wassname/DeepRL
1
0
Fork 0
mirror of https://github.com/wassname/DeepRL.git synced 2026-09-09 11:13:47 +08:00
Code Issues Packages Projects Releases Wiki Activity

Support roboschool

Browse Source
This commit is contained in:
Shangtong Zhang
2017-10-27 14:23:54 -06:00
parent e832e2e89a
commit af360bb9ea
2 changed files with 40 additions and 16 deletions
Download Patch File Download Diff File Expand all files Collapse all files
component/task.py
+28 -5
View File
@@ -7,6 +7,10 @@ import gym
import sys
import numpy as np
from .atari_wrapper import *
try:
import roboschool
except:
gym.logger.info('Roboschool not found')
class BasicTask:
def __init__(self):
@@ -79,11 +83,11 @@ class PixelAtari(BasicTask):
class ContinuousMountainCar(BasicTask):
name = 'MountainCarContinuous-v0'
success_threshold = 90
default_max_episode = 999
def __init__(self):
BasicTask.__init__(self)
self.env = gym.make(self.name)
self.max_episode_steps = self.env._max_episode_steps
self.env._max_episode_steps = sys.maxsize
self.action_dim = self.env.action_space.shape[0]
self.state_dim = self.env.observation_space.shape[0]
@@ -97,11 +101,11 @@ class ContinuousMountainCar(BasicTask):
class Pendulum(BasicTask):
name = 'Pendulum-v0'
success_threshold = -10
default_max_episode = 200
def __init__(self):
BasicTask.__init__(self)
self.env = gym.make(self.name)
self.max_episode_steps = self.env._max_episode_steps
self.env._max_episode_steps = sys.maxsize
self.action_dim = self.env.action_space.shape[0]
self.state_dim = self.env.observation_space.shape[0]
@@ -114,11 +118,11 @@ class Pendulum(BasicTask):
class BipedalWalker(BasicTask):
name = 'BipedalWalker-v2'
success_threshold = 300
default_max_episode = 999
def __init__(self):
BasicTask.__init__(self)
self.env = gym.make(self.name)
self.max_episode_steps = self.env._max_episode_steps
self.env._max_episode_steps = sys.maxsize
self.action_dim = self.env.action_space.shape[0]
self.state_dim = self.env.observation_space.shape[0]
@@ -131,11 +135,11 @@ class BipedalWalker(BasicTask):
class BipedalWalkerHardcore(BasicTask):
name = 'BipedalWalkerHardcore-v2'
success_threshold = 300
default_max_episode = 2000
def __init__(self):
BasicTask.__init__(self)
self.env = gym.make(self.name)
self.max_episode_steps = self.env._max_episode_steps
self.env._max_episode_steps = sys.maxsize
self.action_dim = self.env.action_space.shape[0]
self.state_dim = self.env.observation_space.shape[0]
@@ -148,11 +152,30 @@ class BipedalWalkerHardcore(BasicTask):
class ContinuousLunarLander(BasicTask):
name = 'LunarLanderContinuous-v2'
success_threshold = 300
default_max_episode = 1000
def __init__(self):
BasicTask.__init__(self)
self.env = gym.make(self.name)
self.max_episode_steps = self.env._max_episode_steps
self.env._max_episode_steps = sys.maxsize
self.action_dim = self.env.action_space.shape[0]
self.state_dim = self.env.observation_space.shape[0]
def step(self, action):
action = np.clip(action, -1, 1)
next_state, reward, done, info = self.env.step(action)
return next_state, reward, done, info
class Roboschool(BasicTask):
def __init__(self, name, success_threshold=sys.maxsize, max_episode_steps=None):
BasicTask.__init__(self)
self.name = name
self.env = gym.make(self.name)
self.success_threshold = success_threshold
if max_episode_steps is None:
self.max_episode_steps = self.env._max_episode_steps
else:
self.max_episode_steps = max_episode_steps
self.env._max_episode_steps = sys.maxsize
self.action_dim = self.env.action_space.shape[0]
self.state_dim = self.env.observation_space.shape[0]
main.py
+12 -11
View File
@@ -213,7 +213,7 @@ def a3c_continuous():
config.policy_fn = lambda: GaussianPolicy()
config.worker = ContinuousAdvantageActorCritic
config.discount = 0.99
config.max_episode_length = task.default_max_episode
config.max_episode_length = task.max_episode_steps
config.num_workers = 8
config.update_interval = 20
config.test_interval = 1
@@ -226,8 +226,9 @@ def a3c_continuous():
def dppo_continuous():
config = Config()
# config.task_fn = lambda: Pendulum()
config.task_fn = lambda: BipedalWalkerHardcore()
config.task_fn = lambda: Pendulum()
# config.task_fn = lambda: BipedalWalkerHardcore()
# config.task_fn = lambda: Roboschool('RoboschoolInvertedPendulum-v1')
task = config.task_fn()
config.actor_network_fn = lambda: GaussianActorNet(task.state_dim, task.action_dim,
gpu=False, unit_std=True)
@@ -244,9 +245,9 @@ def dppo_continuous():
config.num_workers = 8
config.test_interval = 1
config.test_repetitions = 1
config.max_episode_length = task.default_max_episode
config.max_episode_length = task.max_episode_steps
config.entropy_weight = 0
config.gradient_clip = 40
config.gradient_clip = 20
config.rollout_length = 10000
config.optimize_epochs = 1
config.ppo_ratio_clip = 0.2
@@ -255,10 +256,10 @@ def dppo_continuous():
agent.run()
def ddpg_continuous():
task_fn = lambda: Pendulum()
task = task_fn()
config = Config()
config.task_fn = task_fn
# config.task_fn = lambda: Pendulum()
config.task_fn = lambda: Roboschool('RoboschoolInvertedPendulum-v1')
task = config.task_fn()
config.actor_network_fn = lambda: DeterministicActorNet(
task.state_dim, task.action_dim, F.tanh, 2, non_linear=F.relu, batch_norm=False)
config.critic_network_fn = lambda: DeterministicCriticNet(
@@ -269,7 +270,7 @@ def ddpg_continuous():
lambda params: torch.optim.Adam(params, lr=1e-3, weight_decay=0.01)
config.replay_fn = lambda: HighDimActionReplay(memory_size=1000000, batch_size=64)
config.discount = 0.99
config.max_episode_length = task.default_max_episode
config.max_episode_length = task.max_episode_steps
config.target_network_mix = 0.001
config.exploration_steps = 100
config.noise_decay_interval = 10000
@@ -289,8 +290,8 @@ if __name__ == '__main__':
# async_cart_pole()
# a3c_cart_pole()
# a3c_continuous()
# dppo_continuous()
ddpg_continuous()
dppo_continuous()
# ddpg_continuous()
# dqn_fruit()
# hrdqn_fruit()