Skip to content

Performance comparison of NearestSearch and radiusSearch function in point cloud.

Notifications You must be signed in to change notification settings

LimHyungTae/OpenFLANN

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

8 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

OpenFLANN

I found PicoFLANN and compare the speed of alorithms, namely PCL FLANN, NanoFLANN, and PicoFLANN.

Characteristics

  • Set NanoFLANN and PicoFLANN to be avaiable on Point Cloud Libarary (PCL).
  • It's simple to use them! Just git pull this repository, then copy & paste the header files in include folder.
  • Usage is totally same with Point Cloud Library, i.e., setInputCloud, nearestSearch, or radiusSearch.
  • Many robotics guys refer to NanoFLANN in LOAM; however, the function radiusSearch in LOAM is actually not in use so the function does not work properly. So, I debugged it and revise the radiusSearch.

Reference

  • PCL FLANN
  • NanoFLANN
  • PicoFLANN (The original developer says it is faster than NanoFLANN, and it actually is when the scale of points becomes larger)

Simulation results

Experiments: Please refer to src/main.cpp. I reran the experiment 1,000 times to measure mean speed of each case.

One-line summary: For robotics application, NanoFLANN is better than other FLANN implementations.

You can show the results via python code (tabulate is necessary)

cd outputs
python viz_output.py

K-Nearest Neighbor

00_knn

01_knn

02_knn

Radius Search

00_raw

01_raw

02_raw

About

Performance comparison of NearestSearch and radiusSearch function in point cloud.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published