#### DMCA

## Process · 3D Point Cloud

### Citations

719 | Gaussian processes for machine learning
- Rasmussen, Williams
- 2009
(Show Context)
Citation Context ... the number of points in segment m. In this paper, we assume that these examples are conditionally independent, and try to seek these hyper-parameters by maximizing the pseudo log marginal likelihood =-=[24]-=-: M∑ m=1 log p(zm|rm, θ) = −1 2 M∑ m=1 (zm)TK−1m z m− 1 2 log ( M∏ m=1 |Km| )− log 2π 2 M∑ m=1 nm, (14) where Km is the covariance matrix for the noisy z m values in segment m, and can be calculated u... |

209 | 3D is here: Point Cloud Library (PCL).
- Rusu, Cousins
- 2011
(Show Context)
Citation Context ...e and the accuracy of the algorithm proposed in this paper, two other algorithms proposed in [5, 13] are adopted to be compared with. We implement them in C++ as well as the point cloud library (PCL) =-=[25]-=-, and conduct several contrastive experiments both in urban and countryside environments. Our experimental platform is a modied Toyota Land Cruiser, as shown in Fig. 6, equipped with a Velodyne HDL-6... |

86 | A comparison of line extraction algorithms using 2D range data for indoor mobile robotics.
- Nguyen, Gächter, et al.
- 2007
(Show Context)
Citation Context ...ed a fast method for segmentation of large-size long range 3D point clouds for a high-speed ALV. The approach divides the polar grid map into many segments, and uses the line extraction algorithms in =-=[19]-=- to represent the ground surface in each segment. The method splits the segmentation problem into many simplestting problems with lower complexity. It can get a real-time performance, but it is not su... |

58 | Nonstationary covariance functions for Gaussian process regression.
- Paciorek, Schervish
- 2004
(Show Context)
Citation Context ...R system [3]. Our method constructs a circular polar grid map, and uses a one-dimensional Gaussian process regression with the non-stationary covariance function proposed by Paciorek and Schervish in =-=[21]-=- to allow for local adaptation to the undulation of the ground in each segment of the map. The adaptation is achieved by calculating the local gradient in each segment. Our method is inspired by refer... |

28 | Object Recognition in 3D Point Clouds Using Web Data and Domain Adaptation
- Lai, Fox
- 2010
(Show Context)
Citation Context ...ail. Title Suppressed Due to Excessive Length 5 Some experimental results are showed in Section 4. Our algorithm are compared with two previous ground segmentation techniques on the public dataset in =-=[16]-=- and the data collected by our ALV in dierent outdoor scenes. The results show a promising performance of our algorithm. Finally, conclusions and an outline of the future work are given in Section 5.... |

17 | On the segmentation of 3D lidar point clouds,”
- Douillard, Underwood, et al.
- 2011
(Show Context)
Citation Context ...ain, many ground segmentation approaches have been proposed, which can roughly be divided into three subgroups: grid cell-based methods [6,29], line-based methods [10,13,26] and surface-based methods =-=[5,11,17,23,31]-=-. Thrun et al. [29] presented a ground segmentation method based on the MinMax elevation map, which was widely used by the teams at the 2007 DARPA (Defense Advanced Research Projects Agency) Urban Cha... |

17 | Adaptive nonstationary kernel regression for terrain modeling
- Lang, Plagemann, et al.
- 2007
(Show Context)
Citation Context ...ain, many ground segmentation approaches have been proposed, which can roughly be divided into three subgroups: grid cell-based methods [6,29], line-based methods [10,13,26] and surface-based methods =-=[5,11,17,23,31]-=-. Thrun et al. [29] presented a ground segmentation method based on the MinMax elevation map, which was widely used by the teams at the 2007 DARPA (Defense Advanced Research Projects Agency) Urban Cha... |

13 | Nonstationary gaussian process regression using point estimates of local smoothness
- Plagemann, Kersting, et al.
- 2008
(Show Context)
Citation Context ...nput location ri has an individual length-scale li. The covariance between two outputs zi and zj is calculated by averaging between the two lengthscales at the corresponding input locations ri and rj =-=[22]-=-. In this way, the covariance function of the model can describe the local characteristics of the ground better, to some extent. In this paper, the length-scales are adapted with the 1D gradient infor... |

13 | Learning Predictive Terrain Models for Legged Robot Locomotion
- Plagemann, Mischke, et al.
- 2008
(Show Context)
Citation Context ...ain, many ground segmentation approaches have been proposed, which can roughly be divided into three subgroups: grid cell-based methods [6,29], line-based methods [10,13,26] and surface-based methods =-=[5,11,17,23,31]-=-. Thrun et al. [29] presented a ground segmentation method based on the MinMax elevation map, which was widely used by the teams at the 2007 DARPA (Defense Advanced Research Projects Agency) Urban Cha... |

9 |
Fast segmentation of 3D point clouds for ground vehicles,”
- Himmelsbach, Hundelshausen, et al.
- 2010
(Show Context)
Citation Context ... frequency domain. In the spatial domain, many ground segmentation approaches have been proposed, which can roughly be divided into three subgroups: grid cell-based methods [6,29], line-based methods =-=[10,13,26]-=- and surface-based methods [5,11,17,23,31]. Thrun et al. [29] presented a ground segmentation method based on the MinMax elevation map, which was widely used by the teams at the 2007 DARPA (Defense Ad... |

6 |
et al.: Stanley: The robot that won the darpa grand challenge
- Thrun
- 2006
(Show Context)
Citation Context ...ee-dimensional scans in the frequency domain. In the spatial domain, many ground segmentation approaches have been proposed, which can roughly be divided into three subgroups: grid cell-based methods =-=[6,29]-=-, line-based methods [10,13,26] and surface-based methods [5,11,17,23,31]. Thrun et al. [29] presented a ground segmentation method based on the MinMax elevation map, which was widely used by the team... |

5 | Durrant-Whyte H. Heteroscedastic Gaussian processes for data fusion in large scale terrain modeling
- Vasudevan, Ramos, et al.
- 2010
(Show Context)
Citation Context ...l coordinate) in segment m, and z∗ is the predictive height of the ground at input r∗. Since Gaussian process regression has powerful approximate ability, it has been widely used in robotics recently =-=[1, 5, 12, 17, 23, 30, 31]-=-. In this paper, a onedimensional Gaussian process regression will be used to build the ground model in every segment of the polar grid map, rather than to build a continuous twodimensional ground mod... |

4 | Y.: Rapid traversability assessment in 2.5D grid-based map on rough terrain
- Gu, Cao, et al.
- 2008
(Show Context)
Citation Context ...ors for their robots or AVLs. The methods can be divided into two groups: the methods based on the Title Suppressed Due to Excessive Length 3 monocular vision [7, 9, 28, 32, 33] and the stereo vision =-=[8, 14, 20]-=-. These methods usually use texture features and color features to segment the ground. Thus, they are highly sensitive to the illumination and shadows. Compared to cameras, laser rangenders contain m... |

4 | Non-stationary dependent Gaussian processes for data fusion in large scale terrain modeling
- Vasudevan, Ramos, et al.
- 2011
(Show Context)
Citation Context |

3 | Calibration of a Rotating Multi-Beam Lidar
- Muhammad, Lacroix
- 2010
(Show Context)
Citation Context ...ectively [15]. Unlike the references [10,26], in this paper, the points that are collected within one scan (required 0.1 second) are treated as if they are collected at the same time. The approach in =-=[18]-=- is adopted to transform the original points to the LIDAR coordinate system. Through simple translation, these points can be transformed into the vehicle's local coordinate system, in which the result... |

2 |
G.: Fft-based terrain segmentation for underwater mapping
- Douillard, Nourani-Vatani, et al.
- 2012
(Show Context)
Citation Context ...parts in their design, which makes them complex and expensive. However, they can provide accurate range information, which makes them popular and widely used in the domain of robots. Douillard et al. =-=[4]-=- presented a method based on the fast Fourier transform for segmenting three-dimensional scans of underwater unstructured terrains in the frequency domain. The lower frequency components represent the... |

2 | Space-carving kernels for accurate rough terrain estimation
- Hadsell, Bagnell, et al.
- 2010
(Show Context)
Citation Context |

2 |
Obstacle-free pathway detection by means of depth maps
- Ortigosa, Morillas, et al.
(Show Context)
Citation Context ...ors for their robots or AVLs. The methods can be divided into two groups: the methods based on the Title Suppressed Due to Excessive Length 3 monocular vision [7, 9, 28, 32, 33] and the stereo vision =-=[8, 14, 20]-=-. These methods usually use texture features and color features to segment the ground. Thus, they are highly sensitive to the illumination and shadows. Compared to cameras, laser rangenders contain m... |

2 |
Neural network based terrain classi using wavelet features
- Sung, Kwak, et al.
- 2010
(Show Context)
Citation Context ...research groups use them as the main sensors for their robots or AVLs. The methods can be divided into two groups: the methods based on the Title Suppressed Due to Excessive Length 3 monocular vision =-=[7, 9, 28, 32, 33]-=- and the stereo vision [8, 14, 20]. These methods usually use texture features and color features to segment the ground. Thus, they are highly sensitive to the illumination and shadows. Compared to ca... |

1 |
Incorporating geometric information into gaussian process terrain models from monocular images
- Abuhashim, Sukkarieh
- 2012
(Show Context)
Citation Context ...l coordinate) in segment m, and z∗ is the predictive height of the ground at input r∗. Since Gaussian process regression has powerful approximate ability, it has been widely used in robotics recently =-=[1, 5, 12, 17, 23, 30, 31]-=-. In this paper, a onedimensional Gaussian process regression will be used to build the ground model in every segment of the polar grid map, rather than to build a continuous twodimensional ground mod... |

1 |
Z.: Lidar-based long range road intersection detection
- Chen, Dai, et al.
- 2011
(Show Context)
Citation Context ...owever, the algorithm in [13] recognizes the intersection as the obstacles as shown in the blue quadrangles of the left image of Fig. 11c, which makes the successive intersection detection impossible =-=[2]-=-. Compared with the result of the algorithm in [5], our segmentation result still has some defects. The stair steps are classied as ground in the blue polygons B in the left image of Fig. 11a. Our al... |

1 |
Q.: 3d lidar-based ground segmentation
- Chen, Dai, et al.
- 2011
(Show Context)
Citation Context ...r an ALV. In this paper, we present a real-time ground segmentation approach based on Gaussian process regression in the polar grid map for an ALV equipped with Velodyne HDL-64E Scanning LIDAR system =-=[3]-=-. Our method constructs a circular polar grid map, and uses a one-dimensional Gaussian process regression with the non-stationary covariance function proposed by Paciorek and Schervish in [21] to allo... |

1 |
A.: Hybrid elevation maps : 3d sruface models for segmentation
- Douillard, Underwood, et al.
- 2010
(Show Context)
Citation Context ... Given a 2D training set D = { (ri, zi) }n i=1 of n samples, according to the denition of the Gaussian process, the joint distribution of these training samples can be expressed as: p(Z|R) ∼ N(µ,K), =-=(6)-=- where R = [r1, . . . , rn] T , Z = [z1, . . . , zn] T , µ is the mean vector, and K is the covariance matrix. For notational simplicity, the mean vector µ is often taken to be zero [24]. The covarian... |

1 |
K.: Feature-based terrain classi for littledog
- Filitchkin, Byl
- 2012
(Show Context)
Citation Context ...research groups use them as the main sensors for their robots or AVLs. The methods can be divided into two groups: the methods based on the Title Suppressed Due to Excessive Length 3 monocular vision =-=[7, 9, 28, 32, 33]-=- and the stereo vision [8, 14, 20]. These methods usually use texture features and color features to segment the ground. Thus, they are highly sensitive to the illumination and shadows. Compared to ca... |

1 |
A combination of terrain prediction and correction for search and rescue robot autonomous navigation
- Guo, Song, et al.
- 2009
(Show Context)
Citation Context ...research groups use them as the main sensors for their robots or AVLs. The methods can be divided into two groups: the methods based on the Title Suppressed Due to Excessive Length 3 monocular vision =-=[7, 9, 28, 32, 33]-=- and the stereo vision [8, 14, 20]. These methods usually use texture features and color features to segment the ground. Thus, they are highly sensitive to the illumination and shadows. Compared to ca... |

1 |
I.: Sequential classi in point clouds of urban scenes
- Hadjiliadis, Stamos
- 2010
(Show Context)
Citation Context ... frequency domain. In the spatial domain, many ground segmentation approaches have been proposed, which can roughly be divided into three subgroups: grid cell-based methods [6,29], line-based methods =-=[10,13,26]-=- and surface-based methods [5,11,17,23,31]. Thrun et al. [29] presented a ground segmentation method based on the MinMax elevation map, which was widely used by the teams at the 2007 DARPA (Defense Ad... |

1 |
S.: Non-parametric uav system identi with dependent gaussian processes
- Hemakumara, Sukkarieh
- 2010
(Show Context)
Citation Context ...l coordinate) in segment m, and z∗ is the predictive height of the ground at input r∗. Since Gaussian process regression has powerful approximate ability, it has been widely used in robotics recently =-=[1, 5, 12, 17, 23, 30, 31]-=-. In this paper, a onedimensional Gaussian process regression will be used to build the ground model in every segment of the polar grid map, rather than to build a continuous twodimensional ground mod... |

1 |
X.: Stereo matching using weighted dynamic programming on a single-direction four-connected tree. Computer Vision and Image Understanding (CVIU
- Hu, Wu, et al.
- 2012
(Show Context)
Citation Context ...pendent, and try to seek these hyper-parameters by maximizing the pseudo log marginal likelihood [24]: M∑ m=1 log p(zm|rm, θ) = −1 2 M∑ m=1 (zm)TK−1m z m− 1 2 log ( M∏ m=1 |Km| )− log 2π 2 M∑ m=1 nm, =-=(14)-=- where Km is the covariance matrix for the noisy z m values in segment m, and can be calculated using Eq. 7. To set the hyper-parameters by maximizing Eq. 14, we seek the partial derivatives of the ps... |

1 |
T.: Online algorithms for classi of urban objects in 3d point clouds
- Stamos, Hadjiliadis, et al.
- 2012
(Show Context)
Citation Context ... frequency domain. In the spatial domain, many ground segmentation approaches have been proposed, which can roughly be divided into three subgroups: grid cell-based methods [6,29], line-based methods =-=[10,13,26]-=- and surface-based methods [5,11,17,23,31]. Thrun et al. [29] presented a ground segmentation method based on the MinMax elevation map, which was widely used by the teams at the 2007 DARPA (Defense Ad... |

1 |
H.: Ribbon model based path tracking method for autonomous land vehicle
- Sun, Chen, et al.
- 2012
(Show Context)
Citation Context ...f Ordos, China, where the ALV needed tosnish the tasks such as avoiding the static obstacles, merging into vehicles, U-turns and so on. With the algorithm in this paper and path planning algorithm in =-=[27]-=-, our ALVsnished the route of ten kilometers autonomously in the urban environment of Ordos in twenty minutes, without any human intervention. 5 Conclusion and Future Work This paper presents a novel ... |

1 | L.: Learning long-range terrain perception for autonomous mobile robots
- Wang, Zhou, et al.
- 2010
(Show Context)
Citation Context |

1 |
Y.: Unstructured road segmentation based on spare representation
- Xiao, Dai, et al.
- 2012
(Show Context)
Citation Context |