Welcome to the ARGoS multi-robot slam simulation. The goal of this simulation is to showcase our distributed implementation of the Gauss-Seidel pose graph optimization technique combined with our distributed inter-robot loop closures outliers rejection technique.
Since the setup is quite complex we provide a Docker Image
lajoiepy/doorslam-argos-simulation to easily run the simulation. Although we
don't recommend it, you can look at the Dockerfile for the setup
instructions.
The following instructions indicate how to start the simulation.
Designate a folder ( HOST_FOLDER ) to share files between the host machine and
the Docker container. The simulation results and logs will be store in this location.
Create a folder named datasets inside the HOST_FOLDER.
Run the following commands:
sudo x11docker --hostdisplay --hostnet --user=RETAIN -- --privileged -v HOST_FOLDER:/home/docker/robust_distributed_slam_simulation/argos_simulation/log -- lajoiepy/doorslam-argos-simulation:latest
Open a shell in the container and run the simulation
docker exec -it $(docker container ls -q) /bin/bash
cd ../home/docker/robust_distributed_slam_simulation/argos_simulation/
argos3 -c robust_distributed_slam_random_walk.argos
Press the simulator play button.
Start a shell in the container and run the following command that will start VSCode.
code --user-data-dir
More details on how to build the image are available here.