Inspiration
- realistic rendering
What it does
- visualizes roads, pavements, road markings, nature (grass) to enable AI models to learn self-driving cars in a virtual environment. The data is given in 2D street data (Open Drive format) and converts it to a nice looking 3D map (See photos).
How we built it
- converted 2D to 3D coordinates
- applying height to vertices by using Perlin noise to generate terrain
- added 3D models and road markings to scene to make it realistic
- Also added sidewalks and street lights for a better ambient
Challenges we ran into
- xodr visualizer does only work with English locales
- rendering textures on 3D objects (we hated Unity for a while)
- limited power on our laptops
Accomplishments that we're proud of
- Having street lamps
- To accomplish more then we expected to especially in the last 10 hours
What we learned
- learned a lot about 3D modeling, rendering and texturing
- learned how to use Unity
What's next for roads.sexy
- Put in more auto generated stuff like houses along the sidewalk
- Better Shaders and Textures
- Improve performance
- But first of all, have a very very good Sleep!
Log in or sign up for Devpost to join the conversation.