93 Hits in 21.3 sec

IRIS: Implicit Reinforcement without Interaction at Scale for Learning Control from Offline Robot Manipulation Data [article]

Ajay Mandlekar, Fabio Ramos, Byron Boots, Silvio Savarese, Li Fei-Fei, Animesh Garg, Dieter Fox
2020 arXiv   pre-print
In this paper, we propose Implicit Reinforcement without Interaction at Scale (IRIS), a novel framework for learning from large-scale demonstration datasets.  ...  Learning from offline task demonstrations is a problem of great interest in robotics.  ...  We thank Kevin Shih for help with training models on image observations. We thank members of the NVIDIA Seattle Robotics Lab for several helpful discussions and feedback.  ... 
arXiv:1911.05321v2 fatcat:q6kp2tjdrreu3oefw5dyczulj4

What Matters in Learning from Offline Human Demonstrations for Robot Manipulation [article]

Ajay Mandlekar, Danfei Xu, Josiah Wong, Soroush Nasiriany, Chen Wang, Rohun Kulkarni, Li Fei-Fei, Silvio Savarese, Yuke Zhu, Roberto Martín-Martín
2021 arXiv   pre-print
Our study analyzes the most critical challenges when learning from offline human data for manipulation.  ...  In this paper, we conduct an extensive study of six offline learning algorithms for robot manipulation on five simulated and three real-world multi-stage manipulation tasks of varying complexity, and with  ...  Acknowledgments We would like to thank Albert Tung for helping with the RoboTurk data collection system, Jim Fan for providing timely lab cluster support, and Helen Roman for helping order items for the  ... 
arXiv:2108.03298v2 fatcat:p7fmn5qjynftbjaymffoifpf5y

Learning to Generalize Across Long-Horizon Tasks from Human Demonstrations [article]

Ajay Mandlekar, Danfei Xu, Roberto Martín-Martín, Silvio Savarese, Li Fei-Fei
2021 arXiv   pre-print
We present a novel imitation learning framework to enable robots to 1) learn complex real world manipulation tasks efficiently from a small number of human demonstrations, and 2) synthesize new behaviors  ...  We validate GTI in both simulated domains and a challenging long-horizon robotic manipulation domain in the real world.  ...  For example, if two trajectories τ 1 and τ 2 intersect at a state s, then there are implicit paths from s 1 0 to g 2 and s 2 0 to g 1 in the demonstration data -even though a demonstrator never explicitly  ... 
arXiv:2003.06085v2 fatcat:tr3u3amg7za23pkczo3c3xshnq

Guided Imitation of Task and Motion Planning [article]

Michael James McDonald, Dylan Hadfield-Menell
2021 arXiv   pre-print
In robotic manipulation tasks with 7-DoF joint control, the partially trained policies reduce the time needed for planning by a factor of up to 2.6.  ...  First, we build an asynchronous distributed TAMP solver that can produce supervision data fast enough for imitation learning.  ...  Iris: Implicit reinforce- ment without interaction at scale for learning control from offline robot manipulation data. 2020 IEEE International Conference on Robotics and Automation (ICRA), pages  ... 
arXiv:2112.03386v1 fatcat:liiwhjl67fao7phu2tuj4bj7yy

Learning Value Functions from Undirected State-only Experience [article]

Matthew Chang, Arjun Gupta, Saurabh Gupta
2022 arXiv   pre-print
This paper tackles the problem of learning value functions from undirected state-only experience (state transitions without action labels i.e. (s,s',r) tuples).  ...  This theoretical result motivates the design of Latent Action Q-learning or LAQ, an offline RL method that can learn effective value functions from state-only experience.  ...  INTRODUCTION Offline or batch reinforcement learning focuses on learning goal-directed behavior from prerecorded data of undirected experience in the form of (s t , a t , s t+1 , r t ) quadruples.  ... 
arXiv:2204.12458v1 fatcat:yxdapx6osrcp3gtyftbq625k4q

A Metaverse: taxonomy, components, applications, and open challenges

Sang-Min Park, Young-Gab Kim
2022 IEEE Access  
With the technological development of deep learning-based high-precision recognition models and natural generation models, Metaverse is being strengthened with various factors, from mobile-based always-on  ...  This paper divides the concepts and essential techniques necessary for realizing the Metaverse into three components (i.e., hardware, software, and contents) and three approaches (i.e., user interaction  ...  [328] studied how expressive learning can accelerate reinforcement learning from rich observations (e.g., images) without relying on domain knowledge.  ... 
doi:10.1109/access.2021.3140175 fatcat:fnraeaz74vh33knfvhzrynesli

COMBO: Conservative Offline Model-Based Policy Optimization [article]

Tianhe Yu, Aviral Kumar, Rafael Rafailov, Aravind Rajeswaran, Sergey Levine, Chelsea Finn
2022 arXiv   pre-print
Model-based algorithms, which learn a dynamics model from logged experience and perform some sort of pessimistic planning under the learned model, have emerged as a promising paradigm for offline reinforcement  ...  learning (offline RL).  ...  Acknowledgments and Disclosure of Funding We thank members of RAIL and IRIS for their support and feedback.  ... 
arXiv:2102.08363v2 fatcat:azvca4wb65gc5aypjpidqphgzi

SIENNA D4.4: Ethical Analysis of AI and Robotics Technologies

Philip Jansen, Philip Brey, Alice Fox, Jonne Maas, Bradley Hillas, Nils Wagner, Patrick Smith, Isaac Oluoch, Laura Lamers, Hero Van Gein, Anaïs Resseguier, Rowena Rodrigues (+2 others)
2020 Zenodo  
(2) their physical technological products and procedures that are designed for practical applications; and (3) the particular uses and applications of these products and procedures.  ...  This SIENNA deliverable offers a broad ethical analysis of artificial intelligence (AI) and robotics technologies.  ...  As machine learning systems develop and refine their algorithms by analysing training data, they are susceptible to reinforcing any implicit or explicit biases contained within that data.  ... 
doi:10.5281/zenodo.4068082 fatcat:xiucqv6opng6rbit25lyfemzm4

On the Opportunities and Risks of Foundation Models [article]

Rishi Bommasani, Drew A. Hudson, Ehsan Adeli, Russ Altman, Simran Arora, Sydney von Arx, Michael S. Bernstein, Jeannette Bohg, Antoine Bosselut, Emma Brunskill, Erik Brynjolfsson, Shyamal Buch (+102 others)
2021 arXiv   pre-print
., BERT, DALL-E, GPT-3) that are trained on broad data at scale and are adaptable to a wide range of downstream tasks.  ...  This report provides a thorough account of the opportunities and risks of foundation models, ranging from their capabilities (e.g., language, vision, robotics, reasoning, human interaction) and technical  ...  In addition, we would like to especially thank Vanessa Parli for helping to organize this effort.  ... 
arXiv:2108.07258v2 fatcat:yktkv4diyrgzzfzqlpvaiabc2m

Robotics and the New Cyberlaw

Ryan Calo
2014 Social Science Research Network  
Robotics combines, for the first time, the promiscuity of data with the capacity to do physical harm; robotic systems accomplish tasks in ways that cannot be anticipated in advance; and robots increasingly  ...  The same public and private institutions that developed the Internet, from the armed forces to search engines, have initiated a significant shift toward developing robotics and artificial intelligence.  ...  Finally, the Internet allows for additional or at least more exquisite forms of observation and manipulation than offline analogs (control). 35 The architecture of networks and interfaces is subject  ... 
doi:10.2139/ssrn.2402972 fatcat:oeptn4tqvvedfnn3kh6t23iwmm

Phasic norepinephrine is a neural interrupt signal for unexpected events in rapidly unfolding sensory sequences – evidence from pupillometry [article]

Sijia Zhao, Maria Chait, Fred Dick, Peter Dayan, Shigeto Furukawa, Hsin-I Liao
2018 biorxiv/medrxiv   pre-print
A prominent theory (Dayan & Yu, 2006) proposes that the brain monitors for 'unexpected uncertainty' – events which deviate substantially from model predictions, indicating model failure.  ...  The results demonstrate that NE tracks 'unexpected uncertainty' on rapid time scales relevant to sensory signals.  ...  Overall, these data provide strong evidence that both conceptual and computational models of reinforcement learning should be revised to incorporate a more prominent role for the amygdala and its interaction  ... 
doi:10.1101/466367 fatcat:a3bquw6n55amhodwmfoa2frboa

Governance of Autonomous Agents on the Web: Challenges and Opportunities [article]

Timotheus Kampik, Adnane Mansour, Olivier Boissier, Sabrina Kirrane, Julian Padget, Terry R. Payne, Munindar P. Singh, Valentina Tamma, Antoine Zimmermann
2022 arXiv   pre-print
The study of autonomous agents has a long tradition in the Multiagent Systems and the Semantic Web communities, with applications ranging from automating business processes to personal assistants.  ...  Towards this end, this paper motivates the need for alignment and joint research across the Multiagent Systems, Semantic Web, and WoT communities, introduces a conceptual framework for governance of autonomous  ...  Singh acknowledges support from the US National Science Foundation under grant IIS-1908374.  ... 
arXiv:2202.02574v1 fatcat:oto6ib442zawta3bneuqwayrhy

Defense Advanced Research Projects Agency (Darpa) Fiscal Year 2015 Budget Estimates

Department Of Defense Comptroller's Office
2014 Zenodo  
DARPA seeks to improve the analysis of large neural data sets by creating interfaces that will allow researchers to generate new models across multiple scales.  ...  Emerging Therapies (SUBNETS)) and better data management.  ...  robotic technologies that will enable autonomous (unmanned) mobile platforms to manipulate objects without human control or intervention.  ... 
doi:10.5281/zenodo.1215345 fatcat:fjzhmynqjbaafk67q2ckcblj2m

Defense Advanced Research Projects Agency (Darpa) Fiscal Year 2016 Budget Estimates

Department Of Defense Comptroller's Office
2015 Zenodo  
FY 2015 2015 Plans: -Develop methods for interactive, iterative, and distributed analysis of diverse data at petabyte scale.  ...  technologies that enabled autonomous (unmanned) mobile platforms to manipulate objects without human control or intervention.  ... 
doi:10.5281/zenodo.1215366 fatcat:cqn5tyfixjanzp5x3tgfkpedri

When I Look into Your Eyes: A Survey on Computer Vision Contributions for Human Gaze Estimation and Tracking

Dario Cazzato, Marco Leo, Cosimo Distante, Holger Voos
2020 Sensors  
estimating gaze target from different perspectives: from the eye of the beholder (first-person view), from an external camera framing the beholder's, from a third-person view looking at the scene where  ...  revolutionized the whole machine learning area, and gaze tracking as well.  ...  problems, including classification, regression, and reinforcement learning.  ... 
doi:10.3390/s20133739 pmid:32635375 pmcid:PMC7374327 fatcat:jwou6gv4f5dy7lrsxvtbnb2fly
« Previous Showing results 1 — 15 out of 93 results