MARCO videos
- Details
- Last Updated: 28 April 2019
- Published: 11 June 2011
- Written by Administrator
- Hits: 4926
"MARCO @ UFreiburg":
In 2014 we started to build "MARCO", the Multimodal Autonomous RoboChess Opponent". Below you find related videos with newest first. Have fun!
We demonstrated MARCO at ICMI2014 in Istanbul, Turkey, and won the Outstanding Demo Award. (Watch this video on youtube, if you like.)
In that configuration we realized:
- a few display movements
- emotional facial expressions driven by WASABI based on the evaluation of the board
- detection of and recovering from "invalid moves" by the human player (at least to some extend)
Next steps / improvements:
- change the display for a bigger one
- implement many more agent behaviors based on a newly invented animation framework
- finish integration of NOVA framework
- possibly integrate the IntraFace tracking software
- devise and conduct empirical studies
My colleague Julien Hué played against the prototype version of MARCO on July 3rd, 2014.
What's still missing:
- The agent on the display in the back is not moving and not reacting (emotionally) to what is happening. But it's "alive" as you can tell by its eyes blinking (thanks to Matthieu Courgeon).
- Human players have to always pick and place pieces on the DGT board, because when pushing pieces over the board, the system receives too many signals. We are working on that part.
- The clock is not in use, but in the end both players should use the clock for Blitz-chess.
An early version of Marco demonstrated to young visitors in May 2014. In the beginning, we use a custom built android app to operate the display rotation. With explanations in German..
Previous version of MARCO:
It can also be controlled from a remote place using the freechess.org server and any client. Thus, someone can play against me using the robot arm to move his or her pieces here in Freiburg.
