Multi-Modal Human-Machine Communication

Posted 28 May 2005 at 20:13 UTC by steve

Just in time to use it with your new Shadow Robot Hand, the authors of a 2002 paper on human-machine coordinated grasping tasks have released their paper online. The full title of the paper is Multi-Modal Human-Machine Communication for Instructing Robot Grasping Tasks (PDF format). The problem facing researchers was that of how a normal, non-technical human could instruct a robot to grasp and manipulate objects. The robot needs to combine speech recognition with non-verbal clues such as gestural commands. The robot also has to coordinate its understanding of the task with its perception of the environment. This work was done by Patrick McGuire and other researchers. You may remember Patrick McGuire from the Cyborg Astrobiologist project that we've reported on in the past.