Presentation 2009-05-25
Reinforcement learning through the active low-dimensional sensory information
Mikihiro KOBAYASHI, Eiji UCHIBE, Kenji DOYA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) The goal of this research is to reduce the dimension of the state spaces in the framework of reinforcement learning using visual servoing. The adaptive visual servoing minimizes the distance between the current and desired image measurements by controlling joint angles of the camera-head mounted on the robot. The final joint angles of camera-head are regarded as the position of the robot in the environment. Therefore, the state space can be constructed from the converged joint angles of the camera head. On the other hand, the role of reinforcement learning is to drive the robot to the target object in the environment. Experimental results show that the efficacy of the proposed method with real robot.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) reinforcement learning / visual servoing / autonomous agent / internal representation / low-dimensional
Paper # NC2009-4
Date of Issue

Conference Information
Committee NC
Conference Date 2009/5/18(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Neurocomputing (NC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Reinforcement learning through the active low-dimensional sensory information
Sub Title (in English)
Keyword(1) reinforcement learning
Keyword(2) visual servoing
Keyword(3) autonomous agent
Keyword(4) internal representation
Keyword(5) low-dimensional
1st Author's Name Mikihiro KOBAYASHI
1st Author's Affiliation Nara Institute of Science and Technology()
2nd Author's Name Eiji UCHIBE
2nd Author's Affiliation Okinawa Institute of Science and Technology Promotion Corporation
3rd Author's Name Kenji DOYA
3rd Author's Affiliation Okinawa Institute of Science and Technology Promotion Corporation
Date 2009-05-25
Paper # NC2009-4
Volume (vol) vol.109
Number (no) 53
Page pp.pp.-
#Pages 6
Date of Issue