env_makers.py

"""
This project was developed by Rocky Duan, Peter Chen, Pieter Abbeel for the Berkeley Deep RL Bootcamp, August 2017. Bootcamp website with slides and lecture videos: https://sites.google.com/view/deep-rl-bootcamp/.

Copyright 2017 Deep RL Bootcamp Organizers.

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

"""


import logging
import numpy as np
import cv2
import gym
from collections import deque
from gym import spaces
import gym.envs.registration
from gym.envs.atari.atari_env import AtariEnv
import logger
import os
import roboschool
import tempfile
import gym.wrappers.monitoring

# Silence the log messages
gym.envs.registration.logger.setLevel(logging.WARNING)
gym.wrappers.monitoring.logger.setLevel(logging.WARNING)


class EnvMaker(object):
    def __init__(self, env_id):
        self.env_id = env_id

    def make(self):
        env = gym.make(self.env_id)
        if logger.get_dir() is not None:
            monitor_dir = os.path.join(logger.get_dir(), "gym_monitor")
            resume = True
            force = False
        else:
            monitor_dir = "/tmp/gym-monitoring"
            resume = False
            force = True
        env = gym.wrappers.Monitor(env, directory=monitor_dir, video_callable=False, force=force, resume=resume,
                                   write_upon_reset=True)
        if isinstance(env.unwrapped, AtariEnv):
            if '-ram-' in self.env_id:
                assert 'NoFrameskip' not in self.env_id
                env = ScaledFloatFrame(env)
            else:
                env = ScaledFloatFrame(wrap_atari_pg(env))
        return env


# Code below are adopted from OpenAI Baselines:
# https://github.com/openai/baselines/blob/master/baselines/common/atari_wrappers_deprecated.py
class NoopResetEnv(gym.Wrapper):
    def __init__(self, env=None, noop_max=30):
        """Sample initial states by taking random number of no-ops on reset.
        No-op is assumed to be action 0.
        """
        super(NoopResetEnv, self).__init__(env)
        self.noop_max = noop_max
        self.override_num_noops = None
        assert env.unwrapped.get_action_meanings()[0] == 'NOOP'

    def _reset(self):
        """ Do no-op action for a number of steps in [1, noop_max]."""
        self.env.reset()
        if self.override_num_noops is not None:
            noops = self.override_num_noops
        else:
            noops = np.random.randint(1, self.noop_max + 1)
        assert noops > 0
        obs = None
        for _ in range(noops):
            obs, _, done, _ = self.env.step(0)
            if done:
                obs = self.env.reset()
        return obs


class NormalizeActionWrapper(gym.Wrapper):
    def __init__(self, env):
        super().__init__(env)
        if isinstance(self.env.action_space, gym.spaces.Box):
            self.action_space = gym.spaces.Box(
                low=-1, high=1, shape=self.env.action_space.shape)
        else:
            self.action_space = self.env.action_space

    def step(self, action):
        if not isinstance(self.action_space, gym.spaces.Box):
            return super().step(action)
        # rescale action
        low = self.env.action_space.low
        high = self.env.action_space.high
        action = np.asarray(action)
        rescaled_action = (action + 1) / 2 * (high - low) + low
        return super().step(rescaled_action)


class FireResetEnv(gym.Wrapper):
    def __init__(self, env=None):
        """For environments where the user need to press FIRE for the game to start."""
        super(FireResetEnv, self).__init__(env)
        assert env.unwrapped.get_action_meanings()[1] == 'FIRE'
        assert len(env.unwrapped.get_action_meanings()) >= 3

    def _reset(self):
        self.env.reset()
        obs, _, done, _ = self.env.step(1)
        if done:
            self.env.reset()
        obs, _, done, _ = self.env.step(2)
        if done:
            self.env.reset()
        return obs


class EpisodicLifeEnv(gym.Wrapper):
    def __init__(self, env=None):
        """Make end-of-life == end-of-episode, but only reset on true game over.
        Done by DeepMind for the DQN and co. since it helps value estimation.
        """
        super(EpisodicLifeEnv, self).__init__(env)
        self.lives = 0
        self.was_real_done = True
        self.was_real_reset = False

    def _step(self, action):
        obs, reward, done, info = self.env.step(action)
        self.was_real_done = done
        # check current lives, make loss of life terminal,
        # then update lives to handle bonus lives
        lives = self.env.unwrapped.ale.lives()
        if lives < self.lives and lives > 0:
            # for Qbert somtimes we stay in lives == 0 condtion for a few frames
            # so its important to keep lives > 0, so that we only reset once
            # the environment advertises done.
            done = True
        self.lives = lives
        return obs, reward, done, info

    def _reset(self):
        """Reset only when lives are exhausted.
        This way all states are still reachable even though lives are episodic,
        and the learner need not know about any of this behind-the-scenes.
        """
        if self.was_real_done:
            obs = self.env.reset()
            self.was_real_reset = True
        else:
            # no-op step to advance from terminal/lost life state
            obs, _, _, _ = self.env.step(0)
            self.was_real_reset = False
        self.lives = self.env.unwrapped.ale.lives()
        return obs


class MaxAndSkipEnv(gym.Wrapper):
    def __init__(self, env=None, skip=4):
        """Return only every `skip`-th frame"""
        super(MaxAndSkipEnv, self).__init__(env)
        # most recent raw observations (for max pooling across time steps)
        self._obs_buffer = deque(maxlen=2)
        self._skip = skip

    def _step(self, action):
        total_reward = 0.0
        done = None
        for _ in range(self._skip):
            obs, reward, done, info = self.env.step(action)
            self._obs_buffer.append(obs)
            total_reward += reward
            if done:
                break

        max_frame = np.max(np.stack(self._obs_buffer), axis=0)

        return max_frame, total_reward, done, info

    def _reset(self):
        """Clear past frame buffer and init. to first obs. from inner env."""
        self._obs_buffer.clear()
        obs = self.env.reset()
        self._obs_buffer.append(obs)
        return obs


class ProcessFrame84(gym.ObservationWrapper):
    def __init__(self, env=None):
        super(ProcessFrame84, self).__init__(env)
        self.observation_space = spaces.Box(low=0, high=255, shape=(84, 84, 1))

    def _observation(self, obs):
        return ProcessFrame84.process(obs)

    @staticmethod
    def process(frame):
        if frame.size == 210 * 160 * 3:
            img = np.reshape(frame, [210, 160, 3]).astype(np.float32)
        elif frame.size == 250 * 160 * 3:
            img = np.reshape(frame, [250, 160, 3]).astype(np.float32)
        else:
            assert False, "Unknown resolution."
        img = img[:, :, 0] * 0.299 + img[:, :, 1] * \
            0.587 + img[:, :, 2] * 0.114
        resized_screen = cv2.resize(
            img, (84, 110), interpolation=cv2.INTER_AREA)
        x_t = resized_screen[18:102, :]
        x_t = np.reshape(x_t, [84, 84, 1])
        return x_t.astype(np.uint8)


class ClippedRewardsWrapper(gym.RewardWrapper):
    def _reward(self, reward):
        """Change all the positive rewards to 1, negative to -1 and keep zero."""
        return np.sign(reward)


class LazyFrames(object):
    def __init__(self, frames):
        """This object ensures that common frames between the observations are only stored once.
        It exists purely to optimize memory usage which can be huge for DQN's 1M frames replay
        buffers.

        This object should only be converted to numpy array before being passed to the model.

        You'd not belive how complex the previous solution was."""
        self._frames = frames

    def __array__(self, dtype=None):
        out = np.concatenate(self._frames, axis=2)
        if dtype is not None:
            out = out.astype(dtype)
        return out


class FrameStack(gym.Wrapper):
    def __init__(self, env, k):
        """Stack k last frames.

        Returns lazy array, which is much more memory efficient.

        See Also
        --------
        baselines.common.atari_wrappers.LazyFrames
        """
        gym.Wrapper.__init__(self, env)
        self.k = k
        self.frames = deque([], maxlen=k)
        shp = env.observation_space.shape
        self.observation_space = spaces.Box(
            low=0, high=255, shape=(shp[0], shp[1], shp[2] * k))

    def _reset(self):
        ob = self.env.reset()
        for _ in range(self.k):
            self.frames.append(ob)
        return self._get_ob()

    def _step(self, action):
        ob, reward, done, info = self.env.step(action)
        self.frames.append(ob)
        return self._get_ob(), reward, done, info

    def _get_ob(self):
        assert len(self.frames) == self.k
        return LazyFrames(list(self.frames))


class ScaledFloatFrame(gym.ObservationWrapper):
    def _observation(self, obs):
        # careful! This undoes the memory optimization, use
        # with smaller replay buffers only.
        return np.array(obs).astype(np.float32) / 255.0


def wrap_atari_pg(env):
    """Apply a common set of wrappers for Atari games."""
    assert 'NoFrameskip' in env.spec.id
    env = EpisodicLifeEnv(env)
    env = NoopResetEnv(env, noop_max=30)
    env = MaxAndSkipEnv(env, skip=4)
    if 'FIRE' in env.unwrapped.get_action_meanings():
        env = FireResetEnv(env)
    env = ProcessFrame84(env)
    env = FrameStack(env, 4)
    env = ClippedRewardsWrapper(env)
    return env