eprllib.ObservationFunctions.fully_shared_parameters#
Fully Shared Parameters Policy#
This module implements a fully shared parameters policy for the observation function.
The flat observation for each agent must be a numpy array and the size must be equal to all agents. To this end, a concatenation of variables is performed.
AgentID one-hot enconde vector
Agent state
Actuators state (include other agents)
Classes
|