Back to Top

A Framework for Robust Cognitive Spatial Mapping

A. Pronobis, K. Sjöö, A. Aydemir, A. Bishop, P. Jensfelt

In 2009 Swedish Workshop on Autonomous Robots (SWAR), 2009.


Spatial knowledge constitutes a fundamental component of the knowledge base of a cognitive, mobile agent. The typical spatial knowledge representations are purely metrical and rely on information extracted from simple, but accurate metric sensors. However, in large-scale, dynamic environments, metrical global maps become harder to control and observe. The agent should be able to exploit sensory information that might be complex and non-metric, yet reflect crucial aspects of the environment. Moreover, it is not clear that the level of detail offered by metric maps is necessary, or even desirable, when the agent is a cognitive system intended to interact with the world in a human-like way. This work introduces a rigorously defined framework for building an abstracted cognitive spatial map that permits high level reasoning about space along with robust navigation and localization while maintaining a description that permits formal proofs and derivations. Although the literature contains many algorithms for spatial mapping, there is little work on the formal analysis of their fundamental requirements and properties. The idea of this work, is to take a step back and see how a rigorous formal treatment can lead the way towards a powerful spatial representation.


  author =       {Pronobis, Andrzej and Sj\"{o}\"{o}, Kristoffer and Aydemir, Alper and Bishop, Adrian N. and Jensfelt, Patric},
  title =        {A Framework for Robust Cognitive Spatial Mapping},
  booktitle =    {2009 Swedish Workshop on Autonomous Robots (SWAR)},
  year =         2009,
  month =        sep,
  address =      {V\"{a}ster\aa{}s, Sweden},
  url =          {}
© 2018. Copyright Andrzej Pronobis
  • stackoverflow
  • scholar