An important challenge for cooperative and cognition-enabled robotics is to develop robots that have the ability to consider different possible courses of action and their likely outcomes before selecting one. This is especially important for robots that are engaged in joint tasks with human partners in which the sub-tasks and individual steps have not yet been fully specified.
To help address this challenge, CoAI JRC co-speaker Michael Beetz has created a research initiative to develop a new software framework FAME (Future-oriented cognitive Action Modeling Engine), with support from an Advanced Grant from the European Research Council. FAME will integrate information from video, text, and perceptual input in order to construct simulations in which digital twins carry out different possible courses of action and observe how they unfold. The system will also eventually be able to integrate instructions and advice offered by a human partner in the context of working together.
You can read more about the FAME project in the latest edition of up2date, the online magazine of the University of Bremen.