A control theoretic formulation of the generalized SLAM problem in robotics
2008 (English)In: 2008 American Control Conference: Vols 1-12, 2008, 2409-2414 p.Conference paper (Refereed)
Simultaneous Localization and Mapping (SLAM) has emerged as a key capability for autonomous mobile robots navigating in unknown environments. The basic idea behind SLAM is to concurrently obtain a map of the environment and an estimate of where the robot is placed within this map. In other words, the map and the robot's pose have to be estimated at the same time, given the same data set. This paper revisits this problem from a control theoretic vantage point by reformulating the SLAM problem as a problem of simultaneously estimating the state and the output map of a controlled, dynamical system. What is different with this formulation is that the map is contained in the output map and not, as previously done, in the state of the system.
Place, publisher, year, edition, pages
2008. 2409-2414 p.
, Proceedings of the American Control Conference, ISSN 0743-1619
Computer and Information Science
IdentifiersURN: urn:nbn:se:kth:diva-38746DOI: 10.1109/ACC.2008.4586852ISI: 000259261501177ScopusID: 2-s2.0-52449100691ISBN: 978-1-4244-2078-0OAI: oai:DiVA.org:kth-38746DiVA: diva2:439344
2008 American Control Conference, ACC; Seattle, WA; 11 June 2008 through 13 June 2008