## H.: Graphical SLAM - a self-correcting map (2004)

Venue: | In: IEEE Intl. Conf. on Robotics and Automation (ICRA |

Citations: | 49 - 2 self |

### BibTeX

@INPROCEEDINGS{Folkesson04h.:graphical,

author = {John Folkesson and Henrik Christensen},

title = {H.: Graphical SLAM - a self-correcting map},

booktitle = {In: IEEE Intl. Conf. on Robotics and Automation (ICRA},

year = {2004},

pages = {383--390}

}

### Years of Citing Articles

### OpenURL

### Abstract

Abstract β In this paper we describe an approach to simultaneous localization and mapping, SLAM. This approach has the highly desirable property of robustness to data association errors. Another important advantage of our algorithm is that non-linearities are computed exactly, so that global constraints can be imposed even if they result in large shifts to the map. We represent the map as a graph and use the graph to find an efficient map update algorithm. We also show how topological consistency can be imposed on the map, such as, closing a loop. The algorithm has been implemented on an outdoor robot and we have experimental validation of our ideas. We also explain how the graph can be simplified leading to linear approximations of sections of the map. This reduction gives us a natural way to connect local map patches into a much larger global map. I.

### Citations

427 | Globally consistent range scan alignment for environment mapping. Autonomous Robots, 1997. BIBLIOGRAPHIE tel-00545774, version 1 - 12 Dec 2010 [LM97b] Feng Lu and Evangelos Milios. Robot pose estimation in unknown environments by matching 2d range scans
- Lu, Milios
- 1997
(Show Context)
Citation Context ... of closing large loops, is much harder. Many methods can not use the information that the robot has closed a loop to fix the map. Some methods are designed to use this type of information explicitly =-=[4]-=-, [5], [6]. The idea of combining topological and metric information in a unified approach was shown in [7] to be a powerful one. Our method is similar to those methods and can also use the global con... |

403 | A probabilistic approach to concurrent mapping and localization for mobile robots
- Thrun, Fox, et al.
- 1998
(Show Context)
Citation Context ... has occurred, many of the existing methods have no way to correct or detect these errors. Others can do limited correction. Matching errors are handled best by expectation maximization, EM, methods, =-=[2]-=-. When applied in the pure form EM will find the best set of matches. In real-time implementations of EM this desirable property must be partially sacrificed as the number of combinations of matches i... |

347 | A solution to the simultaneous localization and map building (slam) problem
- Gamini, Newman, et al.
- 2001
(Show Context)
Citation Context ... to be able to learn these maps as they move about the environment. This problem is known as simultaneous mapping and localization, (SLAM). The SLAM problem is central to autonoumous mobile robotics, =-=[1]-=-. This paper presents a new way to look at the problem focusing on the issues that have caused the most trouble for other methods. We have tried to combine the best features of existing approaches in ... |

281 | Incremental mapping of large cyclic environments
- Gutmann, Konolige
- 1999
(Show Context)
Citation Context ...losing large loops, is much harder. Many methods can not use the information that the robot has closed a loop to fix the map. Some methods are designed to use this type of information explicitly [4], =-=[5]-=-, [6]. The idea of combining topological and metric information in a unified approach was shown in [7] to be a powerful one. Our method is similar to those methods and can also use the global consiste... |

92 | Simultaneous localization and mapping with unknown data association using FastSLAM, in
- Montemerlo, Thrun
- 2003
(Show Context)
Citation Context ...has no way to find and jump to another region of the state space. Therefore, it can only be part of a larger solution to the SLAM problem and not the ultimate solution. Multiple hypothesis solutions, =-=[9]-=-, will suffer from the fact that over large loops the number of hypothesis that will be needed will be way too large. One needs a single real-time solution for many applications but one would like to ... |

36 | Correction of dead-reckoning errors in map building for mobile robots,β Robotics and Automation
- Golfarelli, Maio, et al.
- 2001
(Show Context)
Citation Context ...g large loops, is much harder. Many methods can not use the information that the robot has closed a loop to fix the map. Some methods are designed to use this type of information explicitly [4], [5], =-=[6]-=-. The idea of combining topological and metric information in a unified approach was shown in [7] to be a powerful one. Our method is similar to those methods and can also use the global consistency c... |

12 |
Outdoor Exploration and SLAM using a Compressed Filter
- Folkesson, Christensen
- 2003
(Show Context)
Citation Context ...ds, (steepest descents and eq. (15)), for per node minimization, as relaxing the node.sHaving implemented only this much we found that the maps were already looking better than our previous SLAM maps,=-=[10]-=-. The updates were rather slow however. One must relax the new node then all the adjacent nodes then if any of those changed one must relax any adjacent to them and so on. This causes the update to mo... |

5 |
et al., βAn atlas framework for scalable mapping
- Bosse
- 2003
(Show Context)
Citation Context ... a loop to fix the map. Some methods are designed to use this type of information explicitly [4], [5], [6]. The idea of combining topological and metric information in a unified approach was shown in =-=[7]-=- to be a powerful one. Our method is similar to those methods and can also use the global consistency constraints to improve the map. Another major difficulty inherent in the SLAM problem is the incon... |

5 |
Results fo outdoor-slam using sparse extended information filters
- Lui, Thrun
(Show Context)
Citation Context ...n put back in the constraint that the stars have features in common to refine the solution. The ideas here result in a representation of the map very similar to the sparse extended information filter =-=[11]-=-. The difference here is that we do not linearize once at the time of the observation and lock the approximation in that frame. Instead we linearize a section of the graph in a relative frame and at a... |

2 |
Real-time acquistion of compact volumetric 3d maps with mobile robots
- Martin, Thurn
- 2002
(Show Context)
Citation Context ...d in the pure form EM will find the best set of matches. In real-time implementations of EM this desirable property must be partially sacrificed as the number of combinations of matches is very large =-=[3]-=-. One reason that data association is such a problem is that the really hard problems are ones that look reasonable locally but lead to global failure. With real data, matching errors often do not lea... |

1 |
A counter examplee to the theory of simultaneous localization and map building
- Julier, Uhlmann
- 2001
(Show Context)
Citation Context ...s, the linearizations of the measurements will be about different points. This leads to inconsistancies in the resulting map after many iterations. Julier and Uhlmann have first explained the problem =-=[8]-=-. The method we propose does not suffer from this problem as we work with the full non-linear problem. The approach does suffer from local minimum. It is possible for our map to become trapped in a lo... |