MarineRoboticsGroup/lcgp

Planning for multi-robot networks localizing via range measurements.

Prioritized, Localizability-Constrained Graph Planning

This repo contains code for path-planning of 2-dimensional range-only multi-robot networks.

Getting started

To run some sample code:

conda create -y --name lcgp python=3.9 numpy scipy matplotlib
conda activate lcgp
conda install numba
conda install -c conda-forge cvxpy
conda install -c conda-forge chaospy

# run the sample code from the base directory of this repository
python trial.py

Useful for debugging

If the planner is having issues planning and it is unclear if it's a bug or if it genuinely is unable to find paths, inside the prioritized_prm.py file we have the ability to plot the trajectories for all robots [0,..,n-1] and show the state of all of the points on the PRM (which set they belong to, if any) of robot n. As default there is a if False: branch in the perform_planning function, to turn this plotting on change False to True. This will plot after each robot plans its path (remember, prioritized planning so the robots plan independently in sequence). The full code looks like this:

if False:
    self.constraintSets.animate_valid_states(
        self._coord_trajs, cur_robot_id + 1
    )