[rllib] Upgrade to OpenAI Gym 0.10.3 (#1601)

python/ray/rllib/examples/multiagent_mountaincar_env.py

@@ -1,4 +1,4 @@
-import math
+from math import cos
 from gym.spaces import Box, Tuple, Discrete
 import numpy as np
 from gym.envs.classic_control.mountain_car import MountainCarEnv
@@ -23,17 +23,17 @@ class MultiAgentMountainCarEnv(MountainCarEnv):
 
         self.action_space = [Discrete(3) for _ in range(2)]
         self.observation_space = Tuple([
-            Box(self.low, self.high) for _ in range(2)])
+            Box(self.low, self.high, dtype=np.float32) for _ in range(2)])
 
-        self._seed()
+        self.seed()
         self.reset()
 
-    def _step(self, action):
+    def step(self, action):
         summed_act = 0.5 * np.sum(action)
 
         position, velocity = self.state
         velocity += (summed_act - 1) * 0.001
-        velocity += math.cos(3 * position) * (-0.0025)
+        velocity += cos(3 * position) * (-0.0025)
         velocity = np.clip(velocity, -self.max_speed, self.max_speed)
         position += velocity
         position = np.clip(position, self.min_position, self.max_position)
@@ -47,6 +47,6 @@ class MultiAgentMountainCarEnv(MountainCarEnv):
         self.state = (position, velocity)
         return [np.array(self.state) for _ in range(2)], reward, done, {}
 
-    def _reset(self):
+    def reset(self):
         self.state = np.array([self.np_random.uniform(low=-0.6, high=-0.4), 0])
         return [np.array(self.state) for _ in range(2)]

python/ray/rllib/examples/multiagent_pendulum_env.py

@@ -22,18 +22,20 @@ class MultiAgentPendulumEnv(PendulumEnv):
 
         high = np.array([1., 1., self.max_speed])
         self.action_space = [Box(low=-self.max_torque / 2,
-                                 high=self.max_torque / 2, shape=(1,))
+                                 high=self.max_torque / 2,
+                                 shape=(1,),
+                                 dtype=np.float32)
                              for _ in range(2)]
         self.observation_space = Tuple([
-            Box(low=-high, high=high) for _ in range(2)])
+            Box(low=-high, high=high, dtype=np.float32) for _ in range(2)])
 
-        self._seed()
+        self.seed()
 
-    def _seed(self, seed=None):
+    def seed(self, seed=None):
         self.np_random, seed = seeding.np_random(seed)
         return [seed]
 
-    def _step(self, u):
+    def step(self, u):
         th, thdot = self.state  # th := theta
 
         summed_u = np.sum(u)
@@ -55,7 +57,7 @@ class MultiAgentPendulumEnv(PendulumEnv):
         self.state = np.array([newth, newthdot])
         return self._get_obs(), -costs, False, {}
 
-    def _reset(self):
+    def reset(self):
         high = np.array([np.pi, 1])
         self.state = self.np_random.uniform(low=-high, high=high)
         self.last_u = None