Abstract:Highly accurate and efficient map in unknown and complex environments is essential for robotics navigation. Traditionally, mobile robot platforms are often computationally constrained when using multiple sensors to process large amounts of input data. In previous works, some of them have been deployed to embedded platforms in real-time. However, how to balance accuracy and efficiency while reducing the computational resources and the memory footprint is still the bottleneck. Motivated by these challenges, we proposed a mapping framework called VDBblox to incrementally build Euclidean Signed Distance Fields (ESDFs) map from Truncated Signed Distance Fields (TSDFs) mapping. We use a novel weight function to update the non-projective TSDFs, thus improving the quality of the mesh reconstruction with higher accuracy than up-to-date methods. Meanwhile, the generated ESDFs map is maintained by the least recently used (LRU) cache to dynamically handle the obstacle changes with less runtime than state-of-the-art. We show VDBblox performance in terms of accuracy and efficiency by benchmark comparison on RGB-D and LiDAR public datasets. Moreover, we demonstrate that VDBblox can be integrated into a completed quadrotor system as a sub-module. Then we validate it through online obstacle avoidance and high-quality mesh reconstruction in real-world experiments. Finally, we release our method as open-source code to the community(1).

Modified Polygonal Mesh Voxelization Based on Euclidean Distance Measurement

Polyhedron-bounded Collision Checks for Robotic Assembly of Structural Components

Adaptive Data Compression for Massive Scan Data

A Smart Voxelization Algorithm Zhao Dong Wei Chen Hujun Bao Hongxin Zhang Qunsheng Peng State Key

Error-Bounded solid voxelization for polygonal model based on heuristic seed filling

VDBblox: Accurate and Efficient Distance Fields for Path Planning and Mesh Reconstruction

Conservative voxelization

Automatic hexahedral mesh generation for many-to-one sweep volumes

Real-time Voxelization for Complex Models

Mesh Reconstruction Algorithm Based on Modulation Intensity

Interactive Hausdorff Distance Computation for General Polygonal Models.

Fast Large Scale Voxelization Using Projection Volume and Octree

Noise Removal Algorithm for Polygonal Meshes

A Unified Framework for Isotropic Meshing Based on Narrow-Band Euclidean Distance Transformation

Fast Voxelization of Triangulated Irregular Network Model Using Geometric Intersection Computation

Depth-Box VDB: Accelerate Sparse Volume Rendering with Depth Maps Through Voxel Database

High Resolution Solid Voxelization Accelerated by GPU

Fuzzy Clustering Based Pseudo-Swept Volume Decomposition for Hexahedral Meshing

Variational Hierarchical Directed Bounding Box Construction for Solid Mesh Models

Point-tessellated Voxelization.

A novel segmentation algorithm for complex 3D mesh model in computer vision.