-dof grasping and pick-place using attention focus." In Conference on Robot Learning, pp. 477-486. PMLR, 2018.2018
22
Dex-Net 1.0 (MV-CNNs)The Dexterity Network (Dex-Net) is a research project including code, datasets, and algorithms for generating datasets of synthetic point clouds, robot parallel-jaw grasps and metrics of grasp robustness based on physics for thousands of 3D object models to train machine learning-based methods to plan robot grasps.Sampling, Exemplar methodsSimTwo-fingerSingulated3D object model mesh3D position and approach vectorDexNet 1.0AUTOLABAlexNetProbability of force closureOverheadPython, TensorFlowhttps://github.com/BerkeleyAutomation/dex-net Apache 2.0Berkeley Automation LabMahler, Jeffrey, Florian T. Pokorny, Brian Hou, Melrose Roderick, Michael Laskey, Mathieu Aubry, Kai Kohlhoff, Torsten Kröger, James Kuffner, and Ken Goldberg. "Dex-net 1.0: A cloud-based network of 3d objects for robust grasp planning using a multi-armed bandit model with correlated rewards." In 2016 IEEE international conference on robotics and automation (ICRA), pp. 1957-1964. IEEE, 2016.2016
23
Dex-Net 2.0 (GQ-CNN)The gqcnn Python package is for training and analysis of Grasp Quality Convolutional Neural Networks (GQ-CNNs). It is part of the ongoing Dexterity-Network (Dex-Net) project created and maintained by the AUTOLAB at UC Berkeley.Direct regressionRealTwo-fingerSingulatedDepth image"2D grasp rectangle (x, y, width, height, angle)", Grasp success probabilityDexNet 2.0AUTOLABNoneImage-based Grasp Quality (IGQ), Grasp success rate, Precision, Robust grasp rate, Planning timeOverheadPython, TensorFlowhttps://github.com/BerkeleyAutomation/gqcnn

https://berkeleyautomation.github.io/gqcnn/ 		​Berkeley Automation LabMahler, Jeffrey, Jacky Liang, Sherdil Niyaz, Michael Laskey, Richard Doan, Xinyu Liu, Juan Aparicio Ojea, and Ken Goldberg. "Dex-net 2.0: Deep learning to plan robust grasps with synthetic point clouds and analytic grasp metrics." arXiv preprint arXiv:1703.09312 (2017).2017
24
Dex-Net 2.1We model bin picking with a discrete-time Partially Observable Markov Decision Process that specifies states of the heap, point cloud observations, and rewards. We collect synthetic demonstrations of bin picking from an algorithmic supervisor uses full state information to optimize for the most robust collision-free grasp in a forward simulator based on pybullet to model dynamic object-object interactions and robust wrench space analysis from the Dexterity Network (Dex-Net) to model quasi-static contact between the gripper and object. Reinforcement learningSimTwo-fingerPiledRolloutsGrasping PolicyDexNet 2.1PyBulletNoneGrasp success rate, Average clearance rate, Picks per Hour (PPH), Average PrecisionOverhead​http://bit.ly/3XxKVro ​Berkeley Automation LabMahler, Jeffrey, and Ken Goldberg. "Learning deep policies for robot bin picking by simulating robust grasping sequences." In Conference on robot learning, pp. 515-524. PMLR, 2017.2017
25
Dex-Net 3.0We propose a compliant suction contact model that computes the quality of the seal between the suction cup and target object and determines whether or not the suction grasp can resist an external wrench (e.g. gravity) on the object. To evaluate a grasp, we measure robustness to perturbations in end-effector and object pose, material properties, and external wrenches. We train a Grasp Quality Convolutional Neural Network (GQ-CNN) on this dataset to classify suction grasp robustness from point clouds. Direct regressionSimSuctionSingulatedPoint cloudGrasping PolicyDexNet 3.0N/ANoneGrasp success rate, Planning time, Average PrecisionOverhead​http://bit.ly/3XEPVuo ​Berkeley Automation LabMahler, Jeffrey, Matthew Matl, Xinyu Liu, Albert Li, David Gealy, and Ken Goldberg. "Dex-net 3.0: Computing robust vacuum suction grasp targets in point clouds using a new analytic model and deep learning." In 2018 IEEE International Conference on robotics and automation (ICRA), pp. 5620-5627. IEEE, 2018.2018
26
Dex-Net 4.0We present Dexterity Network (Dex-Net) 4.0, a substantial extension to previous versions of Dex-Net that learns policies for a given set of grippers by training on synthetic datasets using domain randomization with analytic models of physics and geometry. We train policies for a parallel-jaw and a vacuum-based suction cup gripper on 5 million synthetic depth images, grasps, and rewards generated from heaps of three-dimensional objects. Reinforcement learningSimTwo-finger, SuctionPiledDepth imageGrasping PolicyDexNet 4.0N/ANoneGrasp success rate, Average clearance rate, Picks per Hour (PPH), Average PrecisionOverhead​http://bit.ly/3ViM4BG ​Berkeley Automation LabMahler, Jeffrey, Matthew Matl, Vishal Satish, Michael Danielczuk, Bill DeRose, Stephen McKinley, and Ken Goldberg. "Learning ambidextrous robot grasping policies." Science Robotics 4, no. 26 (2019): eaau4984.2019
27
Edge Grasp NetworkEdge Grasp Network is a novel method and neural network model that enables better grasp success rates relative to what is available in the literature. The method takes standard point cloud data as input and works well with single-view point clouds observed from arbitrary viewing directions.SamplingSimTwo-fingerPiledPoint cloud"6-DoF grasp pose (x, y, z, r, p, y)"CustomPyBulletPointNetGrasp success rate, Declutter rateEye-in-handPythonhttps://github.com/HaojHuang/Edge-Grasp-Network MITHaojie HuangHuang, Haojie, Dian Wang, Xupeng Zhu, Robin Walters, and Robert Platt. "Edge grasp network: A graph-based se (3)-invariant approach to grasp detection." In 2023 IEEE International Conference on Robotics and Automation (ICRA), pp. 3882-3888. IEEE, 2023.2023
28
EquiGraspFlowEquiGraspFlow is a flow-based SE(3)-equivariant 6-DoF grasp pose generative model that can learn complex conditional distributions on the SE(3) manifold while guaranteeing SE(3)-equivariance. Our model achieves the equivariance without relying on data augmentation, by using network architectures that guarantee it by construction.SamplingSim, RealTwo-fingerStructuredPoint cloud"6-DoF grasp pose (x, y, z, r, p, y)"ACRONYMIssac-GymVN-DGCNNGrasp success rate, Earth Mover’s Distance (EMD)Eye-in-handPythonhttps://github.com/bdlim99/EquiGraspFlow

https://equigraspflow.github.io/ 		MITLim, Byeongdo, Jongmin Kim, Jihwan Kim, Yonghyeon Lee, and Frank C. Park. "Equigraspflow: Se (3)-equivariant 6-dof grasp pose generative flows." In 8th Annual Conference on Robot Learning. 2024.2024
29
GCNGraspThe GCNGrasp framework uses the semantic knowledge of objects and tasks encoded in a knowledge graph to generalize to new object instances, classes and even new tasks. Our framework shows a significant improvement of around 12% on held-out settings compared to baseline methods which do not use semantics.SamplingSimTwo-fingerSingulatedPoint cloud"6-DoF grasp pose (x, y, z, r, p, y)"TaskGraspOpen3DPointNet++​Eye-in-handPython, PyTorchhttps://github.com/adithyamurali/TaskGrasp

https://sites.google.com/view/taskgrasp 		MITAdithya MuraliMurali, Adithyavairavan, Weiyu Liu, Kenneth Marino, Sonia Chernova, and Abhinav Gupta. "Same object, different grasps: Data and semantic knowledge for task-oriented grasping." In Conference on robot learning, pp. 1540-1557. PMLR, 2021.2021
30
geometric-object-grasperThis repository is an implementation of the paper 'A Geometric Approach for Grasping Unknown Objects with Multi-Fingered Hands' in PyBullet. Note that this is a Python implementation and as a result it runs slower than the initial C++ implementation.AnalyticalSimThree-fingerPiledPoint cloudGrasp preshapeCustomMuJoCoVAEGrasp success rateOblique view Python, PyTorchhttps://github.com/mkiatos/geometric-object-grasper ​Marios KiatosKiatos, Marios, Sotiris Malassiotis, and Iason Sarantopoulos. "A geometric approach for grasping unknown objects with multifingered hands." IEEE Transactions on Robotics 37, no. 3 (2020): 735-746.2021
31
GeomPickPlaceThis approach uses (a) object instance segmentation and shape completion to model the objects and (b) a regrasp planner to decide grasps and places displacing the models to their goals. It is critical for the planner to account for uncertainty in the perceived models, as object geometries in unobserved areas are just guesses, so we account for perceptual uncertainty by incorporating it into the regrasp planner's cost function.
AnalyticalSimTwo-fingerStructuredPoint cloud"6-DoF grasp pose (x, y, z, r, p, y)"CustomOpenRAVE PointNetPlace success rate, Grasp success rate, Number of grasp attempts, Number of re-grasps, Packing height of 5 cmOverheadPython, TensorFlowhttps://github.com/mgualti/GeomPickPlace MITMarcus GualteriGualtieri, Marcus, and Robert Platt. "Robotic pick-and-place with uncertain object instance segmentation and shape completion." IEEE robotics and automation letters 6, no. 2 (2021): 1753-1760.2021
32
ggcnn_plusIn this work, we focus on improving an existing shallow network, GG-CNN, and propose a new orthographic pipeline to enable the use of these models independently of the orientation of the camera.Direct regressionSimTwo-fingerSingulatedRGBD image"2D grasp rectangle (x, y, width, height, angle)"Cornell Grasping DatasetPyBullet​​​Python, Kerashttps://github.com/m-rios/ggcnn_plus ​Mario Rios MunozRios Munoz, Mario. "Grasping in 6DoF: An Orthographic Approach to Generalized Grasp Affordance Predictions." PhD dissertation, 2021.2021
33
Goal-Auxiliary Deep Deterministic Policy Gradient (GA-DDPG)We propose a new method for learning closed-loop control policies for 6D grasping. Our policy takes a segmented point cloud of an object from an egocentric camera as input, and outputs continuous 6D control actions of the robot gripper for grasping the object. We combine imitation learning and reinforcement learning and introduce a goal-auxiliary actor-critic algorithm for policy learning.Reinforcement learningSimTwo-fingerSingulatedPoint cloudGrasping PolicyACRONYMPyBulletPointNet++Grasp success rate, Number of grasp attemptsEye-in-handPython, PyTorchhttps://github.com/liruiw/GA-DDPG

https://sites.google.com/view/gaddpg 		MITLirui (Leroy) WangWang, Lirui, Yu Xiang, Wei Yang, Arsalan Mousavian, and Dieter Fox. "Goal-auxiliary actor-critic for 6d robotic grasping with point clouds." In Conference on Robot Learning, pp. 70-80. PMLR, 2022.2022
34
GRAFF (Grasp-Affordances)GRAFF is a deep RL dexterous robotic grasping policy that uses visual affordance priors learnt from humans for functional grasping. Our proposed model, called GRAFF for Grasp-Affordances, consists of two stages. First, we train a network to predict affordance regions from static images. Second, we train a dynamic grasping policy using the learned affordances. The key upshots of our approach are better grasping, faster learning, and generalization to successfully grasp objects unseen during policy training.Reinforcement learningSimMulti-fingerSingulatedRGBD imageGrasping PolicyCustomMuJoCo, Open3DResNet-50​Eye-in-handPython, TensorFlow, PyTorchhttps://github.com/priyankamandikal/graff

https://vision.cs.utexas.edu/projects/graff-dexterous-affordance-grasp/ 		MITPriyanka MandikalMandikal, Priyanka, and Kristen Grauman. "Learning dexterous grasping with object-centric visual affordances." In 2021 IEEE international conference on robotics and automation (ICRA), pp. 6169-6176. IEEE, 2021.2021
35
Grasp detection via Implicit Geometry and Affordance (GIGA)GIGA is a network that jointly detects 6 DOF grasp poses and reconstruct the 3D scene. GIGA takes advantage of deep implicit functions, a continuous and memory-efficient representation, to enable differentiable training of both tasks. GIGA takes as input a Truncated Signed Distance Function (TSDF) representation of the scene, and predicts local implicit functions for grasp affordance and 3D occupancy. By querying the affordance implict functions with grasp center candidates, we can get grasp quality, grasp orientation and gripper width at these centers.SamplingSimTwo-fingerPiledTruncated Signed Distance Function (TSDF)"6-DoF grasp pose (x, y, z, r, p, y)"CustomPyBullet, Open3DConvONets​Eye-in-handPython, PyTorchhttps://github.com/UT-Austin-RPL/GIGA

https://sites.google.com/view/rpl-giga2021 		MITUT Robot Perception and Learning LabJiang, Zhenyu, Yifeng Zhu, Maxwell Svetlik, Kuan Fang, and Yuke Zhu. "Synergies between affordance and geometry: 6-dof grasp detection via implicit representations." arXiv preprint arXiv:2104.01542 (2021).2021
36
Grasp Pose Detection (GPD)Grasp Pose Detection (GPD) is a package to detect 6-DOF grasp poses (3-DOF position and 3-DOF orientation) for a 2-finger robot hand (e.g., a parallel jaw gripper) in 3D point clouds. GPD takes a point cloud as input and produces pose estimates of viable grasps as output. The main strengths of GPD are: works for novel objects (no CAD models required for detection), works in dense clutter, and outputs 6-DOF grasp poses (enabling more than just top-down grasps).SamplingSimTwo-fingerPiledPoint cloud"6-DoF grasp pose (x, y, z, r, p, y)"CustomOpenRAVE3DNetNumber of objects, Number of grasp attempts, Number of successful grasps, Percentage of successful grasps, Percentage of objects removedEye-in-handC++, Python, PyTorchhttps://github.com/atenpas/gpd BSD 2-ClauseAndreas Ten PasTen Pas, Andreas, Marcus Gualtieri, Kate Saenko, and Robert Platt. "Grasp pose detection in point clouds." The International Journal of Robotics Research 36, no. 13-14 (2017): 1455-1473.2017
37
Grasp Proposal Network (GPNet)The end-to-end Grasp Proposal Network (GPNet) predicts a diverse set of 6-DOF grasps for an unseen object observed from a single and unknown camera view. GPNet builds on a key design of grasp proposal module that defines anchors of grasp centers at discrete but regular 3D grid corners, which is flexible to support either more precise or more diverse grasp predictions. Direct regressionSimTwo-fingerSingulatedPoint cloud"6-DoF grasp pose (x, y, z, r, p, y)"Grasp Proposal Network (GPNet)PyBulletPointNet++Grasp success rate, Coverage rateAgnosticPython, PyTorchhttps://github.com/Gorilla-Lab-SCUT/GPNet ​South China University of TechnologyWu, Chaozheng, Jian Chen, Qiaoyu Cao, Jianchi Zhang, Yunxin Tai, Lin Sun, and Kui Jia. "Grasp proposal networks: An end-to-end solution for visual learning of robotic grasps." Advances in Neural Information Processing Systems 33 (2020): 13174-13184.2020
38
GraspXLWe unify the generation of hand-object grasping motions across multiple motion objectives, diverse object shapes and dexterous hand morphologies in a policy learning framework GraspXL. The objectives are composed of the graspable area, heading direction, wrist rotation, and hand position. Reinforcement learningSimMulti-fingerSingulated3D object model mesh"6-DoF grasp pose (x, y, z, r, p, y)", Dexterous graspGraspXL RaiSim PPOGrasp success rateNAC++, Python, PyTorchhttps://github.com/zdchan/graspxl

https://eth-ait.github.io/graspxl/ 		MIT, BSD 3-ClauseHui ZhangZhang, Hui, Sammy Christen, Zicong Fan, Otmar Hilliges, and Jie Song. "Graspxl: Generating grasping motions for diverse objects at scale." In European Conference on Computer Vision, pp. 386-403. Cham: Springer Nature Switzerland, 2024.2024
39
ICG-NetThe method uses pointcloud data from a single arbitrary viewing direction as an input and generates an instance-centric representation for each partially observed object in the scene. This representation is further used for object reconstruction and grasp detection in cluttered table-top scenes.SamplingSim, RealTwo-fingerStructuredPoint cloud"6-DoF grasp pose (x, y, z, r, p, y)"VGNPyBulletUNetGrasp success rate, Declutter rateOblique viewPythonhttps://github.com/renezurbruegg/icg_net MITZurbrügg, René, Yifan Liu, Francis Engelmann, Suryansh Kumar, Marco Hutter, Vaishakh Patil, and Fisher Yu. "ICGNet: a unified approach for instance-centric grasping." In 2024 IEEE International Conference on Robotics and Automation (ICRA), pp. 4140-4146. IEEE, 2024.2024
40
Multi-FinGANMulti-FinGAN is a fast generative multi-finger grasp sampling method that synthesizes high quality grasps directly from RGB-D images in about a second. We achieve this by training in an end-to-end fashion a coarse-to-fine model composed of a classification
network that distinguishes grasp types according to a specific taxonomy and a refinement network that produces refined grasp poses and joint angles.		SamplingSimThree-fingerSingulatedRGBD image"6-DoF grasp pose (x, y, z, r, p, y)"Multi-FinGANGraspIt!ResNet-50ε-quality grasps, Grasp success rate, Grasp evaluation timeOblique view Python, PyTorchhttps://github.com/aalto-intelligent-robotics/Multi-FinGAN MITAalto University Intelligent RoboticsLundell, Jens, Enric Corona, Tran Nguyen Le, Francesco Verdoja, Philippe Weinzaepfel, Grégory Rogez, Francesc Moreno-Noguer, and Ville Kyrki. "Multi-fingan: Generative coarse-to-fine sampling of multi-finger grasps." In 2021 IEEE International Conference on Robotics and Automation (ICRA), pp. 4495-4501. IEEE, 2021.2021
41
MultiModalGraspingWe propose an approach to multi-modal grasp detection that jointly predicts the probabilities that several types of grasps succeed at a given grasp pose. Given a partial point cloud of a scene, the algorithm proposes a set of feasible grasp candidates, then estimates the probabilities that a grasp of each type would succeed at each candidate pose. Predicting grasp success probabilities directly from point clouds makes our approach agnostic to the number and placement of depth sensors at execution time.SamplingSimMulti-fingerStructuredPoint cloud"6-DoF grasp pose (x, y, z, r, p, y)"CustomDrakePointNet++Grasp success rate, Number of grasp attempts, Object removal rateAgnosticPython, TensorFlowhttps://github.com/mattcorsaro1/MultiModalGrasping MITMatt CorsaroCorsaro, Matt, Stefanie Tellex, and George Konidaris. "Learning to detect multi-modal grasps for dexterous grasping in dense clutter." In 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 4647-4653. IEEE, 2021.2021
42
OrbitGrasp (EquiFormerV2)Our main contribution is to propose an SE(3)-equivariant model that maps each point in the cloud to a continuous grasp quality function over the 2-sphere S 2 using spherical harmonic basis functions. Compared with reasoning about a finite set of samples, this formulation improves the accuracy and efficiency of our model when a large number of samples would otherwise be needed. In order to accomplish this, we propose a novel variation on EquiFormerV2 that leverages a UNet-style encoder-decoder architecture to enlarge the number of points the model can handle.Direct regressionSimTwo-fingerStructuredPoint cloud"6-DoF grasp pose (x, y, z, r, p, y)"CustomPyBulletUNetGrasp success rate, Grasp success rates averaged over independent episodesOblique viewPythonhttps://github.com/BoceHu/orbitgrasp

https://orbitgrasp.github.io/ 		MITBoce HuHu, B., Zhu, X., Wang, D., Dong, Z., Huang, H., Wang, C., ... & Platt, R. (2024). OrbitGrasp: $ SE (3) $-Equivariant Grasp Learning. arXiv preprint arXiv:2407.03531.2024
43
PointNet++ GraspingPointNet++ Grasping is an end-to-end approach to directly predict the poses, categories and scores (qualities) of all the grasps. It takes the whole sparse point clouds as the input and requires no sampling or search process. Moreover, to generate training data of multi-object scene, we propose a fast multi-object grasp detection algorithm based on Ferrari Canny metrics.Direct regressionSimTwo-fingerPiledPoint cloud"7-DoF grasp pose (x, y, z, r, p, y, approach angle)"PointNet++ GraspingPyBulletPointNet++Grasp success rate, Percentage of objects removedEye-in-handPythonhttps://github.com/pyni/PointNet2_Grasping_Data_Part/tree/master ​P. Ni, W. Zhang, X. Zhu, and Q. Cao, “PointNet++ Grasping: Learning An End-to-end Spatial Grasp Generation Algorithm from Sparse Point Clouds,” Mar. 21, 2020, arXiv: arXiv:2003.09644. doi: 10.48550/arXiv.2003.09644.2021
44
PointNetGPDPointNetGPD is an end-to-end grasp evaluation model to address the challenging problem of localizing robot grasp configurations directly from the point cloud. PointNetGPD is light-weighted and can directly process the 3D point cloud that locates within the gripper for grasp evaluation. Taking the raw point cloud as input, our proposed grasp evaluation network can capture the complex geometric structure of the contact area between the gripper and the object even if the point cloud is very sparse.SamplingSimTwo-fingerPiledPoint cloud"6-DoF grasp pose (x, y, z, r, p, y)"PointNetGPDGraspIt!PointNetGrasp success rate, Completion rateMulti-viewPython, PyTorchhttps://github.com/lianghongzhuo/PointNetGPD

https://lianghongzhuo.github.io/PointNetGPD/ 		MITHongzhuo LiangLiang, Hongzhuo, Xiaojian Ma, Shuang Li, Michael Görner, Song Tang, Bin Fang, Fuchun Sun, and Jianwei Zhang. "Pointnetgpd: Detecting grasp configurations from point sets." In 2019 International Conference on Robotics and Automation (ICRA), pp. 3629-3635. IEEE, 2019.2019
45
Probablistic Multi-fingered Grasp PlannerThis repo has several learning-based multi-fingered grasp planners implemented. We proposed multiple machine leanring models to predict the probability of grasp success from visual information of the object and grasp configuration. We then formulated grasp planning as inferring the grasp configuration which maximizes the probability of grasp success inside the grasp prediction deep networks.Active learningRealMulti-fingerSingulatedPoint cloudGrasp preshapeCustom​​Grasp success rateOblique view Python, TensorFlowhttps://github.com/QingkaiLu/multi-fingered_grasp_planners

https://robot-learning.cs.utah.edu/project/grasp_active_learning 		​Qingkai LuLu, Qingkai, Mark Van der Merwe, and Tucker Hermans. "Multi-fingered active grasp learning." In 2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 8415-8422. IEEE, 2020.2020
46
REgion-based Grasp Network (REGNet)REGNet is a REgion-based Grasp Network for End-to-end Grasp Detection in Point Clouds. It aims at generating the optimal grasp of novel objects from partial noisy observations.Direct regressionSimTwo-fingerPiledPoint cloud"7-DoF grasp pose (x, y, z, r, p, y, approach angle)"REgion-based Grasp Network (REGNet)MuJoCoPointNet++Grasp success rate, Completion rateAgnosticPython, PyTorchhttps://github.com/zhaobinglei/REGNet_for_3D_Grasping MITXi'an Jiaotong UniversityZhao, B., H. Zhang, X. Lan, H. Wang, Z. Tian, and N. Zheng. "REGNet: REgion-based grasp network for single-shot grasp detection in point clouds. arXiv." arXiv preprint arXiv:2002.12647 (2020).2020
47
RGBD-GraspRGBD-Grasp is a pipeline that decouples 7-DoF grasp detection into two sub-tasks where RGB and depth information are processed separately. In the first stage, an encoder-decoder like convolutional neural network Angle-View Net(AVN) is proposed to predict the SO(3) orientation of the gripper at every location of the image. Consequently, a Fast Analytic Searching (FAS) module calculates the opening width and the distance of the gripper to the grasp point.Direct regressionSim, RealTwo-fingerStructuredRGBD image"7-DoF grasp pose (x, y, z, r, p, y, gripper width)"GraspNet-1Billion​ResNet-50Grasp success rateEye-in-handPython, PyTorchhttps://github.com/GouMinghao/rgb_matters MITShanghai Jiao Tong UniversityGou, Minghao, Hao-Shu Fang, Zhanda Zhu, Sheng Xu, Chenxi Wang, and Cewu Lu. "Rgb matters: Learning 7-dof grasp poses on monocular rgbd images." In 2021 IEEE International Conference on Robotics and Automation (ICRA), pp. 13459-13466. IEEE, 2021.2021
48
Robust Grasp Planning Over Uncertain Shape CompletionsWe present a method for planning robust grasps over uncertain shape completed objects. For shape completion, a deep neural network is trained to take a partial view of the object as input and outputs the completed shape as a voxel grid. The key part of the network is dropout layers which are enabled not only during training but also at run-time to generate a set of shape samples representing the shape uncertainty through Monte Carlo sampling. AnalyticalSimThree-fingerSingulatedvoxelized occupancy grid"6-DoF grasp pose (x, y, z, r, p, y)"CustomGraspIt!​​Oblique view, Side viewPython, PyTorchhttps://github.com/aalto-intelligent-robotics/robust_grasp_planning_over_uncertain_shape_completions MITJens Lundell, Intelligent Robotics Lab, Aalto UniversityLundell, Jens, Francesco Verdoja, and Ville Kyrki. "Robust grasp planning over uncertain shape completions." In 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 1526-1532. IEEE, 2019.2019
49
ROI-GDROI-GD is proposed to provide a feasible solution to this problem based on Region of Interest (ROI), which is the region proposal for objects. ROI-GD uses features from ROIs to detect grasps instead of the whole scene. It has two stages: the first stage is to provide ROIs in the input image and the second-stage is the grasp detector based on ROI features. ROI-basedRealTwo-fingerPiledRGBD image"2D grasp rectangle (x, y, width, height, angle)"Visual Manipulation Relationship Dataset (VMRD)N/AResNet-101Prediction success rate, Grasp success rateOblique view Python, PyTorchhttps://github.com/ZhangHanbo/Visual-Manipulation-Relationship-Network-Pytorch ​H. Zhang, X. Lan, S. Bai, X. Zhou, Z. Tian, and N. Zheng, “ROI-based Robotic Grasp Detection for Object Overlapping Scenes,” Mar. 14, 2019, arXiv: arXiv:1808.10313. 2019
50
Single-Shot SE(3) Grasp Detection (S4G)S4G is a grasping proposal algorithm to regress SE(3) pose from single camera point cloud (depth only, no RGB information). S4G is trained only on synthetic dataset with YCB Objects, it can generate to real world grasping with unseen objects that has never been used in the training.SamplingSimTwo-fingerPiledPoint cloud"6-DoF grasp pose (x, y, z, r, p, y)"CustomMuJoCoPointNet ​Oblique view Python, PyTorchhttps://github.com/yzqin/s4g-release

https://sites.google.com/view/s4ggrapsing		​Yuzhe QinQin, Yuzhe, Rui Chen, Hao Zhu, Meng Song, Jing Xu, and Hao Su. "S4g: Amodal single-view single-shot se (3) grasp detection in cluttered scenes." In Conference on robot learning, pp. 53-65. PMLR, 2020.2020
51
UniGraspWe propose UniGrasp, an efficient data-driven grasp synthesis method that considers both the object geometry and gripper attributes as inputs. UniGrasp is based on a novel deep neural network architecture that selects sets of contact points from the input point cloud of the object. The proposed model is trained on a large dataset to produce contact points that are in force closure and reachable by the robot hand. By using contact points as output, we can transfer between a diverse set of multifingered robotic hands.Direct regressionSimMulti-fingerSingulatedPoint cloudGrasp PointsUniGraspBulletPointNet++​Eye-in-handPython, TensorFlowhttps://github.com/stanford-iprl-lab/UniGrasp

https://sites.google.com/view/unigrasp		MITStanford Interactive Perception and Robot Learning LabShao, Lin, Fabio Ferreira, Mikael Jorda, Varun Nambiar, Jianlan Luo, Eugen Solowjow, Juan Aparicio Ojea, Oussama Khatib, and Jeannette Bohg. "Unigrasp: Learning a unified model to grasp with multifingered robotic hands." IEEE Robotics and Automation Letters 5, no. 2 (2020): 2286-2293.2020
52
Volumetric Grasping Network (VGN)VGN is a 3D convolutional neural network for real-time 6 DOF grasp pose detection. The network accepts a Truncated Signed Distance Function (TSDF) representation of the scene and outputs a volume of the same spatial resolution, where each cell contains the predicted quality, orientation, and width of a grasp executed at the center of the voxel. The network is trained on a synthetic grasping dataset generated with physics simulation.Direct regressionSimTwo-fingerPiledTruncated Signed Distance Function (TSDF)"6-DoF grasp pose (x, y, z, r, p, y)"VGNPyBullet​​Eye-in-handPython, PyTorchhttps://github.com/ethz-asl/vgn BSD 3-ClauseAutonomous Systems Lab @ ETH ZürichBreyer, Michel, Jen Jen Chung, Lionel Ott, Roland Siegwart, and Juan Nieto. "Volumetric grasping network: Real-time 6 dof grasp detection in clutter." In Conference on Robot Learning, pp. 1602-1611. PMLR, 2021.2021