search

geek-ai / MAgent

A Platform for Many-Agent Reinforcement Learning
MIT License
1.68k stars 332 forks source link

Mode that agents cannot penetrate each other? #25

Closed sunfanyunn closed 6 years ago

sunfanyunn commented 6 years ago

Is there any support for environments where agents cannot go through each other?

For example, in the following scenario, even if agent a has a moving speed of 2, it still cannot move away.

 b
bab
 b
merrymercy commented 6 years ago

Sorry, there is no such mode.

You need to modify it by yourself. The related code is https://github.com/geek-ai/MAgent/blob/872680a8b9fee02c527306e65c4aeee1641b11a5/src/gridworld/GridWorld.cc#L573-L577

sunfanyunn commented 6 years ago

OK. Thanks