Labyrinth 2 is the follow up from the Labyrinth 1 project. A BRIO labyrinth game was equipped with two servo motors and two potentiometers that allow for the remote control and measurement of the orientation of the game’s board.
|Duration:||07.01.2008 till 31.01.2009|
|Donee:||German Research Center for Artificial Intelligence GmbH|
|Application Field:||Assistance- and Rehabilitation Systems|
Development of Learning Architectures and Experiments in Sensory Motor Learning (06.2007- 12.2007)
Testbed for the development of learning architectures
Maja Machine Learning Framework
A BRIO labyrinth game was equipped with two servo motors and two potentiometers that allow for the remote control and measurement of the orientation of the game’s board. The board is mounted on a platform, and a camera is placed above this platform which allows the current position of the ball on the board to be estimated. For this purpose, a vision algorithm hasbeen developed which segments the ball in the camera image and maps the position of the ball inthe image onto a position in the labyrinth coordinate system. The mapping is calibrated automatically at the begining of each session using landmarks. Furthermore, a ball depot is currently beingdeveloped that will be mounted on the game and will allow more than one session to be playedwithout having to put the ball back manually.
Moreover, we have developed a physical simulation of this game using the Open Dynamics Engine.The goal of studies in the field of machine learning is to investigate how for example Reinforcement Learning (RL) algorithms can be appliedto large-scale, real-world problems. These problems are challenging for standard RL algorithmssince they are inherently noisy, have continuous state and/or action spaces and are potentially onlypartially observable.
The BRIO labyrinth game is also used as a testbed for neurobiological studies ofsensorimotori integration, where human subjects are confronted with different tasks related to the BRIO labyrinth game (prediction, control, etc.). For example, electroencephalography (EEG) andfunctional magnetic resonance imaging (fMRI) studies will be carried out where human subjectslearn to play the game. This might provide some insights into the way the human brain automatesbehavior and in this way learns certain skills.
BRIO Labyrinth: Brain Reading (Demonstration)
In a telemanipulation-like scenario