Add NeurIPS 2022 Discrete Soft Actor Critic agent to agent zoo (#1860)

CHANGELOG.md

@@ -11,6 +11,7 @@ Copy and pasting the git commit messages is __NOT__ enough.
 ## [Unreleased]
 ### Added
 - Added a zoo agent, named Control-and-Supervised-Learning, from NeurIPS 2022 submission. This zoo agent runs in benchmark `driving_smarts==0.0`.
+- Added a zoo agent, named Discrete Soft Actor Critic, from NeurIPS 2022 submission. This zoo agent runs in benchmark `driving_smarts==0.0`.
 ### Changed
 ### Deprecated
 ### Fixed

docs/benchmarks/agent_zoo.rst

@@ -60,3 +60,9 @@ Available zoo agents
      - :attr:`~smarts.core.controllers.ActionSpaceType.TargetPose`
      - `code <https://github.com/smarts-project/smarts-project.rl/tree/master/control_and_supervised_learning>`__
      - Contributed as part of `NeurIPS 2022 Driving SMARTS <https://smarts-project.github.io/archive/2022_nips_driving_smarts/>`__ competition.
+   * - | zoo.policies:discrete-soft-actor-critic-agent-v0
+       | zoo/policies/discrete_soft_actor_critic
+     - driving_smarts==0.0
+     - :attr:`~smarts.core.controllers.ActionSpaceType.TargetPose`
+     - `code <https://github.com/smarts-project/smarts-project.rl/tree/master/discrete_soft_actor_critic>`__
+     - Contributed as part of `NeurIPS 2022 Driving SMARTS <https://smarts-project.github.io/archive/2022_nips_driving_smarts/>`__ competition.

docs/conf.py

@@ -38,7 +38,7 @@
     "sphinx.ext.coverage",  # to generate documentation coverage reports
     "sphinx.ext.extlinks",  # shorten external links
     "sphinx.ext.napoleon",  # support Numpy and Google doc style
-    "sphinx.ext.todo", # support for todo items
+    "sphinx.ext.todo",  # support for todo items
     "sphinx.ext.viewcode",  # link to sourcecode from docs
     "sphinx_rtd_theme",  # Read The Docs theme
     "sphinx_click",  # extract documentation from a `click` application

zoo/policies/__init__.py

@@ -134,6 +134,7 @@ def entry_point_iamp(**kwargs):
     locator="interaction-aware-motion-prediction-agent-v0", entry_point=entry_point_iamp
 )
 
+
 def entry_point_casl(**kwargs):
     pkg = "control_and_supervised_learning"
     module = ".policy"
@@ -146,6 +147,24 @@ def entry_point_casl(**kwargs):
         agent_builder=lib.Policy,
     )
 
+
 register(
-    locator="control-and-supervised-learning-agent-v0", entry_point=entry_point_casl,
-)
+    locator="control-and-supervised-learning-agent-v0",
+    entry_point=entry_point_casl,
+)
+
+
+def entry_point_dsac(**kwargs):
+    pkg = "discrete_soft_actor_critic"
+    module = ".policy"
+    lib = _verify_installation(pkg=pkg, module=module)
+
+    return AgentSpec(
+        interface=AgentInterface(
+            action=ActionSpaceType.TargetPose,
+        ),
+        agent_builder=lib.Policy,
+    )
+
+
+register(locator="discrete-soft-actor-critic-agent-v0", entry_point=entry_point_dsac)

zoo/policies/discrete_soft_actor_critic/setup.cfg

@@ -0,0 +1,19 @@
+[metadata]
+name = smarts_agent_dsac_stub
+version = 0.1.0
+url = https://github.com/huawei-noah/SMARTS
+description = SMARTS zoo agent.
+long_description = Trained Discrete Soft Actor Critic agent. See [SMARTS](https://github.com/huawei-noah/SMARTS).
+long_description_content_type=text/markdown
+classifiers=
+    Programming Language :: Python
+    Programming Language :: Python :: 3 :: Only
+    Programming Language :: Python :: 3.8
+
+[options]
+packages = find:
+include_package_data = True
+zip_safe = True
+python_requires = == 3.8.*
+install_requires = 
+    smarts_agent_dsac==0.1.0

zoo/policies/discrete_soft_actor_critic/setup.py

@@ -0,0 +1,4 @@
+from setuptools import setup
+
+if __name__ == "__main__":
+    setup()