matrx.agents.agent_brain.AgentBrain

class AgentBrain(memorize_for_ticks=None)

An artificial agent whose behaviour can be programmed to be, for example, (semi-)autonomous.

Attributes
memorize_for_ticks
state

Methods

create_context_menu_for_other(self, …)

Generate options for a context menu for a specific object/location that a user NOT controlling this human agent opened.

decide_on_action(self, state)

Contains the decision logic of the agent.

filter_observations(self, state)

Filters the world state before deciding on an action.

get_log_data(self)

Provides a dictionary of data for any Logger

initialize(self)

Method called by any world when it starts.

is_action_possible(self, action, action_kwargs)

Checks if an action would be possible.

send_message(self, message)

Sends a Message from this agent to others

Defines the behavior of an agent.

This class is the place where all the decision logic of an agent is contained. This class together with the matrx.objects.agent_body.AgentBody class represent a full agent.

This agent brain simply selects a random action from the possible actions it can do.

When you wish to create a new agent, this is the class you need to extend. In specific these are the functions you should override:

  • matrx.agents.agent_brain.initialize()

    Called before a world starts running. Can be used to initialize variables that can only be initialized after the brain is connected to its body (which is done by the world).

  • matrx.agents.agent_brain.filter_observations()

    Called before deciding on an action to allow detailed and agent specific filtering of the received world state.

  • matrx.agents.agent_brain.decide_on_action()

    Called to decide on an action.

  • matrx.agents.agent_brain.get_log_data()

    Called by data loggers to obtain data that should be logged from this agent internal reasoning.

Attributes
action_set: [str, …]

List of actions this agent can perform.

agent_id: str

The unique identified of this agent’s body in the world.

agent_name: str

The name of this agent.

agent_properties: dict

A dictionary of this agent’s matrx.objects.agent_body.AgentBody properties. With as keys the property name, and as value the property’s value.

These can be adjusted iff they are said to be adjustable (e.g. inside the attribute keys_of_agent_writable_props).

keys_of_agent_writable_props: [str, …]

List of property names that this agent can adjust.

messages_to_send: [Message, …]

List of messages this agent will send. Use the method matrx.agents.agent_brain.AgentBrain.send_message() to append to this list.

previous_action: str

The name of the previous performed or attempted action.

previous_action_result: ActionResult

The matrx.actions.action.ActionResult of the previously performed or attempted action.

received_messages: [Message, …]

The list of received messages.

rnd_gen: Random

The random generator for this agent.

rnd_seed: int

The random seed with which this agent’s rnd_gen was initialized. This seed is based on the master random seed given of the matrx.grid_world.GridWorld.

Methods

create_context_menu_for_other(self, …)

Generate options for a context menu for a specific object/location that a user NOT controlling this human agent opened.

decide_on_action(self, state)

Contains the decision logic of the agent.

filter_observations(self, state)

Filters the world state before deciding on an action.

get_log_data(self)

Provides a dictionary of data for any Logger

initialize(self)

Method called by any world when it starts.

is_action_possible(self, action, action_kwargs)

Checks if an action would be possible.

send_message(self, message)

Sends a Message from this agent to others

__init__(self, memorize_for_ticks=None)

Defines the behavior of an agent.

This class is the place where all the decision logic of an agent is contained. This class together with the matrx.objects.agent_body.AgentBody class represent a full agent.

This agent brain simply selects a random action from the possible actions it can do.

When you wish to create a new agent, this is the class you need to extend. In specific these are the functions you should override:

  • matrx.agents.agent_brain.initialize()

    Called before a world starts running. Can be used to initialize variables that can only be initialized after the brain is connected to its body (which is done by the world).

  • matrx.agents.agent_brain.filter_observations()

    Called before deciding on an action to allow detailed and agent specific filtering of the received world state.

  • matrx.agents.agent_brain.decide_on_action()

    Called to decide on an action.

  • matrx.agents.agent_brain.get_log_data()

    Called by data loggers to obtain data that should be logged from this agent internal reasoning.

Attributes
action_set: [str, …]

List of actions this agent can perform.

agent_id: str

The unique identified of this agent’s body in the world.

agent_name: str

The name of this agent.

agent_properties: dict

A dictionary of this agent’s matrx.objects.agent_body.AgentBody properties. With as keys the property name, and as value the property’s value.

These can be adjusted iff they are said to be adjustable (e.g. inside the attribute keys_of_agent_writable_props).

keys_of_agent_writable_props: [str, …]

List of property names that this agent can adjust.

messages_to_send: [Message, …]

List of messages this agent will send. Use the method matrx.agents.agent_brain.AgentBrain.send_message() to append to this list.

previous_action: str

The name of the previous performed or attempted action.

previous_action_result: ActionResult

The matrx.actions.action.ActionResult of the previously performed or attempted action.

received_messages: [Message, …]

The list of received messages.

rnd_gen: Random

The random generator for this agent.

rnd_seed: int

The random seed with which this agent’s rnd_gen was initialized. This seed is based on the master random seed given of the matrx.grid_world.GridWorld.

Methods

__init__(self[, memorize_for_ticks])

Defines the behavior of an agent.

create_context_menu_for_other(self, …)

Generate options for a context menu for a specific object/location that a user NOT controlling this human agent opened.

decide_on_action(self, state)

Contains the decision logic of the agent.

filter_observations(self, state)

Filters the world state before deciding on an action.

get_log_data(self)

Provides a dictionary of data for any Logger

initialize(self)

Method called by any world when it starts.

is_action_possible(self, action, action_kwargs)

Checks if an action would be possible.

send_message(self, message)

Sends a Message from this agent to others

Attributes

memorize_for_ticks

state