dodox/Columbus
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

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

Browse Source
This commit is contained in:
Dominik Moritz Roth 2022-07-19 10:05:28 +02:00
parent 54adebfe9b
commit a9b53b6eca
1 changed files with 4 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
columbus/env.py
View File

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