This paper applies reinforcement learning to train a predator to hunt multiple prey, which are able to reproduce, in a 2D simulation. It is shown that, using methods of curriculum learning, long-term reward discounting and stacked observations, a reinforcement-learning-based predator can achieve an economic strategy: Only hunt when there is still prey left to reproduce in order to maintain the population. Hence, purely selfish goals are sufficient to motivate a reinforcement learning agent for long-term planning and keeping a certain balance with its environment by not depleting its resources. While a comparably simple reinforcement learning algorithm achieves such behavior in the present scenario, providing a suitable amount of past and predictive information turns out to be crucial for the training success.
@inproceedings{ ritzALIFE21,
author = "Fabian Ritz and Daniel Ratke and Thomy Phan and Lenz Belzner and Claudia Linnhoff-Popien",
title = "A Sustainable Ecosystem through Emergent Cooperation in Multi-Agent Reinforcement Learning",
year = "2021",
abstract = "This paper applies reinforcement learning to train a predator to hunt multiple prey, which are able to reproduce, in a 2D simulation. It is shown that, using methods of curriculum learning, long-term reward discounting and stacked observations, a reinforcement-learning-based predator can achieve an economic strategy: Only hunt when there is still prey left to reproduce in order to maintain the population. Hence, purely selfish goals are sufficient to motivate a reinforcement learning agent for long-term planning and keeping a certain balance with its environment by not depleting its resources. While a comparably simple reinforcement learning algorithm achieves such behavior in the present scenario, providing a suitable amount of past and predictive information turns out to be crucial for the training success.",
url = "https://direct.mit.edu/isal/proceedings/isal/74/102891",
eprint = "https://direct.mit.edu/isal/proceedings-pdf/isal/33/74/1930024/isal a 00399.pdf",
publisher = "MIT Press Direct",
booktitle = "Conference on Artificial Life",
pages = "74--83",
doi = "https://doi.org/10.1162/isal_a_00399"
}
Related Articles
Relevant Research Areas
