VisualEnv: visual Gym environments with Blender
This tool addresses the need for visually realistic environments in reinforcement learning research, though it appears incremental as it combines existing software.
The authors introduced VisualEnv, a tool that integrates Blender and OpenAI Gym to create custom reinforcement learning environments with photorealistic rendering, and demonstrated its functionality through example problems.
In this paper VisualEnv, a new tool for creating visual environment for reinforcement learning is introduced. It is the product of an integration of an open-source modelling and rendering software, Blender, and a python module used to generate environment model for simulation, OpenAI Gym. VisualEnv allows the user to create custom environments with photorealistic rendering capabilities and full integration with python. The framework is described and tested on a series of example problems that showcase its features for training reinforcement learning agents.