图 书 馆
用于快速可伸缩三维重建和映射的因子图Factor Graphs for Fast and Scalable 3D Reconstruction and Mapping |
|
| 课程网址: | http://videolectures.net/bmvc2013_dellaert_factor_graphs/ |
| 主讲教师: | Frank Dellaert |
| 开课单位: | 乔治亚理工学院 |
| 开课时间: | 2014-04-03 |
| 课程语种: | 英语 |
| 中文简介: |  同时定位和制图（SLAM）和运动结构（SFM）是机器人技术和视觉领域中重要且密切相关的问题。我将展示如何通过图形模型，因子图来构成SLAM和SFM实例，并且这些图中的推论可以理解为变量消除。演讲的总体主题将是强调从图形模型的角度来看这些问题所带来的优势和直觉。例如，常见的计算技巧（例如SFM中的Schur补充技巧）是消除图的顺序的简单选择。另外，虽然图形模型的观点是完全通用的，但线性化非线性因子并假设高斯噪声会产生熟悉的直接线性求解器，例如Cholesky和QR分解。基于这些见解，我们开发了在SLAM / SFM域中的图上定义的批处理和增量算法。除了直接方法之外，我们最近还研究了有效的迭代方法，该方法使用这些因子图的子图作为共轭梯度方案中的预处理器。最后，我们现在正在研究如何将最佳控制与自动驾驶汽车中的估计算法无缝集成。 |
| 课程简介: | Simultaneous Localization and Mapping (SLAM) and Structure from Motion (SFM) are important and closely related problems in robotics and vision. I will show how both SLAM and SFM instances can be posed in terms of a graphical model, a factor graph, and that inference in these graphs can be understood as variable elimination. The overarching theme of the talk will be to emphasize the advantages and intuition that come with seeing these problems in terms of graphical models. For example, common computational tricks, such as the Schur complement trick in SFM, are simple choices about the order in which to eliminate the graph. In addition, while the graphical model perspective is completely general, linearizing the non-linear factors and assuming Gaussian noise yields the familiar direct linear solvers such as Cholesky and QR factorization. Based on these insights, we have developed both batch and incremental algorithms defined on graphs in the SLAM/SFM domain. In addition to direct methods, we recently worked on efficient iterative methods that use subgraphs of these factor graphs as pre-conditioners in a conjugate gradient scheme. Finally, we are now looking into how optimal control can be seamlessly integrated with the estimation algorithms for use in autonomous vehicles. |
| 关 键 词: | 图形模型; 增量算法 |
| 课程来源: | 视频讲座网 |
| 数据采集: | 2021-03-25：zyk |
| 最后编审: | 2021-03-25：zyk |
| 阅读次数: | 25 |