roboweldar-3d-reconstruction

This is the 3D reconstruction module of the RoboWeldAR project. It takes as input a series of images and outputs a 3D mesh of the reconstructed model expressed in the world coordinate frame.

The module can be called independently, as a standalone entity, or started as a client and spoken to via websocket and http.

The module can be called in web client mode, or as a standalone script.

Getting Started

These instructions will get you a copy of the project up and running on your local machine for development and testing purposes. See deployment for notes on how to deploy the project on a live system.

Steps:

• git clone --recursive git@github.com:ikh-innovation/roboweldar-3d-reconstruction.git
• cd roboweldar-3d-reconstruction
• chmod a+x setup.sh
• ./setup.sh
• python3.6 -m venv .venv
• source .venv/bin/activate
• pip install -r requirements.txt
• pip install -r src/rest/roboweldar_networking/interfaces/requirements.txt

Running the client

The module is able to talk to the main RoboWeldAR server instance, and receive commands delegated by the end user through the RoboWeldAR UI. To run the 3D reconstruction as a service, run:

• cd roboweldar-3d-reconstruction
• source .venv/bin/activate
• export PYTHONPATH=$PYTHONPATH:$(pwd); python src/rest/client.py --host "localhost"

The service will open a websocket port, listening to incoming messages from the server, perform a 3D reconstruction of the images sent through http, and return the reconstructed model to the server.

Running the standalone module

To run the standalone Meshroom wrapper, from

• cd roboweldar-3d-reconstruction
• source .venv/bin/activate
• export PYTHONPATH=$PYTHONPATH:$(pwd); python src/reconstruction/reconstruction.py

You can instantiate the Meshroom wrapper via ThreeDReconstruction(path_to_meshroom_root, path_to_images_dir, path_to_output_dir, path_to_cache_dir) where path_to_meshroom_root is the path to where the Meshroom binary files are located, path_to_images_dir is the path to the directory containing the raw images, path_to_output_dir is the path where the reconstructed model will be placed, path_to_cache_dir is the path where the Meshroom cache files are saved for one session.

Transforming the model to world coordinates

By default, the reconstructed model is placed in some coordinate frame that depends on the convergence of the Meshroom algorithm. To transform this model into the world frame, a script has been written, that uses the known camera positions provided by the robot, in the world frame, to perform a series of linear transformation (traslation, rotation, scaling) in order to bring the model into the world frame.

More info on usage is found in the associated README.md.

Running the tests

Explain how to run the automated tests for this system

Break down into end to end tests

under construction Explain what these tests test and why

Give an example

Deployment using Docker

under construction.

Authors

• Orfeas Kypris - Initial work - orphefs

See also the list of contributors who participated in this project.

License

This project is licensed under the MIT License - see the LICENSE.md file for details

Acknowledgments

• Props to the AliceVision Meshroom development community
• Props to my cats for being cool