Presently we are working on the problem of qualitative robot navigation based on vision. The robots will be equipped with one or more cameras to acquire images from the environment, which will be shown in a control screen. An operator will indicate on the observed image the target place where the robot should go. The robot will have to reach that target in a fully autonomous way by means of qualitative landmark-based navigation and using only visual information in a way similar to that used by humans. The work will be carried out on real mobile robots (wheeled and legged) in unknown environments of increasing complexity. Wheeled robots will be used in office-like environments or smooth outdoors. In more abrupt terrains, an hexapod will be used.

This research is being done in collaboration with the Institute of Robotics and Industrial Informatics (IRII) of CSIC-UPC (http://www-iri.upc.es). At the IIIA we are developping the qualitative navigation system which is based on a multiagent bidding mechanism. More information on the navigation system and a demo of a simulation presented at the Agents'2000 Conference can be found at http://www.iiia.csic.es/~didac/demo-robot/ . Some images and videos of the real robot we are using for this project can be seen here.

The IIIA is partner of the International Quality Network (IQN) on Spatial Cognition.

• C. Sierra, R.López de Màntaras and D. Busquets. Multiagent Bidding Mechanisms for Robot Qualitative Navigation. Intelligent Agents VII. Lecture Notes in Artificial Intelligence, vol. 1986 (Proceedings of ATAL'00), pp 198-212, Springer, Verlag, 2001.
  [abstract] [gzipped PostScript] [gzipped PDF]
• D. Busquets, R.López de Màntaras and C. Sierra. A robust MAS coordination mechanism for action selection. Proceedings of 2001 AAAI Spring Symposium, Stanford, CA. Robust Autonomy Serie, pp. 38-40.
  [abstract] [gzipped PostScript] [gzipped PDF]
• D. Busquets, C. Sierra, R.López de Màntaras. A Multiagent Approach to Qualitative Landmark-based Navigation Technical Report, 2001.
  [abstract] [gzipped PostScript] [gzipped PDF]
• D. Busquets, R.López de Màntaras, C. Sierra and T.G. Dietterich. Reinforcement Learning for Landmark-based Robot Navigation. Proceedings of the first International Joint Conference on Autonomous Agents and Multiagent Systems (AAMAS 2002), pp. 841-842, ACM press, 2002.
  [abstract] [gzipped PostScript] [gzipped PDF]
• D. Busquets, T.G. Dietterich, R.López de Màntaras and C. Sierra. A mulit-agent architecture integrating learning and fuzzy techniques for landmark-based robot navigation. Topics in Artificial Intelligence. Lecture Notes in Artificial Intelligence (Proceedings of CCIA'02), vol. 2504, pp. 269-281, Springer, Verlag, 2002.
  [abstract] [gzipped PostScript] [gzipped PDF]
• T.G. Dietterich, D. Busquets, R.López de Màntaras and C. Sierra. Action Refinement in Reinforcement Learning by Probability Smoothing Proceedings of the 19th International Conference on Machine Learning (ICML'02), pp. 107-114.
  [abstract] [gzipped PostScript] [gzipped PDF]
• D. Busquets, C. Sierra and R. López de Màntaras. A multi-agent approach to fuzzy landmark-based navigation. Journal of Multiple-Valued Logic and Soft Computing, Vol. 9, pp. 195-220, Old City Publishing, 2003.
  [abstract] [gzipped PostScript] [gzipped PDF]
• D. Busquets, C. Sierra and R. López de Màntaras. A Multi-agent approach to qualitative landmark-based navigation. Autonomous Robots Journal, Vol. 15, pp. 129-153, 2003.
  [abstract] [gzipped PostScript] [gzipped PDF]
• D. Busquets, R.López de Màntaras and C. Sierra. Evolving a Multiagent System for Landmark-Based Robot Navigation. Artificial Intelligence Research and Development (Proceedings of CCIA'03), pp. 87-98, IOS Press, 2003.
  [abstract] [gzipped PostScript] [gzipped PDF]
• M. Mohan, D. Busquets, R.López de Màntaras and C. Sierra. Integrating a Potential Field Based Pilot Into a Multiagent Navigation Architecture for Autonomous Robots. Proceedings of ICINCO'04 (Vol. 2), pp. 287-290, INSTICC Press, 2004.
  [abstract] [gzipped PostScript] [gzipped PDF]