# grab map information
mission_xml = BeautifulSoup(env.params['mission_xml'], features="xml")
map_spec = mission_xml.find('specification')
placement = mission_xml.find('Placement')
map_dimension = [
int(map_spec.contents[1].text),
int(map_spec.contents[2].text),
int(map_spec.contents[3].text)
]
mission_available_moves = env.params['comp_all_commands']
num_episodes = 300
gamma = [1, .6, .3]
alpha = [1, .6, .3]
max_simulation_time = 120
# Input learning method
# MC - monte carlo, Q - Q learning
algorithm = 'Q'
for g in gamma:
for a in alpha:
if algorithm == 'MC':
# instantiate an Agent object
mc = MonteCarlo(mission_name, env, num_episodes, g,
max_simulation_time, a)
mc.mc_prediction(filename='', iteration_number=0)
elif algorithm == 'Q':
# instantiate an Agent object
q = Q(mission_name, env, num_episodes, g, a, max_simulation_time)
q.q_prediction()
