T
Tomoaki Oiki
Researcher at Mitsubishi Electric
Publications - 7
Citations - 63
Tomoaki Oiki is an academic researcher from Mitsubishi Electric. The author has contributed to research in topics: Reinforcement learning & Physics engine. The author has an hindex of 4, co-authored 7 publications receiving 30 citations.
Papers
More filters
Proceedings ArticleDOI
Trajectory Optimization for Unknown Constrained Systems using Reinforcement Learning
Kei Ota,Devesh K. Jha,Tomoaki Oiki,Miura Mamoru,Takashi Nammoto,Daniel Nikovski,Toshisada Mariyama +6 more
TL;DR: A reinforcement learning-based algorithm for trajectory optimization for constrained dynamical systems, trained with a reference path and parameterize the policies with goal locations, so that the agent can be trained for multiple goals simultaneously.
Proceedings Article
Can Increasing Input Dimensionality Improve Deep Reinforcement Learning
TL;DR: In this paper, an online feature extractor network (OFENet) is proposed to produce good representations to be used as inputs to deep RL algorithms, and the authors show that the RL agents learn more efficiently with the high-dimensional representation than with the lower-dimensional state observations.
Journal ArticleDOI
Data-Efficient Learning for Complex and Real-Time Physical Problem Solving Using Augmented Simulation
Kei Ota,Devesh K. Jha,Diego Romeres,Jeroen van Baar,Kevin A. Smith,Takayuki Semitsu,Tomoaki Oiki,Alan Sullivan,Daniel Nikovski,Joshua B. Tenenbaum +9 more
TL;DR: In this article, the authors presented a model that learns to move a marble in a complex environment within minutes of interacting with the real system, using a hybrid model consisting of a full physics engine along with a statistical function approximator.
Posted Content
Trajectory Optimization for Unknown Constrained Systems using Reinforcement Learning.
Kei Ota,Devesh K. Jha,Tomoaki Oiki,Miura Mamoru,Takashi Nammoto,Daniel Nikovski,Toshisada Mariyama +6 more
TL;DR: In this paper, a reinforcement learning-based algorithm for trajectory optimization for constrained dynamical systems is proposed, which is motivated by the fact that for most robotic systems, the dynamics may not always be known.
Posted Content
Towards Human-Level Learning of Complex Physical Puzzles.
Kei Ota,Devesh K. Jha,Diego Romeres,Jeroen van Baar,Kevin A. Smith,Takayuki Semitsu,Tomoaki Oiki,Alan Sullivan,Daniel Nikovski,Joshua B. Tenenbaum +9 more
TL;DR: This paper presents a model that learns to move a marble in the complex environment within minutes of interacting with the real system, and is the first time that a hybrid model consisting of a full physics engine along with a statistical function approximator has been used to control a complex physical system in real-time using nonlinear model-predictive control (NMPC).