In this paper, we present a semantic mapping approach with multiple hypothesis tracking for data association. As semantic information has the potential to overcome ambiguity in measurements and place recognition, it forms an eminent modality for autonomous systems. This is particularly evident in urban scenarios with several similar looking surroundings. Nevertheless, it requires the handling of a non-Gaussian and discrete random variable coming from object detectors. Previous methods facilitate semantic information for global localization and data association to reduce the instance ambiguity between the landmarks. However, many of these approaches do not deal with the creation of complete globally consistent representations of the environment and typically do not scale well. We utilize multiple hypothesis trees to derive a probabilistic data association for semantic measurements by means of position, instance and class to create a semantic representation. We propose an optimized mapping method and make use of a pose graph to derive a novel semantic SLAM solution. Furthermore, we show that semantic covisibility graphs allow for a precise place recognition in urban environments. We verify our approach using real-world outdoor dataset and demonstrate an average drift reduction of 33 % w.r.t. the raw odometry source. Moreover, our approach produces 55 % less hypotheses on average than a regular multiple hypotheses approach.
Spatio-Temporal Metric-Semantic Mapping for Persistent Orchard Monitoring: Method and Dataset
GRASP Laboratory, University of Pennsylvania（GRASP实验室，宾夕法尼亚大学）；Department of Aeronautics and Astronautics, Stanford University（航空航天工程系，斯坦福大学）；Department of Mechanical Engineering, University of California, Merced（机械工程系，加州大学默塞德分校）
SLAM with Objects using a Nonparametric Pose Graph
Massachusetts Institute of Technology（麻省理工学院）
在未知环境中的映射和自定位是许多机器人应用中的基本能力。这些任务通常涉及将对象识别为独特特征或地标，这需要检测对象，然后为对象分配一个独特的标识符，当从不同的角度和在不同的图像中查看时，该标识符可以保持不变。\textit{data association}和\textit{simultaneous localization and mapping}（SLAM）问题在文献中分别得到了很好的研究。但这两个问题本质上是紧密相连的，而且还没有得到很好的解决。如果没有准确的SLAM，可能的数据关联是组合的，并且很容易变得难以处理。SLAM算法在没有精确数据关联的情况下，误差容易发散。本文提出了一种新的非参数姿态图模型数据关联和SLAM在一个单一的框架。进一步引入算法以在推断数据关联和执行SLAM之间交替。实验结果表明，我们的方法具有关联目标检测和定位对象在同一时间的新能力，导致显着更好的性能上的数据关联和SLAM问题比只考虑一个，忽略其他的缺陷。
PAS-SLAM: A Visual SLAM System for Planar Ambiguous Scenes
School of Electronic Information and Electrical Engineering, Dalian University of Technology（大连理工大学电子信息与电气工程学部）；Department of Automation, Tsinghua University（清华大学自动化系）；School of Electronic and Computer Engineering, Peking University（北京大学电子与计算机工程学院）；College of Information Science and Engineering, Northeastern University（东北大学信息科学与工程学院）
Monocular Camera Based Fruit Counting and Mapping with Semantic Data Association
GRASP Lab, University of Pennsylvania（宾夕法尼亚大学GRASP实验室）；School of Earth and Space Exploration, Arizona State University（亚利桑那州立大学地球与空间探索学院）；Australian Centre for Field Robotics, The University of Sydney（悉尼大学澳大利亚野外机器人中心）
Online Monocular Lane Mapping Using Catmull-Rom Spline
The Hong Kong University of Science and Technology（香港科技大学）；HKUST-DJI Joint Innovation Laboratory（香港科技大学-大疆创新联合创新实验室）；Hong Kong Center for Construction Robotics（香港建造机器人中心）
