ll7/gym-docker

OpenAI Gym in a docker environment

I modified a docker environment from jderobot, s.t. I can run OpenAI Gym inside a docker container (alongside Jupyter Lab) and I can see the rendered results.

Docker

Install

If necessary: https://docs.docker.com/get-docker/

Linux Postinstall

https://docs.docker.com/engine/install/linux-postinstall/

nvidia docker

https://github.com/NVIDIA/nvidia-docker

Docker Hub download

https://hub.docker.com/repository/docker/llx7/gym_docker

docker pull llx7/gym_docker

docker build

only if you need to: ./.docker/docker_build.sh

docker run

./docker_run.sh

VNC Viewer

VNC to connect a virtual desktop to docker

https://chrome.google.com/webstore/detail/vnc%C2%AE-viewer-for-google-ch/iabmpiboiopbgfabjmgeedhcmjenhbla?hl=en

Address: localhost:5900

pw: jderobot

(can be modified if docker image is rebuild and pw is adapted in .docker/vnc_startup.sh)

docker logs

docker logs gym_docker to get the jupyter website

further information

• JdeRobot/BehaviorStudio Docker Readme
• .docker/README.md

gym-docker's Issues

env.render() appears to be faster than realtime