An Object SLAM Framework for Association, Mapping, and High-Level Tasks

Object SLAM is considered increasingly significant for robot high-level perception and decision-making. Existing studies fall short in terms of data association, object representation, semantic mapping frequently rely on additional assumptions, limiting their performance. In this paper, we present a comprehensive framework that focuses object-based object-oriented tasks. First, propose an ensemble association approach associating objects complicated conditions by incorporating parametric nonparametric statistic testing. addition, suggest outlier-robust centroid scale estimation algorithm modeling based the iForest line alignment. Then lightweight map represented estimated general models. Taking into consideration invariance objects, convert to topological provide descriptors enable multi-map matching. Finally, object-driven active exploration strategy achieve autonomous grasping scenario. A range public datasets real-world results mapping, augmented reality, scene matching, relocalization, robotic manipulation have been used evaluate proposed its efficient