fix new minitaur envs for newer Gym (_step -> step etc),

add extra sleep to sync Minitaur walk with RTC clock

examples/pybullet/gym/pybullet_envs/minitaur/envs/minitaur_gym_env.py

@@ -17,6 +17,7 @@ import minitaur_logging
 import minitaur_logging_pb2
 import minitaur_rainbow_dash
 import motor
+from pkg_resources import parse_version
 
 NUM_MOTORS = 8
 MOTOR_ANGLE_OBSERVATION_INDEX = 0
@@ -563,6 +564,13 @@ class MinitaurGymEnv(gym.Env):
     """
     return len(self._get_observation())
 
+  if parse_version(gym.__version__)>=parse_version('0.9.6'):
+                close = _close
+                render = _render
+                reset = _reset
+                seed = _seed
+                step = _step
+
   def set_time_step(self, control_step, simulation_step=0.001):
     """Sets the time step of the environment.
 
@@ -604,3 +612,5 @@ class MinitaurGymEnv(gym.Env):
   @property
   def env_step_counter(self):
     return self._env_step_counter
+
+

examples/pybullet/gym/pybullet_envs/minitaur/envs/minitaur_gym_env_example.py

@@ -9,6 +9,7 @@ import argparse
 import numpy as np
 import tensorflow as tf
 import minitaur_gym_env
+import time
 
 
 #FLAGS = flags.FLAGS
@@ -53,6 +54,7 @@ def ResetPoseExample(log_path=None):
   action = [math.pi / 2] * 8
   for _ in range(steps):
     _, _, done, _ = environment.step(action)
+    time.sleep(1./100.)
     if done:
       break
 
@@ -96,6 +98,7 @@ def MotorOverheatExample(log_path=None):
     observation, _, _, _ = environment.step(action)
     current_row.extend(observation.tolist())
     actions_and_observations.append(current_row)
+    time.sleep(1./100.)
 
   if FLAGS.output_filename is not None:
     WriteToCSV(FLAGS.output_filename, actions_and_observations)
@@ -142,6 +145,7 @@ def SineStandExample(log_path=None):
     observation, _, _, _ = environment.step(action)
     current_row.extend(observation.tolist())
     actions_and_observations.append(current_row)
+    time.sleep(1./100.)
 
   if FLAGS.output_filename is not None:
     WriteToCSV(FLAGS.output_filename, actions_and_observations)
@@ -189,6 +193,8 @@ def SinePolicyExample(log_path=None):
     a4 = math.sin(t * speed + math.pi) * amplitude2
     action = [a1, a2, a2, a1, a3, a4, a4, a3]
     _, reward, done, _ = environment.step(action)
+    time.sleep(1./100.)
+
     sum_reward += reward
     if done:
       tf.logging.info("Return is {}".format(sum_reward))