2006
DOI: 10.1299/kikaic.72.3525
|View full text |Cite
|
Sign up to set email alerts
|

Study on Obtained Advance Motion Forms of a Caterpillar Robot by using Reinforcement Learning

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1

Citation Types

0
3
0

Year Published

2006
2006
2011
2011

Publication Types

Select...
3
1

Relationship

2
2

Authors

Journals

citations
Cited by 4 publications
(3 citation statements)
references
References 0 publications
0
3
0
Order By: Relevance
“…In addition, the discount rate significantly affects the form of the maximum reward group. In references [7] and [8], our previous study mentioned that it is easy to produce the final motion form with a very few motion patterns when the discount rate y is set to a large value; vice versa, an inverse effect was confirmed when y is set to a smaller value. It is assumed that these functions often generate various motion forms during the learning process.…”
Section: B Emergence Ofmotion Formsmentioning
confidence: 95%
See 1 more Smart Citation
“…In addition, the discount rate significantly affects the form of the maximum reward group. In references [7] and [8], our previous study mentioned that it is easy to produce the final motion form with a very few motion patterns when the discount rate y is set to a large value; vice versa, an inverse effect was confirmed when y is set to a smaller value. It is assumed that these functions often generate various motion forms during the learning process.…”
Section: B Emergence Ofmotion Formsmentioning
confidence: 95%
“…Further, Kimura and his colleagues demonstrated that a robot can acquire advance actions by the application of reinforcement learning [6]. In our previous works, we have studied the advance actions of a caterpillar robot using reinforcement learning [7] [8]. The results demonstrated that the reinforcement learning enables the robot to achieve an unexpected motion pattern and exhibit good performance for a given task; in this paper, the motion pattern refers to the robot state during one step.…”
Section: Introductionmentioning
confidence: 97%
“…This is a big feature of unsupervised www.intechopen.com (1) and (2), it should be noted that the discount rate γ significantly affects the transition of motion forms. In fact, our previous studies demonstrated that it is easier to produce the optimal motion form with a very few actions when γ is configured at a large value; vice versa, an inverse result is observed when γ is a smaller value (Yamashina et al, 2006;Motoyama et al, 2006). Hence, it is assumed that the discount rate is a significant factor to generate the series of motion forms in the learning process.…”
Section: Transition Of Motion Forms During Q-learningmentioning
confidence: 99%