multi-rl-crowd-sim

Multi-Agent Reinforcement Learning of Crowd Simulation in GPU (PyTorch)
Also, developing Predator-Prey Behaviours and a Battlefield Simulation!

Check the following videos for behaviour learnt by each simulation type:

main_random.py https://www.youtube.com/watch?v=VC5WT70MBQU
main_circle.py https://www.youtube.com/watch?v=B48TkaBFe-c
main_hunt.py https://www.youtube.com/watch?v=Ds9O9wPyF8g
main_army.py https://www.youtube.com/watch?v=IZ1Ht6k2U5E

gym_hunt.py Gym environment for Predator-Prey Behaviours (in-progress)

Ranged army self-play on a battlefield with obstacles is also in-progress!
Note: This is like a Sniper army battle where they can hide behind each other.
The strength of hitting also reduces as their health decrease by the attacks.