Procgen Integration (#197)

Procgen library change in https://github.com/Trinkle23897/procgen
- Fix Qt library include error
- Fix globalGameRegistry == nullptr issue by adding another explicit make_* function (thanks to Yukun!)
- Disable auto-reset because envpool has already handled this feature

Changes:
- Add new dependency `qtdeclarative5-dev`, rebuild docker image
- Update pypi wheel from manylinux_2_17_x86_64 to manylinux_2_24_x86_64 because of libstdc++.so.6 symbol
- Add 42 procgen envs
- Refactor `base_path`

Co-authored-by: Jiayi Weng <trinkle23897@gmail.com>
Co-authored-by: Leo Guo <jiongtig@andrew.cmu.edu>

Author: 3 people

.github/workflows/release.yml

@@ -9,7 +9,7 @@ on:
 jobs:
   release:
     runs-on: ubuntu-latest
-    container: trinkle23897/envpool-release:2022-12-29-03c3d95
+    container: trinkle23897/envpool-release:2023-01-02-5f1a5fd
     strategy:
       matrix:
         python-version: ['3.7', '3.8', '3.9', '3.10', '3.11']
@@ -27,7 +27,7 @@ jobs:
     - name: Build
       run: |
         make pypi-wheel
-        pip3 install wheelhouse/*.whl --force-reinstall
+        pip3 install dist/*.whl --force-reinstall
     - name: Test
       run: |
         make release-test

Makefile

@@ -180,7 +180,7 @@ docker-release-launch: docker-release
 	docker run --network=host -v /:/host --shm-size=4gb -it $(PROJECT_NAME)-release:$(DOCKER_TAG) bash
 
 pypi-wheel: auditwheel-install bazel-release
-	ls dist/*.whl -Art | tail -n 1 | xargs auditwheel repair --plat manylinux_2_17_x86_64
+	ls dist/*.whl -Art | tail -n 1 | xargs auditwheel repair --plat manylinux_2_24_x86_64
 
 release-test1:
 	cd envpool && python3 make_test.py

README.md

@@ -16,7 +16,7 @@
 - [x] [ViZDoom single player](https://envpool.readthedocs.io/en/latest/env/vizdoom.html)
 - [x] [DeepMind Control Suite](https://envpool.readthedocs.io/en/latest/env/dm_control.html)
 - [x] [Box2D](https://envpool.readthedocs.io/en/latest/env/box2d.html)
-- [ ] Procgen
+- [x] [Procgen](https://envpool.readthedocs.io/en/latest/env/procgen.html)
 - [ ] Minigrid
 
 Here are EnvPool's several highlights:

WORKSPACE

@@ -8,6 +8,20 @@ load("//envpool:workspace1.bzl", workspace1 = "workspace")
 
 workspace1()
 
+# QT special, cannot move to workspace2.bzl, not sure why
+
+load("@local_config_qt//:local_qt.bzl", "local_qt_path")
+
+new_local_repository(
+    name = "qt",
+    build_file = "@com_justbuchanan_rules_qt//:qt.BUILD",
+    path = local_qt_path(),
+)
+
+load("@com_justbuchanan_rules_qt//tools:qt_toolchain.bzl", "register_qt_toolchains")
+
+register_qt_toolchains()
+
 load("//envpool:pip.bzl", pip_workspace = "workspace")
 
 pip_workspace()

docker/dev-cn.dockerfile

@@ -7,7 +7,7 @@ ARG HOME=/root
 ARG PATH=$PATH:$HOME/go/bin
 
 RUN apt-get update \
-    && apt-get install -y python3-pip python3-dev golang-1.18 clang-format-11 git wget swig tmux clang-tidy vim \
+    && apt-get install -y python3-pip python3-dev golang-1.18 clang-format-11 git wget swig tmux clang-tidy vim qtdeclarative5-dev \
     && rm -rf /var/lib/apt/lists/*
 RUN ln -s /usr/bin/python3 /usr/bin/python
 RUN ln -sf /usr/lib/go-1.18/bin/go /usr/bin/go

docker/dev.dockerfile

@@ -7,7 +7,7 @@ ARG HOME=/root
 ARG PATH=$PATH:$HOME/go/bin
 
 RUN apt-get update \
-    && apt-get install -y python3-pip python3-dev golang-1.18 clang-format-11 git wget swig tmux clang-tidy vim \
+    && apt-get install -y python3-pip python3-dev golang-1.18 clang-format-11 git wget swig tmux clang-tidy vim qtdeclarative5-dev \
     && rm -rf /var/lib/apt/lists/*
 RUN ln -s /usr/bin/python3 /usr/bin/python
 RUN ln -sf /usr/lib/go-1.18/bin/go /usr/bin/go

docker/release.dockerfile

@@ -14,7 +14,8 @@ RUN apt-get update && apt-get install -y software-properties-common && add-apt-r
 RUN apt-get update \
     && apt-get install -y git curl wget gcc-9 g++-9 build-essential swig make \
     zlib1g-dev libbz2-dev libreadline-dev libsqlite3-dev libncursesw5-dev libffi-dev liblzma-dev \
-    llvm xz-utils tk-dev libxml2-dev libxmlsec1-dev
+    llvm xz-utils tk-dev libxml2-dev libxmlsec1-dev qtdeclarative5-dev \
+    && rm -rf /var/lib/apt/lists/*
 # use self-compiled openssl instead of system provided (1.0.2)
 RUN apt-get remove -y libssl-dev
 

docs/env/procgen.rst

@@ -0,0 +1,97 @@
+Procgen
+=======
+
+We use ``procgen==0.10.7`` as the codebase.
+See https://github.com/openai/procgen/tree/0.10.7
+
+
+Options
+-------
+
+* ``task_id (str)``: see available tasks below;
+* ``num_envs (int)``: how many environments you would like to create;
+* ``batch_size (int)``: the expected batch size for return result, default to
+  ``num_envs``;
+* ``num_threads (int)``: the maximum thread number for executing the actual
+  ``env.step``, default to ``batch_size``;
+* ``seed (int)``: the environment seed, default to ``42``;
+* ``max_episode_steps (int)``: the maximum number of steps for one episode,
+  each procgen game has different timeout value;
+* ``env_name (str)``: one of 16 procgen env name;
+* ``num_levels (int)``: default to ``0``;
+* ``start_level (int)``: default to ``0``;
+* ``use_sequential_levels (bool)``: default to ``False``;
+* ``center_agent (bool)``: default to ``True``;
+* ``use_backgrounds (bool)``: default to ``True``;
+* ``use_monochrome_assets (bool)``: default to ``False``;
+* ``restrict_themes (bool)``: default to ``False``;
+* ``use_generated_assets (bool)``: default to ``False``;
+* ``paint_vel_info (bool)``: default to ``False``;
+* ``use_easy_jump (bool)``: default to ``False``;
+* ``distribution_mode (int)``: one of ``(0, 1, 2, 10)``; ``0`` stands for easy
+  mode, ``1`` stands for hard mode, ``2`` stands for extreme mode, ``10``
+  stands for memory mode. The default value is determined by ``task_id``.
+
+Note: arguments after ``env_name`` are provided by procgen environment itself.
+We keep the default value as-is. We haven't tested the setting of
+``use_sequential_levels == True``, and have no promise it is aligned with the
+original version of procgen (PRs for fixing this issue are highly welcome).
+
+
+Observation Space
+-----------------
+
+The observation image size is ``(64, 64, 3)``.
+
+
+Action Space
+------------
+
+15 action buttons in total, ranging from 0 to 14.
+
+
+Available Tasks
+---------------
+
+* ``BigfishEasy-v0``
+* ``BigfishHard-v0``
+* ``BossfightEasy-v0``
+* ``BossfightHard-v0``
+* ``CaveflyerEasy-v0``
+* ``CaveflyerHard-v0``
+* ``CaveflyerMemory-v0``
+* ``ChaserEasy-v0``
+* ``ChaserHard-v0``
+* ``ChaserExtreme-v0``
+* ``ClimberEasy-v0``
+* ``ClimberHard-v0``
+* ``CoinrunEasy-v0``
+* ``CoinrunHard-v0``
+* ``DodgeballEasy-v0``
+* ``DodgeballHard-v0``
+* ``DodgeballExtreme-v0``
+* ``DodgeballMemory-v0``
+* ``FruitbotEasy-v0``
+* ``FruitbotHard-v0``
+* ``HeistEasy-v0``
+* ``HeistHard-v0``
+* ``HeistMemory-v0``
+* ``JumperEasy-v0``
+* ``JumperHard-v0``
+* ``JumperMemory-v0``
+* ``LeaperEasy-v0``
+* ``LeaperHard-v0``
+* ``LeaperExtreme-v0``
+* ``MazeEasy-v0``
+* ``MazeHard-v0``
+* ``MazeMemory-v0``
+* ``MinerEasy-v0``
+* ``MinerHard-v0``
+* ``MinerMemory-v0``
+* ``NinjaEasy-v0``
+* ``NinjaHard-v0``
+* ``PlunderEasy-v0``
+* ``PlunderHard-v0``
+* ``StarpilotEasy-v0``
+* ``StarpilotHard-v0``
+* ``StarpilotExtreme-v0``

docs/index.rst

@@ -91,6 +91,7 @@ stable version through `envpool.readthedocs.io/en/stable/
    env/classic_control
    env/dm_control
    env/mujoco_gym
+   env/procgen
    env/toy_text
    env/vizdoom
 

docs/spelling_wordlist.txt

@@ -68,3 +68,4 @@ api
 jit
 mins
 lidar
+procgen