Zobrazeno 1 - 10
of 96
pro vyhledávání: '"Karol Hausman"'
Publikováno v:
IEEE Robotics and Automation Letters. 6:4329-4336
One of the most challenging aspects of real-world reinforcement learning (RL) is the multitude of unpredictable and ever-changing distractions that could divert an agent from what was tasked to do in its training environment. While an agent could lea
Autor:
Zhanpeng He, Eric Heiden, Hejia Zhang, Joseph J. Lim, Ryan Julian, Gaurav S. Sukhatme, Stefan Schaal, Karol Hausman
Publikováno v:
The International Journal of Robotics Research. 39:1259-1278
We present a strategy for simulation-to-real transfer, which builds on recent advances in robot skill decomposition. Rather than focusing on minimizing the simulation–reality gap, we propose a method for increasing the sample efficiency and robustn
Publikováno v:
ISRR
Representing the environment is a fundamental task in enabling robots to act autonomously in unknown environments. In this work, we present confidence-rich mapping (CRM), a new algorithm for spatial grid-based mapping of the 3D environment. CRM augme
Publikováno v:
Robotics: Science and Systems
The distributional perspective on reinforcement learning (RL) has given rise to a series of successful Q-learning algorithms, resulting in state-of-the-art performance in arcade game environments. However, it has not yet been analyzed how these findi
Autor:
Joseph J. Lim, Ryan Julian, Stefan Schaal, Gaurav S. Sukhatme, Karol Hausman, Eric Heiden, Zhanpeng He, Hejia Zhang
Publikováno v:
Springer Proceedings in Advanced Robotics ISBN: 9783030339494
ISER
ISER
We present a novel solution to the problem of simulation-to-real transfer, which builds on recent advances in robot skill decomposition. Rather than focusing on minimizing the simulation-reality gap, we learn a set of diverse policies that are parame
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_________::71f68d8d2203d62490bf85e2d99c9e10
https://doi.org/10.1007/978-3-030-33950-0_24
https://doi.org/10.1007/978-3-030-33950-0_24
Publikováno v:
Robotics: Science and Systems
Reinforcement learning provides a general framework for learning robotic skills while minimizing engineering effort. However, most reinforcement learning algorithms assume that a well-designed reward function is provided, and learn a single behavior
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::9fa23c929f6347c2b55759f42840331b
Publikováno v:
Springer Proceedings in Advanced Robotics ISBN: 9783030286187
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_________::90b441c8ddcb23776a7f6aa85ab97970
https://doi.org/10.1007/978-3-030-28619-4_45
https://doi.org/10.1007/978-3-030-28619-4_45
Publikováno v:
IEEE Robotics and Automation Letters. 2:1770-1777
We study the nonlinear observability of a system's states in view of how well they are observable and what control inputs would improve the convergence of their estimates. We use these insights to develop an observability-aware trajectory-optimizatio
Publikováno v:
AAAI
When deploying autonomous agents in the real world, we need effective ways of communicating objectives to them. Traditional skill learning has revolved around reinforcement and imitation learning, each with rigid constraints on the format of informat
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::a47947d29bf59135ad962eaf145734d1
http://arxiv.org/abs/1906.10187
http://arxiv.org/abs/1906.10187
Publikováno v:
The International Journal of Robotics Research. 34:1660-1677
We consider the cooperative control of a team of robots to estimate the position of a moving target using onboard sensing. In this setting, robots are required to estimate their positions using relative onboard sensing while concurrently tracking the