Filters








8 Hits in 3.5 sec

A Real-Time Model-Based Reinforcement Learning Architecture for Robot Control [article]

Todd Hester, Michael Quinlan, Peter Stone
2011 arXiv   pre-print
In this paper, we present a novel parallel architecture for model-based RL that runs in real-time by 1) taking advantage of sample-based approximate planning methods and 2) parallelizing the acting, model  ...  Reinforcement Learning (RL) is a method for learning decision-making tasks that could enable robots to learn and adapt to their situation on-line.  ...  ACKNOWLEDGMENTS This work has taken place in the Learning Agents Research Group (LARG) at the Artificial Intelligence Laboratory, The University of Texas at Austin.  ... 
arXiv:1105.1749v2 fatcat:7j7dnqntt5ac5p7oej5vk5obxa

RTMBA: A Real-Time Model-Based Reinforcement Learning Architecture for robot control

Todd Hester, Michael Quinlan, Peter Stone
2012 2012 IEEE International Conference on Robotics and Automation  
In this paper, we present a novel parallel architecture for model-based RL that runs in real-time by 1) taking advantage of sample-based approximate planning methods and 2) parallelizing the acting, model  ...  For an RL algorithm to be practical for robotic control tasks, it must learn in very few samples, while continually taking actions in real-time.  ...  Fig. 2 . 2 A diagram of the Real-Time Model Based Architecture (RTMBA) Average reward versus clock time.  ... 
doi:10.1109/icra.2012.6225072 dblp:conf/icra/HesterQS12 fatcat:ifxetlh4qnaz5h6n5degashuxy

The Open-Source TEXPLORE Code Release for Reinforcement Learning on Robots [chapter]

Todd Hester, Peter Stone
2014 Lecture Notes in Computer Science  
For an RL algorithm to be practical for robotic control tasks, it must learn in very few samples, while continually taking actions in real-time.  ...  We demonstrate texplore learning to control the velocity of an autonomous vehicle in real-time. texplore has been released as an open-source ROS repository, enabling learning on a variety of robot tasks  ...  Acknowledgements This work has taken place in the Learning Agents Research Group (LARG) at UT Austin.  ... 
doi:10.1007/978-3-662-44468-9_47 fatcat:75j3f6puhjbgdpbugarlmd2fim

TEXPLORE: real-time sample-efficient reinforcement learning for robots

Todd Hester, Peter Stone
2012 Machine Learning  
For an RL algorithm to be practical for robotic control tasks, it must learn in very few samples, while continually taking actions in real-time.  ...  With sample-based planning and a novel parallel architecture, TEXPLORE can select actions continually in real-time whenever necessary.  ...  Parallel Architecture In addition to using MCTS for planning, we have developed a multi-threaded architecture, called the Real-Time Model Based Architecture (RTMBA), for the agent to learn while acting  ... 
doi:10.1007/s10994-012-5322-7 fatcat:5bac6cigxbgnzg6krsrbq2ywba

Continuous reinforcement learning with incremental Gaussian mixture models

Rafael Coimbra Pinto
2019 Figshare  
Then, this same function approximator was employed to model the joint state and Q-values space, all in a single FIGMN, resulting in a concise and data-efficient algorithm, i.e., a reinforcement learning  ...  This algorithm, called Fast Incremental Gaussian Mixture Network (FIGMN), was employed as a sample-efficient function approximator for the state space of continuous reinforcement learning tasks, which,  ...  RTMBA Real-Time Model-Based Architecture (RTMBA) [Hester, Quinlan and Stone 2012] aims to make model-based reinforcement learning practical for real-time applications like robotics.  ... 
doi:10.6084/m9.figshare.9942521 fatcat:jly4alfvtjgy7ktm7pheayvwni

Intrinsically motivated model learning for developing curious robots

Todd Hester, Peter Stone
2017 Artificial Intelligence  
Reinforcement Learning (RL) agents are typically deployed to learn a specific, concrete task based on a pre-defined reward function.  ...  Novel methods for obtaining intrinsic rewards from a random-forest-based model of the world. 2.  ...  In order to be applicable to robots, texplore-vanir uses the Real-Time Model Based Architecture [24] .  ... 
doi:10.1016/j.artint.2015.05.002 fatcat:uthcqhezpzgdpnlpflpizugqrq

Intrinsically motivated model learning for a developing curious agent

Todd Hester, Peter Stone
2012 2012 IEEE International Conference on Development and Learning and Epigenetic Robotics (ICDL)  
Reinforcement Learning (RL) agents are typically deployed to learn a specific, concrete task based on a pre-defined reward function.  ...  The algorithm learns models of the transition dynamics of a domain using random forests.  ...  In order to be applicable to robots, TEXPLORE-VANIR uses the Real-Time Model Based Architecture [4] .  ... 
doi:10.1109/devlrn.2012.6400802 dblp:conf/icdl-epirob/HesterS12 fatcat:jke4h6mpsrgl3hrmp42bqukp4i

Flexible Heuristic Dynamic Programming for Reinforcement Learning in Quad-Rotors

Alexander Helmer, Coen C. de Visser, Erik-Jan Van Kampen
2018 2018 AIAA Information Systems-AIAA Infotech @ Aerospace   unpublished
Figure 2 . 8 : 28 Algorithm for the Real-Time Model-Based Architecture. Taken from[27].  ...  In [27] , a Real-Time Model Based Architecture (RTMBA) for RL is proposed that parallelizes the modellearning, planning and action within an agent into three separate threads that can be run on separate  ...  cr do set trace of nearest neighbors to 1 17: z t (p) ← 1 18: for p in M cr do update values of critic nearest neighbors 19 : add [s t , a t |s t +1 ] to M pm add sample to plant model memory 21  ... 
doi:10.2514/6.2018-2134 fatcat:o6ttgzc5drcoxnu7owmbhdhulq