Giter Site home page Giter Site logo

explain-icp-uncertainty's Introduction

Explain ICP Uncertainty

This codebase implements my Interdisciplinary Project (IDP) Explanation of Uncertainty in Point Cloud Registration at Technical University of Munich. It uses kernel SHAP to find the associations between uncertainty sources and estimated uncertainty in Iterative Closest Point (ICP) algorithms.

Installation

After creating a Python virtual environment using conda, pyenv virtualenv, or venv, do pip install -r requirements.txt.

Then, clone the fork of libpointmatcher by 3D ICP.

cd 3d-icp-cov
git clone https://github.com/CAOR-MINES-ParisTech/libpointmatcher.git
cd libpointmatcher
mkdir build && cd build
cmake ..
make
sudo make install
cd ../..

Data Preparation

First, then download global_frame and local_frame datasets from the Challenging datasets at .

Then, create a directory data/<sequence>/, and put the global_frame and local_frame datasets under data/<sequence>/global_frame/ and data/<sequence>/local_frame/0_0/.

Last, change variables lpm_path, dir_path and results_base of class Param in utils.py according to your paths.

After running ICP algorithms with different perturbations (see next section), the file structure would look like this:

├── data
│   ├── Apartment
│   │   ├── global_frame
│   │   │   ├── overlap_apartment.csv
│   │   │   ├── PointCloud<id>.csv
│   │   │   ├── pose_scanner_leica.csv
│   │   │   ├── xsens_Compass_mean.csv
│   │   │   ├── xsens_Gps_mean.csv
│   │   │   └── xsens_Gravity_mean.csv
│   │   ├── local_frame
│   │   │   ├── 0_0
│   │   │   │   ├── Hokuyo_<id>.csv
│   │   │   │   ├── <partial overlap>
│   │   │   │   │   ├── Hokuyo_<id>.csv
│   │   │   ├── <other sensor noise>
│   │   │   │   ├── <partial overlap>
│   │   │   │   │   ├── Hokuyo_<id>.csv

Run Pipeline

Both experiments in the paper can be run. Experiment 1 determines the relationship between perturbation levels of uncertainty sources and uncertainty estimate for point clouds scan_ref and scan_ref+1 in sequence Apartment. Experiment 2 determines the effects of uncertainty sources for contiguous point clouds in all sequences.

To run experiment 1, do:

python background.py <scan_ref> # run background dataset
python kernel_shap.py <scan_ref> # use kernel SHAP to generate plots

To run experiment 2, do:

python avg_parallel.py <seq> # the parallel version of average.py
python ks_avg_parallel.py <seq> # the parallel version of ks_avg.py
python plot_avg.py # visualize the inlier SHAP values

Note

One could set logger in /home/parallels/Desktop/idp/libpointmatcher/martin/config/base_config.yaml to NullLogger to avoid seeing output from ICP.

explain-icp-uncertainty's People

Contributors

targo-benchmark avatar franklinqin0 avatar

Watchers

 avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.