## A real-time algorithm for mobile robot mapping with applications to multi-robot and 3D mapping (2000)

### Cached

### Download Links

- [www.ai.mit.edu]
- [www-2.cs.cmu.edu]
- [robots.stanford.edu]
- [www.cs.cmu.edu]
- DBLP

### Other Repositories/Bibliography

Venue: | In IEEE International Conference on Robotics and Automation |

Citations: | 256 - 37 self |

### BibTeX

@INPROCEEDINGS{Thrun00areal-time,

author = {Sebastian Thrun and Wolfram Burgard and Dieter Fox},

title = {A real-time algorithm for mobile robot mapping with applications to multi-robot and 3D mapping},

booktitle = {In IEEE International Conference on Robotics and Automation},

year = {2000}

}

### Years of Citing Articles

### OpenURL

### Abstract

We present an incremental method for concurrent mapping and localization for mobile robots equipped with 2D laser range finders. The approach uses a fast implementation of scan-matching for mapping, paired with a sample-based probabilistic method for localization. Compact 3D maps are generated using a multi-resolution approach adopted from the computer graphics literature, fed by data from a dual laser system. Our approach builds 3D maps of large, cyclic environments in real-time. It is remarkably robust. Experimental results illustrate that accurate maps of large, cyclic environments can be generated even in the absence of any odometric data. 1

### Citations

1131 | CONDENSATION conditional density propagation for visual tracking
- Isard, Blake
- 1998
(Show Context)
Citation Context ...t−1,st−1,mt−1) importance sampler [14] and, in the context of temporal posterior estimation is known as particle filters [12]. It has, with great success, been applied for tracking in computer vision =-=[7]-=- and mobile robot localization [2, 3]. As argued in the statistical literature, this representation can approximate almost arbitrary posteriors at a convergence rate of 1 √ N [17]. It is convenient fo... |

945 | P.: Surface simplification using quadric error metrics
- GARLAND, HECKBERT
- 1997
(Show Context)
Citation Context ...part from the expected measurement under the robot motion (and assuming the robot faces a straight wall), this measurement does not become part of a polygone. We also apply a simplification algorithm =-=[4, 5]-=-, previously developed to simplify polygonal models for real-time rendering in computer graphics. 1 In a nutshell, this approach iteratively simplifies multi-polygon surface models by fusing polygons ... |

514 | Filtering via simulation: Auxiliary particle filters
- Pitt, Shephard
- 1999
(Show Context)
Citation Context ...,dt−1,at−1,mt−1) P(st | dt−1,at−1,mt−1) = ηP(ot | st,mt−1) (10) � P(st | dt−1,at−1,st−1,mt−1) importance sampler [14] and, in the context of temporal posterior estimation is known as particle filters =-=[12]-=-. It has, with great success, been applied for tracking in computer vision [7] and mobile robot localization [2, 3]. As argued in the statistical literature, this representation can approximate almost... |

406 |
High Resolution Maps from Wide Angle Sonar
- Moravec, Elfes
- 1985
(Show Context)
Citation Context ...lem, to indicate its chicken-and-egg nature: Building maps when a robot’s locations are known is relatively straightforward, as early work by Moravec and Elfes has demonstrated more than a decade ago =-=[10]-=-. Conversely, localizing a robot when a map is readily available is also relatively well-understood, as a flurry of algorithms has successfully demonstrated [1]. In combination, however, the problem i... |

403 | A probabilistic approach to concurrent mapping and localization for mobile robots
- Thrun, Fox, et al.
- 1998
(Show Context)
Citation Context ...and when closing the cycle error has to be corrected backwards in time (which most existing methods are incapable of doing). A recent family of probabilistic methods based on EM overcome this problem =-=[15, 18]-=-. EM searches the most likely map by simultaneously considering the locations of all past scans, using a probabilistic argument for iterative refinement during map construction. While these approaches... |

277 | Monte Carlo Localization: Efficient position estimation for mobile robots
- Fox, Burgard, et al.
- 1999
(Show Context)
Citation Context ... to the EM family of mapping algorithms [15, 18]. 2.3 Incremental Mapping Using Posteriors Following the literature on probabilistic mapping with EM [15, 18] and the literature on Markov localization =-=[16, 3]-=-, our approach computes the full posterior over robot poses, instead of the maximum likelihood pose only (as given in (6)). The posterior is a probability distributionover poses conditioned on past se... |

253 | Probabilistic robot navigation in partially observable environments
- Simmons, Koenig
- 1995
(Show Context)
Citation Context ... to the EM family of mapping algorithms [15, 18]. 2.3 Incremental Mapping Using Posteriors Following the literature on probabilistic mapping with EM [15, 18] and the literature on Markov localization =-=[16, 3]-=-, our approach computes the full posterior over robot poses, instead of the maximum likelihood pose only (as given in (6)). The posterior is a probability distributionover poses conditioned on past se... |

230 | Robot pose estimation in unknown environments by matching 2D range scans
- Lu, Milios
- 1997
(Show Context)
Citation Context ...in the complete absence of odometry data, and it also extends to the generation of 3D with laser range finders. 2.1 Likelihood Function Let m be a map. Following the literature on laser scan matching =-=[6, 9]-=-, we assume that a map is a collection of scans and their poses; the term pose refers to the x-y location relative to some hypothetical coordinate system and a scan’s orientation θ. At time t, the map... |

181 |
Dynamic map building for an autonomous mobile robot
- Leonard, Durrant-Whyte, et al.
- 1992
(Show Context)
Citation Context ...rd. Recent progress has led to a range of new methods. Most of these approaches build maps incrementally, by iterating localization and incremental mapping for each new sensor scan the robot receives =-=[8, 13, 19, 20]-=-. While these methods are fast and can well be applied in real-time, they typically fail when mapping large cyclic environments. This is because in environments with cycles, the robot’s cumulative err... |

169 | Simplifying surfaces with color and texture using quadric error metrics
- Garland, Heckbert
- 1998
(Show Context)
Citation Context ...part from the expected measurement under the robot motion (and assuming the robot faces a straight wall), this measurement does not become part of a polygone. We also apply a simplification algorithm =-=[4, 5]-=-, previously developed to simplify polygonal models for real-time rendering in computer graphics. 1 In a nutshell, this approach iteratively simplifies multi-polygon surface models by fusing polygons ... |

168 |
Tools for Statistical Inference
- Tanner
- 1993
(Show Context)
Citation Context ...ng in computer vision [7] and mobile robot localization [2, 3]. As argued in the statistical literature, this representation can approximate almost arbitrary posteriors at a convergence rate of 1 √ N =-=[17]-=-. It is convenient for robotics, since it is easy tos(a) ✁ ✁✕ robot and samples (b) ✁ ✁✕ robot and samples (c) ✛ robot and samples Figure 6: Incremental algorithm for concurrent mapping and localizati... |

153 |
Using the SIR Algorithm to Simulate Posterior Distributions
- Rubin
- 1988
(Show Context)
Citation Context ... dt−1,at−1,ot,mt−1) ure 2). The idea of using samples goes back to Rubin’s = ηP(ot | st,dt−1,at−1,mt−1) P(st | dt−1,at−1,mt−1) = ηP(ot | st,mt−1) (10) � P(st | dt−1,at−1,st−1,mt−1) importance sampler =-=[14]-=- and, in the context of temporal posterior estimation is known as particle filters [12]. It has, with great success, been applied for tracking in computer vision [7] and mobile robot localization [2, ... |

119 |
Navigating Mobile Robots: Systems and Techniques
- Borenstein, Everett, et al.
- 1996
(Show Context)
Citation Context ...as demonstrated more than a decade ago [10]. Conversely, localizing a robot when a map is readily available is also relatively well-understood, as a flurry of algorithms has successfully demonstrated =-=[1]-=-. In combination, however, the problem is hard. Recent progress has led to a range of new methods. Most of these approaches build maps incrementally, by iterating localization and incremental mapping ... |

116 | Using the condensation algorithm for robust, vision-based mobile robot localization
- Dellaert, Burgard, et al.
- 1999
(Show Context)
Citation Context ...[14] and, in the context of temporal posterior estimation is known as particle filters [12]. It has, with great success, been applied for tracking in computer vision [7] and mobile robot localization =-=[2, 3]-=-. As argued in the statistical literature, this representation can approximate almost arbitrary posteriors at a convergence rate of 1 √ N [17]. It is convenient for robotics, since it is easy tos(a) ✁... |

75 | Spatial learning for navigation in dynamic environments
- Yamauchi, Beer
- 1996
(Show Context)
Citation Context ...rd. Recent progress has led to a range of new methods. Most of these approaches build maps incrementally, by iterating localization and incremental mapping for each new sensor scan the robot receives =-=[8, 13, 19, 20]-=-. While these methods are fast and can well be applied in real-time, they typically fail when mapping large cyclic environments. This is because in environments with cycles, the robot’s cumulative err... |

67 | AMOS: Comparison of scan matching approaches for self-localization in indoor environments
- Gutmann, Schlegel
- 1996
(Show Context)
Citation Context ...in the complete absence of odometry data, and it also extends to the generation of 3D with laser range finders. 2.1 Likelihood Function Let m be a map. Following the literature on laser scan matching =-=[6, 9]-=-, we assume that a map is a collection of scans and their poses; the term pose refers to the x-y location relative to some hypothetical coordinate system and a scan’s orientation θ. At time t, the map... |

44 |
Concurrent localisation and map building for mobile robots using ultrasonic sensors
- Rencken
- 1993
(Show Context)
Citation Context ...rd. Recent progress has led to a range of new methods. Most of these approaches build maps incrementally, by iterating localization and incremental mapping for each new sensor scan the robot receives =-=[8, 13, 19, 20]-=-. While these methods are fast and can well be applied in real-time, they typically fail when mapping large cyclic environments. This is because in environments with cycles, the robot’s cumulative err... |

39 |
Comparison of scan matching approaches for self-localization in indoor environments
- Gutmann, Schlegel, et al.
- 1996
(Show Context)
Citation Context ...in the complete absence of odometry data, and it also extends to the generation of 3D with laser range finders. 2.1 Likelihood Function Let m be a map. Following the literature on laser scan matching =-=[6, 9]-=-, we assume that a map is a collection of scans and their poses; the term pose refers to the x-y location relative to some hypothetical coordinate system and a scan’s orientation θ. At time t, the map... |

26 | Learning Models for Robot Navigation
- Shatkay
- 1998
(Show Context)
Citation Context ...and when closing the cycle error has to be corrected backwards in time (which most existing methods are incapable of doing). A recent family of probabilistic methods based on EM overcome this problem =-=[15, 18]-=-. EM searches the most likely map by simultaneously considering the locations of all past scans, using a probabilistic argument for iterative refinement during map construction. While these approaches... |

19 | Magellan: An Integrated Adaptive Architecture for Mobile Robots
- Yamauchi, Langley, et al.
- 1998
(Show Context)
Citation Context |

16 | Robot navigation by 3D spatial evidence grids - Moravec, Martin - 1994 |