34 Hits in 10.6 sec

Sim-to-Real 6D Object Pose Estimation via Iterative Self-training for Robotic Bin Picking [article]

Kai Chen, Rui Cao, Stephen James, Yichuan Li, Yun-Hui Liu, Pieter Abbeel, Qi Dou
2022 arXiv   pre-print
In this paper, we propose an iterative self-training framework for sim-to-real 6D object pose estimation to facilitate cost-effective robotic grasping.  ...  Our method is also able to improve robotic bin-picking success by 19.54%, demonstrating the potential of iterative sim-to-real solutions for robotic applications.  ...  The work was supported by the Hong Kong Centre for Logistics Robotics.  ... 
arXiv:2204.07049v2 fatcat:ex5fmjaphnh6bicou2bndkxkbi

DemoGrasp: Few-Shot Learning for Robotic Grasping with Human Demonstration [article]

Pengyuan Wang, Fabian Manhardt, Luca Minciullo, Lorenzo Garattoni, Sven Meie, Nassir Navab, Benjamin Busam
2021 arXiv   pre-print
To this end, most approaches either compute the full 6D pose for the object of interest or learn to predict a set of grasping points.  ...  Finally, we transfer the a-priori knowledge from the relative pose between object and human hand with the estimate of the current object pose in the scene into necessary grasping instructions for the robot  ...  Bousmalis, “Sim-to-real via sim- from rgb images,” in ECCV, 2018. to-sim: Data-efficient robotic grasping via randomized-to-canonical [28] Y.  ... 
arXiv:2112.02849v1 fatcat:kpwrzj6r4re6rodinzlrer25si

Physics-based Scene-level Reasoning for Object Pose Estimation in Clutter [article]

Chaitanya Mitash, Abdeslam Boularias, Kostas Bekris
2019 arXiv   pre-print
Furthermore, confident estimates are used to label online real images from multiple views and re-train the process in a self-learning pipeline.  ...  This work proposes an autonomous process for pose estimation that spans from data generation to scene-level reasoning and self-learning.  ...  An example of such a setup exists in current day warehouses, where robots are being deployed for tasks such as picking from bins, packing and sorting.  ... 
arXiv:1806.10457v2 fatcat:hngnpzbfejhqbk6nez6pkeu6m4

You Only Demonstrate Once: Category-Level Manipulation from Single Visual Demonstration [article]

Bowen Wen, Wenzhao Lian, Kostas Bekris, Stefan Schaal
2022 arXiv   pre-print
The demonstration is reprojected to a target trajectory tailored to a novel object via the canonical representation.  ...  Nevertheless, it often requires expensive real-world data collection and manual specification of semantic keypoints for each object category and task.  ...  These methods range from training 6D pose estimators with CAD models to end-to-end reinforcement learning from repeated robot interaction with the task object.  ... 
arXiv:2201.12716v2 fatcat:qbgchamnlvfvhhu6avtulp6yq4

Visual Foresight Tree for Object Retrieval from Clutter with Nonprehensile Rearrangement [article]

Baichuan Huang, Shuai D. Han, Jingjin Yu, Abdeslam Boularias
2021 arXiv   pre-print
We first show that a deep neural network can be trained to accurately predict the poses of the packed objects when the robot pushes one of them.  ...  Object retrieval in dense clutter is an important skill for robots to operate in households and everyday environments effectively.  ...  To train GN, we set the reward to be 1 for grasps where the robot successfully picks up only the target object, and 0 otherwise. GN is the reward estimator for states in VFT (in Section V-C).  ... 
arXiv:2105.02857v2 fatcat:lfmfxh7e4navdnudnzak2oxt6i

2021 Index IEEE Robotics and Automation Letters Vol. 6

2021 IEEE Robotics and Automation Letters  
Departments and other items may also be covered if they have been judged to have archival value. The Author Index contains the primary entry for each item, listed under the first author's name.  ...  ., +, LRA Oct. 2021 7791-7798 Adversarial Training on Point Clouds for Sim-to-Real 3D Object Detection.  ...  ., +, LRA Oct. 2021 8363-8370 Adversarial Training on Point Clouds for Sim-to-Real 3D Object Detection.  ... 
doi:10.1109/lra.2021.3119726 fatcat:lsnerdofvveqhlv7xx7gati2xu

Learning Dexterous Manipulation from Suboptimal Experts [article]

Rae Jeong, Jost Tobias Springenberg, Jackie Kay, Daniel Zheng, Yuxiang Zhou, Alexandre Galashov, Nicolas Heess, Francesco Nori
2021 arXiv   pre-print
It represents the optimal policy via importance sampling from a learned prior and is well-suited to take advantage of mixed data distributions.  ...  reference behaviors to bootstrap a complex manipulation task on a simulated bimanual robot with human-like hands.  ...  Self- supervised sim-to-real adaptation for visual robotic manipulation, 2019. [21] A. Kurenkov, A. Mandlekar, R. Martin-Martin, S. Savarese, and A. Garg.  ... 
arXiv:2010.08587v2 fatcat:l3o7m2ht6fakhiuzxbu66trg2i

Table of Contents

2022 IEEE Robotics and Automation Letters  
Chen A Sim-to-Real Object Recognition and Localization Framework for Industrial Robotic Bin Picking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .X. Li, R.  ...  Wensing iCaps: Iterative Category-Level Object Pose and Shape Estimation . . . . . X. Deng, J. Geng, T. Bretl, Y. Xiang, and D.  ... 
doi:10.1109/lra.2022.3165102 fatcat:enjzebowe5hn7hsfwklc7nieuy

2020 Index IEEE Robotics and Automation Letters Vol. 5

2020 IEEE Robotics and Automation Letters  
., +, LRA Oct. 2020 6686-6693 A Bin-Picking Benchmark for Systematic Evaluation of Robotic Pick-and- Place Systems.  ...  Sim-To-Real Transfer Learning Approach for Tracking Multi-DOF Ankle Motions Using Soft Strain Sensors.  ... 
doi:10.1109/lra.2020.3032821 fatcat:qrnouccm7jb47ipq6w3erf3cja

Interactive Open-Ended Learning for 3D Object Recognition [article]

S. Hamidreza Kasaei
2019 arXiv   pre-print
In this context, "open-ended" implies that the set of categories to be learned is not known in advance, and the training instances are extracted from actual experiences of a robot, and thus become gradually  ...  In particular, this architecture provides perception capabilities that will allow robots to incrementally learn object categories from the set of accumulated experiences and reason about how to perform  ...  In this step, the robot captures a point cloud of the scene and computes a list of object hypotheses containing both objects6D pose and recognized label (i.e., object recognition and pose estimation  ... 
arXiv:1912.09539v1 fatcat:aksfik7earfafpe7n5amsehc2e

Solving Rubik's Cube with a Robot Hand [article]

OpenAI, Ilge Akkaya, Marcin Andrychowicz, Maciek Chociej, Mateusz Litwin, Bob McGrew, Arthur Petron, Alex Paino, Matthias Plappert, Glenn Powell, Raphael Ribas, Jonas Schneider (+7 others)
2019 arXiv   pre-print
We demonstrate that models trained only in simulation can be used to solve a manipulation problem of unprecedented complexity on a real robot.  ...  The combination of ADR with our custom robot platform allows us to solve a Rubik's cube with a humanoid robot hand, which involves both control and state estimation problems.  ...  Acknowledgements We would like to thank Shadow Robot Company Ltd. for building, maintaining, and improving the Shadow Dexterous Hand, PhaseSpace Inc. for building custom Rubik's cubes and supporting our  ... 
arXiv:1910.07113v1 fatcat:pqy6tcdd6jfuximspvl6nia2su

Strategy for an Autonomous Behavior that Guarantees a Qualitative Fine Adjustment at the Target Pose of a Collaborating Mobile Robot

Kai Waelti
2021 Zenodo  
This thesis introduces a strategy that leverages a pose graph-based localization approach to reduce positioning errors at target poses of collaborating mobile robots.  ...  As a basis for this strategy, an extensible software architecture for socially acceptable navigation of mobile robots is designed and implemented using ROS components.  ...  It al- lows clients to subscribe to topics using sim- ple JSON or BSON objects.  ... 
doi:10.5281/zenodo.4742013 fatcat:637baqamlffkfda5cvvxzegyuy

Learning Neural Textual Representations for Citation Recommendation

Binh Thanh Kieu, Inigo Jauregi Unanue, Son Bao Pham, Hieu Xuan Phan, Massimo Piccardi
2021 2020 25th International Conference on Pattern Recognition (ICPR)  
366 TSDM: Tracking by SiamRPN++ with a Depth-Refiner and a Mask- Generator DAY 3 -Jan 14, 2021 Feng, Hangtao; Zhang, Lu; Yang, Xu; Liu, Zhiyong 369 MixedFusion: 6D Object Pose Estimation from  ...  DAY 3 -Jan 14, 2021 Guo, Haifeng; Lu, Tong; Wu, Yirui 1529 Dynamic Low-Light Image Enhancement for Object Detection Via End-To-End Training DAY 3 -Jan 14, 2021 Zhu, Yongpei; Zhou, Zicong; Liao  ... 
doi:10.1109/icpr48806.2021.9412725 fatcat:3vge2tpd2zf7jcv5btcixnaikm

Cost-Efficient Global Robot Navigation in Rugged Off-Road Terrain

T. Braun
2011 Künstliche Intelligenz  
Abstract This thesis addresses the problem of finding a global robot navigation strategy for rugged off-road terrain which is robust against inaccurate self-localization and scalable to large environments  ...  Thanks to Norbert Schmitz for maintaining and fixing many things in our software framework, Daniel Schmidt for putting up with the chores of teaching and undergraduate education, Jochen Hirth for his work  ...  The mathematical technique to convert between a 6D robot pose (x, y, z, roll, pitch, yaw) specified in the W ORLD coordinate system and the corresponding transformation matrix M W ORLD ROBOT is documented  ... 
doi:10.1007/s13218-011-0088-9 fatcat:u4ssainmcvc3jmqdon6gpjg674

Scalable, physics-aware 6D pose estimation for robot manipulation

Chaitanya Mitash
Then it self-improves over time by acquiring and labeling real-world images via a search-based pose estimation process.  ...  Robot Manipulation often depend on some form of pose estimation to represent the state of the world and allow decision making both at the task-level and for motion or grasp planning.  ...  Self-Learning via Multi-view Pose Estimation Given access to an object detector and a pose estimation process trained with the physics-based simulator, the self-learning pipeline labels real-world images  ... 
doi:10.7282/t3-s5fk-ht60 fatcat:lndswo56sfhdvkmsmuigsvr4hi
« Previous Showing results 1 — 15 out of 34 results