Hierarchical, Dense and Dynamic 3D Reconstruction Based on VDB Data Structure for Robotic Manipulation Tasks

This paper presents a novel approach to implement hierarchical, dense and dynamic reconstruction of 3D objects based on the VDB (Variational Dynamic B + Trees) data structure for robotic applications. The scene is done by integration depth-images using Truncated Signed Distance Field (TSDF). proposed method trees in order provide similar results current state-of-the-art methods (i.e., complete volumes, hashing voxels hierarchical volumes) terms execution time but with direct multi-level representation that remains real-time. provides two major advantages: it unbounded space representation. optimally implemented be used GPU architecture, exploiting parallelism skills this hardware. A series experiments will presented prove performance robot arm platform.