"""
Proximal Policy Optimization (PPO)
----------------------------
A simple version of Proximal Policy Optimization (PPO) using single thread.
PPO is a family of first-order methods that use a few other tricks to keep new policies close to old.
PPO methods are significantly simpler to implement, and empirically seem to perform at least as well as TRPO.
Reference
---------
Proximal Policy Optimization Algorithms, Schulman et al. 2017
High Dimensional Continuous Control Using Generalized Advantage Estimation, Schulman et al. 2016
Emergence of Locomotion Behaviours in Rich Environments, Heess et al. 2017
MorvanZhou's tutorial page: https://morvanzhou.github.io/tutorials
MorvanZhou's code: https://github.com/MorvanZhou/Reinforcement-learning-with-tensorflow/
Prerequisites
--------------
tensorflow >=2.0.0a0
tensorflow-probability 0.6.0
tensorlayer >=2.0.0
"""
import time
from rlzoo.common.utils import *
from rlzoo.common.policy_networks import *
from rlzoo.common.value_networks import *
EPS = 1e-8 # epsilon
############################### PPO ####################################
[docs]class PPO_CLIP(object):
"""
PPO class
"""
def __init__(self, net_list, optimizers_list, epsilon=0.2):
"""
:param net_list: a list of networks (value and policy) used in the algorithm, from common functions or customization
:param optimizers_list: a list of optimizers for all networks and differentiable variables
:param epsilon: clip parameter
"""
assert len(net_list) == 2
assert len(optimizers_list) == 2
self.name = 'PPO_CLIP'
self.epsilon = epsilon
self.critic, self.actor = net_list
assert isinstance(self.critic, ValueNetwork)
assert isinstance(self.actor, StochasticPolicyNetwork)
self.critic_opt, self.actor_opt = optimizers_list
[docs] def a_train(self, tfs, tfa, tfadv, oldpi_prob):
"""
Update policy network
:param tfs: state
:param tfa: act
:param tfadv: advantage
:param oldpi_prob: old policy distribution
:return: None
"""
try:
tfs = np.array(tfs, np.float32)
tfa = np.array(tfa, np.float32)
tfadv = np.array(tfadv, np.float32)
except:
pass
with tf.GradientTape() as tape:
_ = self.actor(tfs)
pi_prob = tf.exp(self.actor.policy_dist.logp(tfa))
ratio = pi_prob / (oldpi_prob + EPS)
surr = ratio * tfadv
aloss = -tf.reduce_mean(
tf.minimum(surr, tf.clip_by_value(ratio, 1. - self.epsilon, 1. + self.epsilon) * tfadv))
a_gard = tape.gradient(aloss, self.actor.trainable_weights)
self.actor_opt.apply_gradients(zip(a_gard, self.actor.trainable_weights))
[docs] def c_train(self, tfdc_r, s):
"""
Update actor network
:param tfdc_r: cumulative reward
:param s: state
:return: None
"""
tfdc_r = np.array(tfdc_r, dtype=np.float32)
with tf.GradientTape() as tape:
v = self.critic(s)
advantage = tfdc_r - v
closs = tf.reduce_mean(tf.square(advantage))
grad = tape.gradient(closs, self.critic.trainable_weights)
self.critic_opt.apply_gradients(zip(grad, self.critic.trainable_weights))
[docs] def cal_adv(self, tfs, tfdc_r):
"""
Calculate advantage
:param tfs: state
:param tfdc_r: cumulative reward
:return: advantage
"""
tfdc_r = np.array(tfdc_r, dtype=np.float32)
advantage = tfdc_r - self.critic(tfs)
return advantage.numpy()
[docs] def update(self, s, a, r, a_update_steps, c_update_steps):
"""
Update parameter with the constraint of KL divergent
:param s: state
:param a: act
:param r: reward
:return: None
"""
adv = self.cal_adv(s, r)
# adv = (adv - adv.mean())/(adv.std()+1e-6) # adv norm, sometimes helpful
_ = self.actor(s)
oldpi_prob = tf.exp(self.actor.policy_dist.logp(a))
oldpi_prob = tf.stop_gradient(oldpi_prob)
# update actor
for _ in range(a_update_steps):
self.a_train(s, a, adv, oldpi_prob)
# update critic
for _ in range(c_update_steps):
self.c_train(r, s)
[docs] def get_action(self, s):
"""
Choose action
:param s: state
:return: clipped act
"""
return self.actor([s])[0].numpy()
[docs] def get_action_greedy(self, s):
"""
Choose action
:param s: state
:return: clipped act
"""
return self.actor([s], greedy=True)[0].numpy()
[docs] def get_v(self, s):
"""
Compute value
:param s: state
:return: value
"""
try:
s = s.astype(np.float32)
if s.ndim < 2: s = s[np.newaxis, :]
except:
pass
res = self.critic(s)[0, 0]
return res
[docs] def save_ckpt(self, env_name):
"""
save trained weights
:return: None
"""
save_model(self.actor, 'actor', self.name, env_name)
save_model(self.critic, 'critic', self.name, env_name)
[docs] def load_ckpt(self, env_name):
"""
load trained weights
:return: None
"""
load_model(self.actor, 'actor', self.name, env_name)
load_model(self.critic, 'critic', self.name, env_name)
[docs] def learn(self, env, train_episodes=200, test_episodes=100, max_steps=200, save_interval=10,
gamma=0.9, mode='train', render=False, batch_size=32, a_update_steps=10, c_update_steps=10,
plot_func=None):
"""
learn function
:param env: learning environment
:param train_episodes: total number of episodes for training
:param test_episodes: total number of episodes for testing
:param max_steps: maximum number of steps for one episode
:param save_interval: timesteps for saving
:param gamma: reward discount factor
:param mode: train or test
:param render: render each step
:param batch_size: udpate batchsize
:param a_update_steps: actor update iteration steps
:param c_update_steps: critic update iteration steps
:param plot_func: additional function for interactive module
:return: None
"""
t0 = time.time()
if mode == 'train':
print('Training... | Algorithm: {} | Environment: {}'.format(self.name, env.spec.id))
reward_buffer = []
for ep in range(1, train_episodes + 1):
s = env.reset()
buffer_s, buffer_a, buffer_r = [], [], []
ep_rs_sum = 0
for t in range(max_steps): # in one episode
if render:
env.render()
a = self.get_action(s)
s_, r, done, _ = env.step(a)
buffer_s.append(s)
buffer_a.append(a)
buffer_r.append(r)
s = s_
ep_rs_sum += r
# update ppo
if (t + 1) % batch_size == 0 or t == max_steps - 1 or done:
if done:
v_s_ = 0
else:
try:
v_s_ = self.get_v(s_)
except:
v_s_ = self.get_v([s_]) # for raw-pixel input
discounted_r = []
for r in buffer_r[::-1]:
v_s_ = r + gamma * v_s_
discounted_r.append(v_s_)
discounted_r.reverse()
# bs = buffer_s if len(buffer_s[0].shape)>1 else np.vstack(buffer_s) # no vstack for raw-pixel input
bs = buffer_s
ba, br = np.vstack(buffer_a), np.array(discounted_r)[:, np.newaxis]
buffer_s, buffer_a, buffer_r = [], [], []
self.update(bs, ba, br, a_update_steps, c_update_steps)
if done:
break
print(
'Episode: {}/{} | Episode Reward: {:.4f} | Running Time: {:.4f}'.format(
ep, train_episodes, ep_rs_sum,
time.time() - t0
)
)
reward_buffer.append(ep_rs_sum)
if plot_func is not None:
plot_func(reward_buffer)
if ep and not ep % save_interval:
self.save_ckpt(env_name=env.spec.id)
plot_save_log(reward_buffer, algorithm_name=self.name, env_name=env.spec.id)
self.save_ckpt(env_name=env.spec.id)
plot_save_log(reward_buffer, algorithm_name=self.name, env_name=env.spec.id)
# test
elif mode == 'test':
self.load_ckpt(env_name=env.spec.id)
print('Testing... | Algorithm: {} | Environment: {}'.format(self.name, env.spec.id))
reward_buffer = []
for eps in range(test_episodes):
ep_rs_sum = 0
s = env.reset()
for step in range(max_steps):
if render:
env.render()
action = self.get_action_greedy(s)
s, reward, done, info = env.step(action)
ep_rs_sum += reward
if done:
break
print('Episode: {}/{} | Episode Reward: {:.4f} | Running Time: {:.4f}'.format(
eps, test_episodes, ep_rs_sum, time.time() - t0)
)
reward_buffer.append(ep_rs_sum)
if plot_func:
plot_func(reward_buffer)
else:
print('unknown mode type')