Learning cooperative behaviours in complex 3D games with multi-agent reinforcement learning
Multi-Agent systems can be used to deal with plenty of real world problems in almost any industry(Robotics, Distributed Control, Telecommunication,etc). In these industries most of these problems would be complex and often the solutions would require a group of agents that must cooperate and coor...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
Nanyang Technological University
2022
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/156606 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | Multi-Agent systems can be used to deal with plenty of real world problems in
almost any industry(Robotics, Distributed Control, Telecommunication,etc). In
these industries most of these problems would be complex and often the solutions
would require a group of agents that must cooperate and coordinate their
action.Through Multi-Agent Reinforcement Learning(MARL) multiple agents
will interact with each other in the same environment, either cooperatively or
competitively using centralized training with decentralized execution. This
project aims to analyse MARL algorithms, selecting the algorithm with the most
potential that would be able to learn cooperative behaviours effectively and how
it would be compared to other RL algorithms. |
|---|