|
|
| RIL Toolbox Examples |
|
The Pole-Balancing
Task |
|
|
| The Task is to
balance a Pole hinged to a cart. The Example was taken from
R.Sutton's Book. This is a tutorial for creating your own
environment and using different learning algorithm. We use a
discrete Q-Value Learning, discrete Actor Critic and a Continuous
Q-Learning algorithm and compare their
performance.
|
|
|
|
|
The Taxi
Task |
|
|
| Tutorial for the
hierarchical architecture of the
toolbox.
|
|
| Extern Projects |
|
Learning
robot behavior |
| author: |
Måns Ullerstam,
The Mizukawa Laboratory, Shibaura Institute of Technology, Tokyo,
Japan.
|
|
| Project used as
Master Thesis. Using the RIL Toolbox to learn robot behavior
patterns. Implemented in a real environment with a Sony AIBO
ERS-210 with Wireless LAN. Learning system executes on a computer
sending actions to the robot. The environment is a collection of
the sensor readings from the robot, sent back to the computer.
Actions are high level such as move forward, turn left, sit, look
up, wag tail.
|
|
|
|
|
