Situated Dialogue Processing

 
 

Talking robots. They speak to our imagination -- C3PO, Sonny, R2-D2. We are fascinated by the idea of technology we can talk to, work with, all taking place in the world we live in, the places we inhabit.


And that is, in a nutshell, what makes it challenging to build such systems -- or even just the capabilities for robots to conduct a dialogue, to interact with a human. Because understanding dialogue is about more than just understanding the speech signal, words, or the utterance. For a robot to understand dialogue when talking with a  human, it ultimately needs to understand how that dialogue relates and refers to the world we live in.


Over the last years, we have been working on making this possible. For publications on how we have built talking robots for exploring indoor environments, learning visual objects or playing games, have a look at http://talkingrobots.dfki.de.

On this website you can download open source code, tools, and resources for helping you build dialogue systems, to make your robots talk.


  1. The SitDial Java API provides a basic class library with representations and functions for
      situated dialogue comprehension, planning and realization.

  2. The SitDial CAST Toolkit turns these into components, to integrate dialogue into a CAST
      -based cognitive architecture.

  3. The SitDial Tools & Resources provide tools for visualizing and debugging aspects of a
      dialogue system, and fundamental resources such as grammars, and data sets for training
      statistical models.


See the news blog for the latest news on the SitDial API.

Making talking robots