pytsc.wrappers package

Submodules

pytsc.wrappers module

pytsc.wrappers module

pytsc.wrappers module

pytsc.wrappers module

pytsc.wrappers.epymarl module

class pytsc.wrappers.epymarl.DomainRandomizedEPyMARLTrafficSignalNetwork(map_names, simulator_backend='sumo', **kwargs)

Bases: smac.env.multiagentenv.MultiAgentEnv

A wrapper for the TrafficSignalNetwork environment that supports domain randomization. It randomly selects a map from a provided list at each reset, and pads outputs so that the observation and action interfaces remain fixed (with a maximum number of agents).

Parameters

• map_names (list) – A list of map names to choose from.

• max_n_agents (int) – The maximum number of agents.

• simulator_backend (str) – The simulator backend to use (default “sumo”).

• **kwargs – Additional keyword arguments for TrafficSignalNetwork.

apply_actions(actions)

Apply actions and remove padding if necessary.

close()

get_avail_actions()

Get available actions from the current environment and pad the list.

get_env_info()

Get environment info from the current environment and pad the adjacency matrix. Also, set n_agents to max_n_agents.

get_local_rewards()

Get local rewards from the current environment and pad the list.

get_obs()

Get observations from the current environment and pad the list of observations.

get_obs_size()

Returns the size of the observation.

get_state()

Get the state from the current environment. The state is padded to match the max_n_agents.

get_state_size()

Returns the size of the global state.

get_stats()

get_total_actions()

Returns the total number of actions an agent could ever take.

reset()

Reset the environment. Reinitialize the underlying TrafficSignalNetwork using a (potentially) different map to achieve domain randomization.

step(actions)

Step through the environment using valid (unpadded) actions, then pad the observations before returning.

step_stats = None

class pytsc.wrappers.epymarl.EPyMARLTrafficSignalNetwork(map_name='pasubio', simulator_backend='sumo', **kwargs)

Bases: smac.env.multiagentenv.MultiAgentEnv

A wrapper for the TrafficSignalNetwork environment that allows for multi-agent reinforcement learning. This environment is designed to be used with the EPyMARL framework, which is a multi-agent reinforcement learning library.

Parameters

• map_name (str) – The name of the map to use (default “pasubio”).

• simulator_backend (str) – The simulator backend to use (default “sumo”).

• **kwargs – Additional keyword arguments for TrafficSignalNetwork.

apply_actions(actions)

close()

get_avail_actions()

Returns the available actions of all agents in a list.

get_domain_class()

get_env_info()

get_local_rewards()

get_network_flow()

get_obs()

Returns all agent observations in a list.

get_obs_size()

Returns the size of the observation.

get_pressures()

get_state()

Returns the global state.

get_state_size()

Returns the size of the global state.

get_stats()

get_total_actions()

Returns the total number of actions an agent could ever take.

is_terminated()

reset()

Returns initial observations and states.

set_domain_class(domain_class)

sim_step()

step(actions)

Returns reward, terminated, info.

step_stats = None

pytsc.wrappers.pymarl module

class pytsc.wrappers.pymarl.PyMARLTrafficSignalNetwork(map_name='monaco', simulator_backend='sumo', **kwargs)

Bases: smac.env.multiagentenv.MultiAgentEnv

apply_actions(actions)

close()

get_avail_actions()

Returns the available actions of all agents in a list.

get_domain_class()

get_env_info()

get_local_rewards()

get_network_flow()

get_obs()

Returns all agent observations in a list.

get_obs_size()

Returns the size of the observation.

get_pressures()

get_state()

Returns the global state.

get_state_size()

Returns the size of the global state.

get_stats()

get_total_actions()

Returns the total number of actions an agent could ever take.

is_terminated()

reset()

Returns initial observations and states.

set_domain_class(domain_class)

sim_step()

step(actions)

Returns reward, terminated, info.

step_stats = None

pytsc.wrappers.rllib module

Module contents