从 Map 数据创建 OpenAI Gym 环境

Question

I am just starting out with reinforcement learning and trying to create a custom environment with OpenAI gym. However, I am stumped with trying to create an environment (with roads and intersections) from map data, say Google Maps.

Would appreciate any help I can get.

Answer 1

The best way that I have found would be to use osmnx to have your map data ready in a graph format. It allows easy access of nodes that make up a street map in a graph or geopandas form. This data can then be used to drive the game logic. With regards to rendering, external rendering tools like PyGame can then be used.