columbus/env.py

@@ -4,9 +4,7 @@ from gym import spaces
 import numpy as np
 import pygame
 import random as random_dont_use
-from os import urandom
 import math
-from time import time as get_time
 from columbus import entities, observables
 
 
@@ -46,14 +44,13 @@ class ColumbusEnv(gym.Env):
         self.void_damage = 100
 
         self.rng = random_dont_use.Random()
-        self._seed(self.env_seed)
         self.reset()
 
         self.observation_space = self.observable.get_observation_space()
 
     def _seed(self, seed):
         if seed == None:
-            seed = urandom(12)
+            seed = random_dont_use.random()
         self.rng.seed(seed)
 
     def random(self):
@@ -168,7 +165,6 @@ class ColumbusEnv(gym.Env):
         pygame.init()
         self._rendered = False
         self.inp = (0.5, 0.5)
-        self.keypress_timeout = 0
         # will get rescaled acording to fps (=reward per second)
         self.new_reward = 0
         self.new_abs_reward = 0  # will not get rescaled. should be used for one-time rewards
@@ -198,18 +194,6 @@ class ColumbusEnv(gym.Env):
                                                               self.joystick_offset[0], 20+int(60*y)+self.joystick_offset[1]), 20, width=0)
 
     def render(self, mode='human', dont_show=False):
-        for event in pygame.event.get():
-            pass
-        keys = pygame.key.get_pressed()
-        if self.keypress_timeout == 0:
-            self.keypress_timeout = int(self.fps/2)
-            if keys[pygame.K_m]:
-                self.draw_entities = not self.draw_entities
-            else:
-                self.keypress_timeout = 0
-        else:
-            self.keypress_timeout -= 1
-
         self.visible = self.visible or not dont_show
         self._ensure_surface()
         pygame.draw.rect(self.surface, (0, 0, 0),
@@ -274,9 +258,9 @@ class ColumbusRayDrone(ColumbusTestRay):
 
 
 class ColumbusCandyland(ColumbusEnv):
-    def __init__(self, observable=observables.RayObservable(chans=[entities.Reward, entities.Void], num_rays=16, include_rand=True), hide_map=False, fps=30, env_seed=None):
+    def __init__(self, observable=observables.RayObservable(chans=[entities.Reward, entities.Void], num_rays=16, include_rand=True), hide_map=False, fps=30):
         super(ColumbusCandyland, self).__init__(
-            observable=observable,  fps=fps, env_seed=env_seed)
+            observable=observable,  fps=fps)
         self.draw_entities = not hide_map
 
     def setup(self):
@@ -294,7 +278,7 @@ class ColumbusCandyland(ColumbusEnv):
 class ColumbusEasyObstacles(ColumbusEnv):
     def __init__(self, observable=observables.RayObservable(num_rays=16), hide_map=False, fps=30, env_seed=None):
         super(ColumbusEasyObstacles, self).__init__(
-            observable=observable,  fps=fps, env_seed=env_seed)
+            observable=observable,  fps=fps)
         self.draw_entities = not hide_map
         self.aux_reward_max = 0.1
 
@@ -310,31 +294,7 @@ class ColumbusEasyObstacles(ColumbusEnv):
             self.entities.append(reward)
         for i in range(1):
             enemy = entities.WalkingChaser(self)
-            enemy.chase_speed = 0.20
+            enemy.chase_speed = 0.1
-            self.entities.append(enemy)
-
-
-class ColumbusEasierObstacles(ColumbusEnv):
-    def __init__(self, observable=observables.RayObservable(num_rays=16), hide_map=False, fps=30, env_seed=None):
-        super(ColumbusEasierObstacles, self).__init__(
-            observable=observable,  fps=fps, env_seed=env_seed)
-        self.draw_entities = not hide_map
-        self.aux_reward_max = 0.5
-
-    def setup(self):
-        self.agent.pos = self.start_pos
-        for i in range(5):
-            enemy = entities.CircleBarrier(self)
-            enemy.radius = 30 + self.random()*70
-            self.entities.append(enemy)
-        for i in range(3):
-            reward = entities.TeleportingReward(self)
-            reward.radius = 30
-            reward.reward *= 2
-            self.entities.append(reward)
-        for i in range(1):
-            enemy = entities.WalkingChaser(self)
-            enemy.chase_speed = 0.20
             self.entities.append(enemy)
 
 
@@ -412,9 +372,3 @@ register(
     entry_point=ColumbusEasyObstacles,
     max_episode_steps=30*60*2,
 )
-
-register(
-    id='ColumbusEasierObstacles-v0',
-    entry_point=ColumbusEasyObstacles,
-    max_episode_steps=30*60*2,
-)

columbus/humanPlayer.py

@@ -7,8 +7,8 @@ from observables import Observable, CnnObservable
 
 
 def main():
-    #env = ColumbusTest3_1(fps=30)
+    #env = ColumbusTest3_1()
-    env = ColumbusEasierObstacles(fps=30)
+    env = ColumbusEasyObstacles(fps=30)
     env.start_pos = [0.6, 0.3]
     playEnv(env)
     env.close()
@@ -22,6 +22,17 @@ def playEnv(env):
         t1 = time()
         env.render()
         pos = (0.5, 0.5)
+        for event in pygame.event.get():
+            pass
+        keys = pygame.key.get_pressed()
+        if to == 0:
+            to = int(env.fps/2)
+            if keys[pygame.K_m]:
+                env.draw_entities = not env.draw_entities
+            else:
+                to = 0
+        else:
+            to -= 1
         pos = pygame.mouse.get_pos()
         pos = (min(max((pos[0]-env.joystick_offset[0]-20)/60, 0), 1),
                min(max((pos[1]-env.joystick_offset[1]-20)/60, 0), 1))