matches. Consider for example the case when a landmark is occluded for a
short period of time. A spatial compatibility test would not have any information
on the history of observations of such landmark, and might still be trying
to wrongly associate it with a neighboring observed feature. If the algorithm
succeeds in incorrectly associating the occluded feature, the new observation
will not be consistent with the initial measurement, thus producing large error
in the estimate for the localization of that landmark, while at the same time
underestimating its covariance. Given that the map covariance is fully correlated,
starting with the next iteration of the algorithm, that wrong value for
the uncertainty would be propagated to the rest of the landmark locations,
and that of the robot as well; leading to pergence in the map, and ultimately
breaking down the entire estimation approach to SLAM.
To aid in those situations in which landmark observations might not be
consistent in time, we propose a new set of temporal landmark quality models,
and show how by incorporating these models, the overall estimation-theoretic
approach to SLAM is improved. With the aid of these models, a new temporal
landmark quality test is presented to aid in differentiating between the
imprecision in the localization of a landmark, and the uncertainty in its very
existence. Thanks to this test we are able to remove weak landmarks from the
map. Landmarks that would most likely be a product of false data association
or spurious observations, and that if considered, would otherwise induce
 
Fig. 1.1. The blue dots indicate sensor raw data coming from a laser range finder,
and the blue ellipses represent 2σ confidence level curves on the wall end point
estimates. The green lines represent walls inferred from consecutive readings. The
red lines indicate the estimated robot trajectory.
undesired localization errors. Temporal landmark compatibility is addressed
in Section 1.3.
Finally, in Section 1.4, our planar mobile robot configuration is used to
evaluate the original full-covariance Extended Kalman Filter algorithm to Simultaneous
Localization and Map Building as reported by Dissanayake et al
[31], including the spatial landmark compatibility tests [70], versus our improved
algorithm, the EKF-SLAM-LV, with both temporal and spatial landmark
quality tests, both in the presence of various noise levels, and ultimately,
in cases with limited field of view and extreme data missassociation.
同时定位和地图构建本地感知发生,以自我为中心的参照系的机器人。为了确保当地的代表之间的通信环境建造的具有里程碑意义的提取过程,和全球表示包含在一个地图,机器人必须估计自己的位置这张地图。论文网
使用随机模型在移动地图构建和本地化机器人技术近年来获得了许多流行。特别兴趣是使用预测估计机器人位置和过滤器不确定性,从传感器读数同时更新这些估计同时建立一个增量的环境地图。
最重要的限制之一estimationtheoretic等的应用地图构建方法和本地化是数据关联问题。数据关联是指匹配问题的观察以前学习环境中的元素。一些技术可以用于缓解数据关联问题,比如地标的跟踪机器人从一个位置到另一个,或者通过使用高效测试sceneto -里程碑式的匹配假说验证模型。显然总是有妥协的可能性完全不变的标志性特征这样的描述特征的提取和困难从原始传感器数据。 同时定位和地图构建英文文献和中文翻译(2):http://www.youerw.com/fanyi/lunwen_31596.html