This project was developed for a paper that studies manually programmed AI in combination with Reinforcement Learning for simulating non-player characters. This scenario is based on the ML Agents Ecosystem. It includes two agents that are manually programmed (no ML involved) for comparison purposes.
The project uses Unity ML-Agents to train ML-based agents.
The following are the goals of this project:
- Integrate manually programmed agents
- Test functionality
- Code finalizing
To run this environment, you will need:
- Python 3.6.1 or greater
- mlagents version 0.16.1
- Unity 2019.2.0f1 or greater
More can be found at the ML Agents Ecosystem repository.