12,995 Hits in 4.4 sec

Curriculum-Based Deep Reinforcement Learning for Adaptive Robotics: A Mini-Review

Gupta Kashish, Najjaran Homayoun
2021 International Journal of Robotic Engineering  
Recent progress in deep reinforcement learning has corroborated to its potential to train such autonomous and robust agents.  ...  At the same time, the introduction of curriculum learning has made the reinforcement learning process significantly more efficient and allowed for training on much broader tasks.  ...  Breyer M, Furrer F, Novkovic T, Siegwart R, Nieto J (2018) Flexible robotic grasping with sim-to-real transfer based reinforcement learning. 25.  ... 
doi:10.35840/2631-5106/4131 fatcat:tnoa4vd4yrgnpjzesxr5a3jq2m

Deep-Reinforcement-Learning-Based Semantic Navigation of Mobile Robots in Dynamic Environments [article]

Linh Kästner, Cornelius Marx, Jens Lambrecht
2020 arXiv   pre-print
On this account, we propose a reinforcement learning based local navigation system which learns navigation behavior based solely on visual observations to cope with highly dynamic environments.  ...  Furthermore, for safety reasons, they often rely on hand-crafted safety guidelines, which makes the system less flexible and slow.  ...  Training Algorithm The basic training loop is based on the suggestions from [17] and employs deep Q-learning.  ... 
arXiv:2008.00516v1 fatcat:gnemrlpn4zckriarj4z6bnn7zi

A Novel Sample-efficient Deep Reinforcement Learning with Episodic Policy Transfer for PID-Based Control in Cardiac Catheterization Robots [article]

Olatunji Mumini Omisore, Toluwanimi Akinyemi, Wenke Duan, Wenjing Du, Lei Wang
2021 arXiv   pre-print
In this study, a sample-efficient deep reinforcement learning with episodic policy transfer is, for the first time, used for motion control during robotic catheterization with fully adaptive PID tuning  ...  Robotic catheterization is typically used for percutaneous coronary intervention procedures nowadays and it involves steering flexible endovascular tools to open up occlusion in the coronaries.  ...  The deep Q-network learns from a state-action space using source and target networks trained on experience replay trained on a vector of given state, action output, next state, and network parameters,  ... 
arXiv:2110.14941v1 fatcat:incjnx6x3vevxfujn6z2daxwdy

Position Control of Cable-Driven Robotic Soft Arm Based on Deep Reinforcement Learning

Qiuxuan Wu, Yueqin Gu, Yancheng Li, Botao Zhang, Sergey A. Chepinskiy, Jian Wang, Anton A. Zhilenkov, Aleksandr Y. Krasnov, Sergei Chernyi
2020 Information  
deep Q learning.  ...  In order to solve slow convergence and unstable effect in the process of simulation and migration when deep reinforcement learning is applied to the actual robot control task, a control strategy learning  ...  The above research focuses on the control method of the model-based soft robot arm, datadriven modeling method and reinforcement learning method of the flexible manipulator.  ... 
doi:10.3390/info11060310 fatcat:ottk7kpaare4nhhe3biuylddla

Correction: Bhagat, S.; et al. Deep Reinforcement Learning for Soft, Flexible Robots: Brief Review with Impending Challenges. Robotics 2019, 8, 4

Sarthak Bhagat, Hritwick Banerjee, Zion Tsz Ho Tse, Hongliang Ren
2019 Robotics  
The manuscript will be updated and the original will remain online on the article webpage, with a reference to this Correction.  ...  reinforcement learning (DRL) and imitation learning techniques presently occur.  ...  In Figure 7 of this paper [1] , the caption was revised with the permission from the publishers as "Soft Robot Simulation on SOFA using Soft-robotics toolkit.  ... 
doi:10.3390/robotics8040093 fatcat:q6eojc2psbatznl3rp4v5cz4ze

Deep Reinforcement Learning for Formation Control

Can Aykin, Martin Knopp, Klaus Diepold
2018 2018 27th IEEE International Symposium on Robot and Human Interactive Communication (RO-MAN)  
Continuing our work on using reinforcement learning for formation control, we present an end-to-end deep learning system which uses only camera images to learn to control the individual system's correct  ...  This published work inspired us to employ a similar approach for processing the camera images and controlling the robot. We repeat the same experiment with two completely different camera positions.  ...  Inspired by the work of Mnih et al. ([4] , [5] ), we present an end-to-end deep reinforcement learning approach for formation control that does not depend on specific distance sensors, but uses conventional  ... 
doi:10.1109/roman.2018.8525765 dblp:conf/ro-man/AykinKD18 fatcat:sam5kytiefbyncvusjpr3hp2te

Deep Reinforcement Learning for Soft, Flexible Robots: Brief Review with Impending Challenges

Sarthak Bhagat, Hritwick Banerjee, Zion Ho Tse, Hongliang Ren
2019 Robotics  
For soft robotic structures possessing countless degrees of freedom, it is at times not convenient to formulate mathematical models necessary for training a deep reinforcement learning (DRL) agent.  ...  Deploying current imitation learning algorithms on soft robotic systems has provided competent results.  ...  The learning from experts' actions (generally a person for soft robotic manipulation problems) can sometimes lead to less-optimal solutions while using deep neural networks to train reinforcement learning  ... 
doi:10.3390/robotics8010004 fatcat:bxlbnniiwrdbhnf3vghkhiew3q

Learning Mobile Manipulation through Deep Reinforcement Learning

Cong Wang, Qifeng Zhang, Qiyan Tian, Shuo Li, Xiaohui Wang, David Lane, Yvan Petillot, Sen Wang
2020 Sensors  
This paper investigates how to leverage deep reinforcement learning to tackle whole-body mobile manipulation tasks in unstructured environments using only on-board sensors.  ...  It has an efficient framework decoupling visual perception from the deep reinforcement learning control, which enables its generalization from simulation training to real-world testing.  ...  ; • picking the object up using the policy generated by deep reinforcement learning.  ... 
doi:10.3390/s20030939 pmid:32050678 pmcid:PMC7039391 fatcat:yg3pdjoosrathioxsg3hbywp44

Composable Action-Conditioned Predictors: Flexible Off-Policy Learning for Robot Navigation [article]

Gregory Kahn, Adam Villaflor, Pieter Abbeel, Sergey Levine
2018 arXiv   pre-print
However, standard reinforcement learning approaches learn separate task-specific policies and assume the reward function for each task is known a priori.  ...  We show that a simulated robotic car and a real-world RC car can gather data and train fully autonomously without any human-provided labels beyond those needed to train the detectors, and then at test-time  ...  We propose a generalization of the reinforcement learning framework that combines flexible multitask learning, off-policy training, and the ability to learn directly from real-world events that can be  ... 
arXiv:1810.07167v1 fatcat:3mjsz4z2tjbmzhyjd2repvnllq

Applications and Challenges of Deep Reinforcement Learning in Multi-robot Path Planning

Tianyun Qiu, Yaxuan Cheng
2021 Journal of Electronic Research and Application  
Path planning in a collision-free environment is essential for many robots to do tasks quickly and efficiently, and path planning for multiple robots using deep reinforcement learning is a new research  ...  With the rapid advancement of deep reinforcement learning (DRL) in multi-agent systems, a variety of practical application challenges and solutions in the direction of multi-agent deep reinforcement learning  ...  Combining model-based algorithms and model-free deep reinforcement learning path planning problems is one of the key ways to improve the efficiency of reinforcement learning in the future.  ... 
doi:10.26689/jera.v5i6.2809 fatcat:ohkwlmyzlrdihpzxpwbufke6ni

Towards Vision-Based Deep Reinforcement Learning for Robotic Motion Control [article]

Fangyi Zhang, Jürgen Leitner, Michael Milford, Ben Upcroft, Peter Corke
2015 arXiv   pre-print
We build upon the success of recent deep reinforcement learning and develop a system for learning target reaching with a three-joint robot manipulator using external visual observation.  ...  A Deep Q Network (DQN) was demonstrated to perform target reaching after training in simulation.  ...  Acknowledgements This research was conducted by the Australian Research Council Centre of Excellence for Robotic Vision (project number CE140100016).  ... 
arXiv:1511.03791v2 fatcat:fiuwfb6dmnerrnzm7pmr4jkn2u

Transfer Learning with Demonstration Forgetting for Robotic Manipulator

Ermek Aitygulov, Aleksandr I. Panov
2021 Procedia Computer Science  
Deep learning and especially deep reinforcement learning usually require huge amount of data for training and simulators using is perspective approach to provide this data.  ...  Abstract Deep learning and especially deep reinforcement learning usually require huge amount of data for training and simulators using is perspective approach to provide this data.  ...  Deep Reinforcement Learning (DRL) is perspective approach for robotic control because it has shown good results in different control problems [1, 2, 3] .  ... 
doi:10.1016/j.procs.2021.04.159 fatcat:ga457ocfizfijldoz3siedyatm

Motion planning in human robot cooperation via deep reinforcement learning

Giorgio Nicola, Nicola Pedrocchi, Stefano Ghidoni
2019 Zenodo  
An approach to motion planning for human robot cooperation based on Deep Reinforcement Learning in simulated environments is proposed.  ...  This approach aims at solving some of the typical problems of motion planning in human robot cooperation such as the need of inferring the human movements or the need of continuous re-planning trajectories  ...  Currently, Deep Reinforcement Learning (DRL) is a hot topic in the field of machine learning and also with a number of applications in robotics.  ... 
doi:10.5281/zenodo.4795818 fatcat:ckdv2rsfxvbpvkrmnty6t7ygpe

Malleable Agents for Re-Configurable Robotic Manipulators [article]

Athindran Ramesh Kumar, Gurudutt Hosangadi
2022 arXiv   pre-print
While deep reinforcement learning has had immense success in robotic manipulation, domain adaptation is a significant problem that limits its applicability to real-world robotics.  ...  Re-configurable robots potentially have more utility and flexibility for many real-world tasks. Designing a learning agent to operate such robots requires adapting to different configurations.  ...  Deep reinforcement learning for robotic manipulation Robotic arm manipulation tasks such as reach, pick-andplace, grasping have become a standard benchmark [27] for testing deep reinforcement learning  ... 
arXiv:2202.02395v1 fatcat:5imnqodlh5gjpmtq74kmy2dib4

Contact-Rich Manipulation of a Flexible Object based on Deep Predictive Learning using Vision and Tactility [article]

Hideyuki Ichiwara, Hiroshi Ito, Kenjiro Yamamoto, Hiroki Mori, Tetsuya Ogata
2022 arXiv   pre-print
The results of experiments using a real robot arm revealed that our method can realize motions responding to the deformation of the bag while reducing the load on the zipper.  ...  Additionally, the flexible fabric bag state constantly changes during operation, so the robot needs to dynamically respond to the change.  ...  ACKNOWLEDGEMENT AI Bridging Cloud Infrastructure of National Institute of Advanced Industrial Science and Technology was used.  ... 
arXiv:2112.06442v2 fatcat:3vhzx5ftk5duxidfceioa3hjtu
« Previous Showing results 1 — 15 out of 12,995 results