BaseRLProtocol

Full path: schola.core.protocols.base.BaseRLProtocol

schola.core.protocols.base.BaseRLProtocol

BaseRLProtocol

BaseRLProtocol

Bases: BaseProtocol

Base class for reinforcement learning protocols.

This class extends BaseProtocol with methods specific to RL environments, including reset, step, and action messaging.

Methods

Item	Description
__init__()	—
close()	Close the protocol connection.
get_definition	Get the environment definition from Unreal Engine.
send_action_msg	Send actions to the environment and receive the next state.
send_reset_msg	Send a reset message to restart the environment.
send_startup_msg	Send the startup message with auto-reset configuration.
start()	Start the protocol connection.

Attributes

Item	Description
properties	Get protocol-specific properties.

get_definition

get_definition()

Get the environment definition from Unreal Engine.

Returns

A tuple containing:

• List of agent IDs per environment
• List of agent groups per environment (used for grouping agents)
• Observation spaces for each environment and agent
• Action spaces for each environment and agent

Return type: Tuple[List[List[str]], List[Dict[int, str]], Dict[int, Dict[str, gym.Space]], Dict[int, Dict[str, gym.Space]]]

---

send_action_msg

send_action_msg(actions, action_space)

Send actions to the environment and receive the next state.

Parameters

actions (Dict[int, Dict[str, Any]]) : Actions to take, indexed by environment ID and agent ID.

action_space (Dict[str, gym.Space]) : The action spaces used to serialize the actions.

Returns

A tuple containing:

• Observations for each environment
• Rewards for each environment
• Termination flags for each environment
• Truncation flags for each environment
• Info dicts for each environment
• Initial observations if auto-reset occurred
• Initial info dicts if auto-reset occurred

Return type: Tuple[List[Dict[str,Any]], List[float], List[Dict[str,bool]], List[Dict[str,bool]], List[Dict[str,str]], Dict[int,Dict[str, Any]], Dict[int,Dict[str, str]]]

---

send_reset_msg

send_reset_msg(seeds=None, options=None)

Send a reset message to restart the environment.

Parameters

seeds (List, optional) : List of random seeds for each environment.

options (List, optional) : List of reset options for each environment.

Returns

A tuple containing:

• List of initial observations for each environment
• List of initial info dicts for each environment

Return type: Tuple[List[Dict[str, Any]], List[Dict[str, Dict[str,str]]]]

---

send_startup_msg

send_startup_msg(auto_reset_type=AutoResetType.SAME_STEP)

Send the startup message with auto-reset configuration.

Parameters

auto_reset_type (AutoResetType, default=AutoResetType.SAME_STEP) : The type of auto-reset behavior to use when episodes end.

---