Autonomous Construction in Unknown Environments with Scarce Resources

This is our (Stéphane Magnenat and Francesco Mondada) entry accepted in the AAAI-11 video competition.

This video presents a marXbot robot with a magnetic manipulator performing autonomous construction in an unknown environment with scarce resources, which requires understanding and reasoning about the environment.

At the beginning of the experiment, the robot knows neither the geometry nor the topology of the environment. It does not know where the resources are located. The robot does know that building a tower requires three resources, and that a remote resource can be fetched by building a bridge. In the video, the decision to build this specific bridge to fetch the remote resources was the robot's own one, done by reasoning using an HTN planner, whose source code is available at http://gitorious.org/planner9 .

More informations about this work are available in my PhD thesis at http://dx.doi.org/10.5075/epfl-thesis-4783

A video explaining the rotating scanner hardware and the lightslam algorithm is available at http://www.youtube.com/watch?v=mMkRRX7Oxc8 .

More information about the Mobots group are available on its web site at http://mobots.epfl.ch .

This video is available for download in h264 and webm formats on http://www.archive.org/details/AutonomousConstructionInUnknownEnvironmentsWit... .

Category:

Science & Technology

Tags:

• robot
• reasoning
• HTN planning
• symbol grounding
• autonomous construction
• miniature mobile robot
• computer technology
• electronics gadget
• marXbot