Skip to content

Latest commit

 

History

History
36 lines (31 loc) · 2.03 KB

unity.md

File metadata and controls

36 lines (31 loc) · 2.03 KB

Unity Development

Details on the unity development environment used for the simulation. Our simulation project can be found under vehicle-sim-env/ and has been developed in Unity version 2019.3.0f6.

Unity Packages

  • The Unity Machine Learning Agents Toolkit
    • version: 0.13.1
    • an open-source Unity plugin that enables games and simulations to serve as environments for training intelligent agents.
  • Vehicle Tools
    • version: 1.9
    • Asset by Unity from Unity's Asset Store. 3 basic vehicle models created using Unity's wheel collider. The family car was modified for our vehicle
  • Road Architect
    • version: 1.7
    • Open source Unity package used to create roads
  • Terrin Tools Sample Asset Pack
    • version: 1.0
    • Asset by Unity from Unity's Asset Store. Includes some helpful tools for creating terrains.

Package Modifications

  • Road Architect
    • Lane Markers (see RL Environment)
      • Used the side walk prefab from Road Architect to construct a custom invisible wall for lane guidance in the reinforcement learning model.
  • Vehicle Tools
    • Family Car: see Vehicle Agent
      • The base model of our vehicle. We altered the included vehicle script to create our agent.
      • Altered the prefab to include wheel trigger collision boxes with ray casting for road positioning.

Project Creation

Note: The following has already been completed for the vehicle-sim-env project. Given the active development of ML-Agents, we are documenting the steps to create our project.

  1. Create new Unity Project
  2. Clone Machine Learning Agents Toolkit
  3. Copy ML-Agents/UnitySDK/ML-Agents => new_proj/Assets/
  4. Import Road Architect .unityProject
  5. Import Vehicle Tools
  6. Import Terrain Tools