| Project/Area Number |
11680405
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Intelligent informatics
|
|---|
| Research Institution | Kinki University |
|---|
Principal Investigator |
IGARASHI Harukazu Kinki University School of Engineering,Associate Professor, 工学部, 助教授 (80288886)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
IOI Kiyoshi Kinki University School of Engineering, Associate Professor, 工学部, 助教授 (90288887)
KUROSE Yoshinobu Kinki University School of Engineering, Professor, 工学部, 教授 (00043802)
|
|---|
| Project Period (FY) |
1999 – 2001
|
| Project Status |
Completed (Fiscal Year 2001)
|
| Budget Amount *help |
¥3,800,000 (Direct Cost: ¥3,800,000)
Fiscal Year 2001: ¥500,000 (Direct Cost: ¥500,000)
Fiscal Year 2000: ¥600,000 (Direct Cost: ¥600,000)
Fiscal Year 1999: ¥2,700,000 (Direct Cost: ¥2,700,000)
|
|---|
| Keywords | Autonomous Mobile Robots / Reinforcement Learning / Multiagent / RoboCup / Global Vision / Remote Brain / Path Planning / マルチユージェント / Robo Cup / 画像歪み / RoBoCup |
|---|
| Research Abstract |
This report describes research results of our JP/S-II Projecct that has been developed at Kinki University since 1997 and was supported by Grant-in-Aid for Scientific Research ? of Japan Society for the Promotion of Science from 1999 to 2001.Our group is the second group to propose a JP/S (Joint Platform for RoboCup Small Size League) project in Japan. The JP/S-II system has the following four design features. First, robots are remotely controlled by control programs (Remote Brains). Each remote brain gets visual information from the global-view camera, analyses the situation and determines its next behavior. Each brain uses a set of virtual roads mapped on the field to search for the shortest path to the goal where it wants to go. Second, the remote brains organize a multi-agent system. Each agent can collaborate with the others to make a team play. Third, the JP/S-II system has a simulator that can simulate the type of game played in the RoboCup Small Size Robot League. Fourth, robots can be made at very low cost.
Moreover, we proposed a new method for motion planning of mobile robots, In our approach, we formulated the problem as a discrete optimization problem at each time step. To solve the optimization problem, we used an objective function consisting of a goal term, a smoothness term and a collision term. The weight parameters in the objective function are adjusted by reinforcement learning. We applied Williams's learning algorithm, episodic REINFORCE, to derive a learning rule for the weight parameters. The learning rule was verified by our some simulations. This approach can be applied to navigation of real mobile robots as well as motion planning simulation.
|
