some new stuff

alr_envs/__init__.py

@@ -3,7 +3,58 @@ from gym.envs.registration import register
 register(
     id='ALRReacher-v0',
     entry_point='alr_envs.mujoco:ALRReacherEnv',
-    max_episode_steps=1000,
+    max_episode_steps=200,
+    kwargs={
+        "steps_before_reward": 0,
+    }
+)
+
+register(
+    id='ALRReacher100-v0',
+    entry_point='alr_envs.mujoco:ALRReacherEnv',
+    max_episode_steps=200,
+    kwargs={
+        "steps_before_reward": 100,
+    }
+)
+
+register(
+    id='ALRReacher180-v0',
+    entry_point='alr_envs.mujoco:ALRReacherEnv',
+    max_episode_steps=200,
+    kwargs={
+        "steps_before_reward": 180,
+    }
+)
+
+register(
+    id='ALRReacher7-v0',
+    entry_point='alr_envs.mujoco:ALRReacherEnv',
+    max_episode_steps=200,
+    kwargs={
+        "steps_before_reward": 0,
+        "n_links": 7,
+    }
+)
+
+register(
+    id='ALRReacher100_7-v0',
+    entry_point='alr_envs.mujoco:ALRReacherEnv',
+    max_episode_steps=200,
+    kwargs={
+        "steps_before_reward": 100,
+        "n_links": 7,
+    }
+)
+
+register(
+    id='ALRReacher180_7-v0',
+    entry_point='alr_envs.mujoco:ALRReacherEnv',
+    max_episode_steps=200,
+    kwargs={
+        "steps_before_reward": 180,
+        "n_links": 7,
+    }
 )
 
 register(

alr_envs/classic_control/simple_reacher.py

@@ -31,7 +31,7 @@ class SimpleReacherEnv(gym.Env):
         self._angle_velocity = None
 
         self.max_torque = 1  # 10
-        self.steps_before_reward = 100
+        self.steps_before_reward = 180
 
         action_bound = np.ones((self.n_links,))
         state_bound = np.hstack([
@@ -69,7 +69,7 @@ class SimpleReacherEnv(gym.Env):
 
     def _add_action_noise(self, action: np.ndarray):
         """
-        add unobserved Gaussian Noise N(0,0.5) to the actions
+        add unobserved Gaussian Noise N(0,0.01) to the actions
         Args:
             action:
 

example.py

@@ -1,15 +1,20 @@
+import time
+
 import gym
 
 if __name__ == '__main__':
 
-    # env = gym.make('alr_envs:ALRReacher-v0')
+    env = gym.make('alr_envs:ALRReacher-v0')
-    env = gym.make('alr_envs:SimpleReacher-v0')
+    # env = gym.make('alr_envs:SimpleReacher-v0')
+    # env = gym.make('alr_envs:ALRReacher7-v0')
     state = env.reset()
 
     for i in range(10000):
         state, reward, done, info = env.step(env.action_space.sample())
-        if i % 5 == 0:
+        if i % 1 == 0:
             env.render()
 
-        if done:
+        # if done:
-            state = env.reset()
+        state = env.reset()
+
+        time.sleep(0.5)