path-finding

This is a repository for a class project in reinforcement learning.

Contributors:
Zoe Kanavas (zkanavas@ucdavis.edu)
Erin Musabandesu (enmusabandesu@ucdavis.edu)
Liam Lynch (wdlynch@ucdavis.edu)

Data can be found here:

UC Davis Google Drive Data Access

Run the file RL_testbed_final.py from the same directory as the following:

1. Sample_A (data folder)
2. heuristic_info_all_samples.csv

Creates a folder for results (pickled dictionary) and figs based on a trial number. The trial number is also taken as the random seed.

Two algorithms implemented:

1. Episodic semigradient SARSA (Sutton and Barto, pg. 244) - with linear approximation function
2. Continuous semigradient SARSA (Sutton and Barto, pg. 251) - with linear approximation function