Current practices in knowledge representation for robotics
General
- Name of the project
- Names of the main contributors
Knowledge model
- Main paradigms (like ontologies, fluents, semantic frames, relational databases...)
- How would be represented the fact: "The bottle (id 1845) is on the table (id 2365)"?
- Which formalisms/standards/standard corpus are used (OWL, FrameNet, Cyc...)
- Expressivity (for Description Logics, cf http://www.cs.man.ac.uk/~ezolin/dl/)
- Open World Assumption or Closed World Assumption?
- Resonning capabilities
- Reasonner available?
- If yes, external tool or part of the KB?
- Capable of representing uncertain knowledge?
- Management of time?
- How time is represented?
- How is it associated to statements?
- What is it used for?
- Has an associated rule language?
- Modifiable at run-time?
- A-Box only?
- or T-Box and A-Box?
Area of use
- Task planning
- Plan representation
- Spatial mapping
- Dialogue grounding
- Learning/Experience modelling
- Human-robot interaction
- Other area?
- References to main experiment conducted with the KB
Features
- Capable of fetching external sources of knowledge (like Wikipedia,...)
- Automaticaly?
- Grounding mechanism?
- Could you briefly describe how perception are grounded in the KB?
- Integration with supervision processes?
- Could you briefly give an example?
- Other features?
Other
- Main programming language
- How does the KB integrate into the robotic architecture? (language bindings, middleware wrappers...)
- Has been deployed on a real, physical platform?
- If yes, has been deployed on more that one platform?
- Project homepage, API documentation, tutorials/examples
- Main publications that present the project