search

Unity-Technologies / ml-agents

The Unity Machine Learning Agents Toolkit (ML-Agents) is an open-source project that enables games and simulations to serve as environments for training intelligent agents using deep reinforcement learning and imitation learning.
https://unity.com/products/machine-learning-agents
Other
16.61k stars 4.09k forks source link

Curriculum Learning set up #2872

Closed the-zer07 closed 2 years ago

the-zer07 commented 4 years ago

I read the documentation about CL but I don't understand how to implement in my project. I'm doing a vehicle AI in a racing game, so actually there are 4 cars that must drive in a track. I'm using camera e raycast sensor to determine other vehicles and the walls. I saw that classic RL it's too hard, for each car it's very difficult concurrently learning the steering, accelerating/breaking and collision avoidance, so I think using CL could be a solution but I don't know what to write in my code. I thought before the cars should learn how to steer, accelerate and break and then collision avoidance. Maybe I should training the cars ignoring the collisions among them during the x% of the max steps, in this case the cars should learn how to steer and at the same time the correct value of acceleration and break. In the end after x% max steps activate the collisions and the cars should learn the collision avoidance. Could someone help me?

hvpeteet commented 3 years ago

Hi thanks for the interest. If you are still looking for advice on CL I would recommend the forums https://forum.unity.com/forums/ml-agents.453/ where other users can help you out with their experience as well.

hvpeteet commented 2 years ago

Closing since this has been stale for over a year and was just a redirect to the forums.

github-actions[bot] commented 2 years ago

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.