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

Fixed incrorrect rng seeding

Browse Source
This commit is contained in:
Dominik Moritz Roth 2022-06-22 13:08:23 +02:00
parent 2132deedcd
commit 7113190e35
1 changed files with 6 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
columbus/env.py
View File

@ -166,6 +166,7 @@ class ColumbusEnv(gym.Env):
def reset(self): def reset(self):
pygame.init() pygame.init()
self._seed(self.env_seed)
self._rendered = False self._rendered = False
self.inp = (0.5, 0.5) self.inp = (0.5, 0.5)
self.keypress_timeout = 0 self.keypress_timeout = 0
@ -178,7 +179,6 @@ class ColumbusEnv(gym.Env):
self.agent = entities.Agent(self) self.agent = entities.Agent(self)
self.setup() self.setup()
self.entities.append(self.agent) # add it last, will be drawn on top self.entities.append(self.agent) # add it last, will be drawn on top
self._seed(self.env_seed)
self.observable._entities = None self.observable._entities = None
return self.observable.get_observation() return self.observable.get_observation()
@ -206,6 +206,8 @@ class ColumbusEnv(gym.Env):
self.keypress_timeout = int(self.fps/2) self.keypress_timeout = int(self.fps/2)
if keys[pygame.K_m]: if keys[pygame.K_m]:
self.draw_entities = not self.draw_entities self.draw_entities = not self.draw_entities
elif keys[pygame.K_r]:
self.reset()
else: else:
self.keypress_timeout = 0 self.keypress_timeout = 0
else: else:
@ -361,7 +363,7 @@ class ColumbusJustState(ColumbusEnv):
class ColumbusStateWithBarriers(ColumbusEnv): class ColumbusStateWithBarriers(ColumbusEnv):
def __init__(self, observable=observables.StateObservable(coordsAgent=True, speedAgent=False, coordsRelativeToAgent=False, coordsRewards=True, rewardsWhitelist=None, coordsEnemys=True, enemysWhitelist=None, enemysNoBarriers=True, rewardsTimeouts=False, include_rand=True), fps=30, env_seed=3.1): def __init__(self, observable=observables.StateObservable(coordsAgent=True, speedAgent=False, coordsRelativeToAgent=False, coordsRewards=True, rewardsWhitelist=None, coordsEnemys=True, enemysWhitelist=None, enemysNoBarriers=True, rewardsTimeouts=False, include_rand=True), fps=30, env_seed=3.141):
super(ColumbusStateWithBarriers, self).__init__( super(ColumbusStateWithBarriers, self).__init__(
observable=observable, fps=fps, env_seed=env_seed) observable=observable, fps=fps, env_seed=env_seed)
self.aux_reward_max = 0.01 self.aux_reward_max = 0.01
@ -373,9 +375,9 @@ class ColumbusStateWithBarriers(ColumbusEnv):
enemy = entities.CircleBarrier(self) enemy = entities.CircleBarrier(self)
enemy.radius = self.random()*25+75 enemy.radius = self.random()*25+75
self.entities.append(enemy) self.entities.append(enemy)
for i in range(3): for i in range(1):
enemy = entities.FlyingChaser(self) enemy = entities.FlyingChaser(self)
enemy.chase_acc = self.random()*0.4+0.3 # *0.6+0.5 enemy.chase_acc = 0.55 # *0.6+0.5
self.entities.append(enemy) self.entities.append(enemy)
for i in range(1): for i in range(1):
reward = entities.TeleportingReward(self) reward = entities.TeleportingReward(self)