Construction Environments

This is the code necessary to run the Construction environments introduced by Bapst et al. (2019) and Hamrick et al. (2020).

Overview

These Construction tasks are provided through pre-packaged Docker containers.

This package consists of support code to run these Docker containers. You interact with the task environment via a dm_env Python interface.

Please see the documentation for more detailed information on the available tasks, actions and observations.

Requirements

The Construction tasks are intended to be run on Linux and are not officially supported on Mac and Windows. However, they can in principle be run on any platform (though installation may be more of a headache). In particular, on Windows, you will need to install and run the Construction tasks with WSL.

The Construction tasks require Docker, Python 3.6.1 or later and a x86-64 CPU with SSE4.2 support. We do not attempt to maintain a working version for Python 2.

Note: We recommend using Python virtual environment to mitigate conflicts with your system's Python environment.

Download and install Docker:

• For Linux, install Docker-CE
• Install Docker Desktop for OSX or Windows.

Quickstart

After installing Docker and Python (see above), install the Construction tasks by running the following command from the root of this repository:

pip install .

Then, from a Python interpreter:

>>> import dm_construction
>>> env = dm_construction.get_environment("covering", difficulty=0)
>>> time_step = env.reset()
>>> print(time_step.step_type)
StepType.FIRST

This may take a little while the first time you run it, as this command will automatically download the Docker container for the environment, start it, and connect to it through Python. For more details, see the documentation and the Environment Loading notebook demo.

Demos

If you wish to run the demos, you will also need to install ffmpeg. Here are some options for how to install ffmpeg:

• Cross-platform with Anaconda: conda install ffmpeg
• Ubuntu: apt-get install ffmpeg
• Mac with Homebrew: brew install ffmpeg

Next, run the following command to install a few additional dependencies:

pip install ".[demos]"

Troubleshooting

See below for some common errors and how to fix them.

Protobuf

AttributeError: module 'google.protobuf.descriptor' has no attribute '_internal_create_key'

Sometimes the protobuf library can get into a bad state if you have tried to update it. To fix this, do pip uninstall protobuf then pip install protobuf to fix it.

Docker

requests.exceptions.ConnectionError: ('Connection aborted.', FileNotFoundError(2, 'No such file or directory'))

This likely means Docker is not installed, or not running. Make sure you have both installed it and started the Docker engine. You can test if the Docker engine is running by running docker info.

Shapely

There is currently a bug in the 1.7.0 version of the Shapely library that may cause problems on Mac and Windows.

• Mac OSX (OSError: Could not find lib c or load any of its variants): see this StackOverflow question for a suggested fix.

• Windows (FileNotFoundError: Could not find module 'geos.dll' (or one of its dependencies)): the Construction tasks are not officially supported on Windows. Please run the library using WSL instead.

References

Bapst, V., Sanchez-Gonzalez, A., Doersch, C., Stachenfeld, K., Kohli, P., Battaglia, P., & Hamrick, J. (2019, May). Structured agents for physical construction. In International Conference on Machine Learning (pp. 464-474).

Hamrick, J. B., Bapst, V., Sanchez-Gonzalez, A., Pfaff, T., Weber, T., Buesing, L., & Battaglia, P. W. (2020). Combining Q-Learning and Search with Amortized Value Estimates. ICLR 2020.