Lowered Aux-Reward (now per-second instead of per frame)

This commit is contained in:
Dominik Moritz Roth 2022-07-19 10:05:28 +02:00
parent 54adebfe9b
commit a9b53b6eca

View File

@ -123,7 +123,7 @@ class ColumbusEnv(gym.Env):
self.aux_reward_discretize / 2
aux_reward -= penalty
return aux_reward
return aux_reward/self.fps
def step(self, action):
# TODO: Just make the range consistent...
@ -381,7 +381,7 @@ class ColumbusCandyland(ColumbusEnv):
class ColumbusCandyland_Aux10(ColumbusCandyland):
def __init__(self, fps=30):
super(ColumbusCandyland_Aux10, self).__init__(fps=fps)
self.aux_reward_max = 10
self.aux_reward_max = 1
class ColumbusEasyObstacles(ColumbusEnv):
@ -389,7 +389,7 @@ class ColumbusEasyObstacles(ColumbusEnv):
super(ColumbusEasyObstacles, self).__init__(
observable=observable, fps=fps, env_seed=env_seed)
self.draw_entities = not hide_map
self.aux_reward_max = 0.1
self.aux_reward_max = 1
def setup(self):
self.agent.pos = self.start_pos
@ -412,7 +412,7 @@ class ColumbusEasierObstacles(ColumbusEnv):
super(ColumbusEasierObstacles, self).__init__(
observable=observable, fps=fps, env_seed=env_seed)
self.draw_entities = not hide_map
self.aux_reward_max = 0.5
self.aux_reward_max = 1
def setup(self):
self.agent.pos = self.start_pos