eprllib.Agents

Agents

This module contains classes for representing and manipulating agents in the environment. The agents are responsible for taking actions in the environment following a specified policy that responses to the current state of the environment.

In this module, you will find:

• AgentSpec: The main class for defining agents, including their observation,

  filter, action, action_mapper, and reward specifications.

• ObservationSpec: Defines the observation space for the agent.

• FilterSpec: Defines filters to preprocess observations before they are fed to the agent.

• ActionSpec: Defines the action space and actuators for the agent.

• ActionMapperSpec: Defines ActionMappers that determine how the agent

  should transform policy actions into actutators an actions.

• RewardSpec: Defines the reward function for the agent.

Note

Filter must to be coordinated with Connector.

Additionally, you will find base classes and some applications for Filters, Rewards, and ActionMappers, which are essential parts of an agent in eprllib.

Modules

ActionMappers	ActionMappers :2: (WARNING/2) Title underline too short. ActionMappers =========
ActionSpec	Specification for the action space and actuators This module defines the ActionSpec class, which is used to specify the configuration of action space and actuators for agents in reinforcement learning environments.
AgentSpec	Defining agents
Filters	Filters
ObservationSpec	Specification for the observation space and parameters This module defines the ObservationSpec class, which is used to specify the configuration of observation space and parameters for agents in reinforcement learning environments.
Rewards	Reward functions in eprllib