Skip to content

Multi-Exit Evacuation simulation; Rainbow DQN application

Notifications You must be signed in to change notification settings

KOWLEWzr/MultiExit-Rainbow

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

16 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

MultiExit-Rainbow DQN

The proposed multi-exit evacuation simulation is based on Rainbow DQN, where RVO2 library is used to avoid collision and Rainbow DQN is applied to guide the pedestrians to evacuate the room with reasonable behaviors.

Keywords

Deep reinforcement learning; Multi-Exit Evacuation Simulation; Rainbow DQN application

Setup

Environment

  • Ubuntu 18.04
  • CPU: Intel i7 + GPU: Nvidia RTX 2080 ti + RAM: 64G
  • Python 3.7 + Cuda 10.0

Requirements

  • Numpy
  • Pytorch 1.0
  • OpenCV
  • xlwt and xlrd (optional)

Usage

The code can be run Directly with default arguments:
'''
python3 main.py '''

Or you can run the code using the following options:
'''
python3 main.py
--noisy True
--double True
--dueling True
--prioritized_replay True
--c51 True
--multi_step 3
--num_agents 12
--read_model None
--evaluate False
'''

Demos

Three video demos are presented:

Acknowledgements

About

Multi-Exit Evacuation simulation; Rainbow DQN application

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%