Home | Accessing Simulation Outputs | Model Parameters | Sampling Custom Data
By default, simulation runs executed via run.py will save output data and additional graphical outputs
in \mesa_spatial_sampling_MRS\results\default\1\.
Within this directory, 1.csv contains data on the performance of the simulated multi-robot system
recorded at each step of the simulated sampling mission.
This file and the directory it resides in are named 1 after the iteration number of the simulated run,
used as a seed for randomised variables such as the starting positions of the robots.
Visual outputs are also saved in this folder. The names of these are prefixed with the simulation
time step at which they were generated. These are figures generated with Matplotlib at every x steps
of the simulation. The frequency of figure generation is controlled by the vis_freq parameter of the
SpatialSamplingModel class, which can be changed in the model_params dictionary in server.py for running the
simulation with the web UI, or in the fixed_parameters dictionary argument of the BatchRunnerMP constructor in
BatchRunner_RR_and_SSI.py for running headless batches of simulated experiments.
The following figures are generated:
Saved as x_Interpolation.png, where x is the time step at which the figure was generated. These values are
soil compaction in KPa.
Saved as x_Variance.png, where x is the time step at which the figure was generated. Kriging
variance is a per-cell measure of the uncertainty of Kriging interpolation's prediction of the soil compaction value.
Saved as x_<robot_id>_visited_cells.png, where x is the time step at which the figure was generated, and <robot_id> is
the ID of the robot, generated from its initial position. The robot's trajectory and sampling points are overlaid on
top of the heatmap of visited cells.
_visited_cells.png "A heatmap of a single robot's visited cells, overlaid with its trajectory and sampling points.")
Saved as x_combined_visited_cells.png, where x is the time step at which the figure was
generated. The trajectories and sampling points of all robots are overlaid on top of a
heatmap of the number of visits made to each cell by the whole robot team.
The 'animate_graphical_output.py' script can be used to
generate animated GIFs of the above figures. Just change the value of path_to_png_files to the path containing the
png images generated from the simulation, e.g. "results/default/1/". To animate the figures generated from a batch of
simulations, run 'animate_batchrunner_vis.py'. You may need to
modify the script to call the animate_batch function, providing the path to the batch's outputs, and the number of
trials you wish to animate, e.g. animate_batch("results/RR_DS_3robs_20x20/", 10).