Benchmarking urban six‐degree‐of‐freedom simultaneous localization and mapping

Abstract

Quite a number of approaches for solving the simultaneous localization and mapping (SLAM) problem exist by now. Some of them have recently been extended to mapping environments with six-degree-of-freedom poses, yielding 6D SLAM approaches. To demonstrate the capabilities of the respective algorithms, it is common practice to present generated maps and successful loop closings in large outdoor environments. Unfortunately, it is nontrivial to compare different 6D SLAM approaches objectively, because ground truth data about the outdoor environments used for demonstration are typically unavailable. We present a novel benchmarking method for generating the ground truth data based on reference maps. The method is then demonstrated by comparing the absolute performance of some previously existing 6D SLAM algorithms that build a large urban outdoor map. C

A comparison of SLAM algorithms based on a graph of relations

“…Wolf et al [29] proposed the idea of using manually supervised Monte Carlo Localization (MCL) for matching 3D scans against a reference map. They suggested to generate the reference maps from independently created CAD data, which can be obtained from the land registry office.…”

A Review of Point Cloud Registration Algorithms for Mobile Robotics

“…Moreover, the construction of the tree requires some time, and often only the reference is used as seeding points. Limiting the number of kD-tree constructions by the use of keyframes or metascans can help to decrease the registration time for a sequence of scans, while limiting the drift of the final path [Wulf et al, 2008].…”

“…Finally, the newly registered scan is concatenated with the global map and filters are applied to the resulting map before making it available to the next scan registration. This type of processing pipeline was first proposed by Wulf et al [2008] under the name of Registration Use Cases metascans, which brings the random walk error as a function of distance newly explored instead of time. All of those processing steps were made using the library libpointmatcher presented in .…”

A New Method for Relative/Quantitative Comparison of Map Built by SLAM

“…야외 환경에서 3D 센서로 측정한 정보를 4가지 서로 다 른 SLAM 방법으로 처리하여 작성한 지도들의 정확도를 절 대좌표계 상에서 비교하는 방법이 제안되었다 [6] . 실내 환경에서 유사한 개념을 사용하여 3가지 서로 다 른 SLAM 기법으로 작성한 지도를 비교하는 방법도 제안 되었다 [7] .…”