Abstract
Mobile robots are useful for environment exploration and rescue operations. In such applications, it is crucial to accurately analyse and represent an environment, providing appropriate inputs for motion planning in order to support robot navigation and operations. 2D mapping methods are simple but cannot handle multilevel or multistory environments. To address this problem, 3D mapping methods generate structural 3D representations of the robot operating environment and its objects by 3D mesh reconstruction. However, they face the challenge of efficiently transmitting those 3D representations to system modules for 3D mapping, motion planning, and robot operation visualization. This paper proposes a quality-driven mesh compression and transmission method to address this. Our method is efficient, as it compresses a mesh by quantizing its transformed vertices without the need to spend time constructing an a-priori structure over the mesh. A visual distortion function is developed to govern the level of quantization, allowing mesh transmission to be controlled under different network conditions or time constraints. Our experiments demonstrate how the visual quality of a mesh can be manipulated by the visual distortion function.
Highlights
Mapping has been a very active research area in Simultaneous Localization and Mapping (SLAM) [1] for robotic applications. 3D mapping and reconstruction [2, 3] produce structural 3D geometric representations for robot operating environments and their objects, providing comprehensive inputs for motion planning to enhance results and guide robot navigation and operations more appropriately
We have developed a simple and quality-driven mesh compression and transmission method, which does not require an a-priori structure to be built on top of a mesh
We found that visual distortion error (VDE) is both influenced by the number of anchors introduced to a reconstructed mesh and the number of bits used for quantizing a mesh
Summary
Mapping has been a very active research area in Simultaneous Localization and Mapping (SLAM) [1] for robotic applications. 3D mapping and reconstruction [2, 3] produce structural 3D geometric representations for robot operating environments and their objects, providing comprehensive inputs for motion planning to enhance results and guide robot navigation and operations more appropriately. One popular representation of such 3D mapping is meshes, which may entail large data size, causing performance issues in passing environment and object representations to system modules for 3D mapping, motion planning, and robot operation visualization Such representations may require continuous updating when a robot is exploring an un‐ known or dynamic environment. There exist many mesh compression and trans‐ mission methods [5, 6, 7] that seek to address the above challenges, most of these require structures to be built on top of 3D meshes before they can be transmitted This is not favourable to robotic applications, where the environment and object representations may be dynamically generated and updated during run-time. The requirement of constructing a-priori structures will repeatedly cause delays in obtaining timely environment and object representations, limiting capacity to produce motion plans or support multi-robot collaboration
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
