Online Object-Oriented Semantic Mapping in Triger Classification Environment

Abstract

Creating and maintaining an accurate representation of the environment is an essential capability for every mobile robot. Especially, semantic information plays an important role in mobile robot navigation and other operations. Some scholars are studying different forms of expression of the environment, including geometric, semantic and other forms. In this paper, we present a semantic mapping framework. Our system is capable of online mapping and object updating given object detections from RGB-D data. The map can provides 2D representations, the locations and labels of the mapped objects. To undo wrong data association, we perform a judgment whether to intersect when updating object shapes. Furthermore, we maintain a merge step to deal with part of object detections and keep the map updated. Our mapping system is highly accurate and efficient. We evaluated our approach in the simulated triger classification environments using turtlebot3 robot. As the experimental results demonstrate, our system is able to generate maps that are close to the simulated environment.

This work was supported partially by the National Key Research and Development Program of China (2020YFB1313900), and partially by the Shezhen Science and Technology Program (No. JCYJ20180508152226630).