def __init__(self, simulator, sim_params):
"""Instantiate a Flow kernel object.
Parameters
----------
simulator : str
simulator type, must be one of {"traci"}
sim_params : flow.core.params.SimParams
simulation-specific parameters
Raises
------
ValueError
if the specified input simulator is not a valid type
"""
self.kernel_api = None
if simulator == "traci":
self.simulation = TraCISimulation(self)
self.scenario = TraCIScenario(self)
self.vehicle = KernelVehicle(self, sim_params)
self.traffic_light = TraCITrafficLight(self)
else:
raise ValueError('Simulator type "{}" is not valid.'.
format(simulator))
def __init__(self,
master_kernel,
sim_params):
"""See parent class."""
KernelVehicle.__init__(self, master_kernel, sim_params)
self.__ids = [] # ids of all vehicles
self.__human_ids = [] # ids of human-driven vehicles
self.__controlled_ids = [] # ids of flow-controlled vehicles
self.__controlled_lc_ids = [] # ids of flow lc-controlled vehicles
self.__rl_ids = [] # ids of rl-controlled vehicles
self.__observed_ids = [] # ids of the observed vehicles
# vehicles: Key = Vehicle ID, Value = Dictionary describing the vehicle
# Ordered dictionary used to keep neural net inputs in order
self.__vehicles = collections.OrderedDict()
# create a sumo_observations variable that will carry all information
# on the state of the vehicles for a given time step
self.__sumo_obs = {}
# total number of vehicles in the network
self.num_vehicles = 0
# number of rl vehicles in the network
self.num_rl_vehicles = 0
# contains the parameters associated with each type of vehicle
self.type_parameters = {}
# contain the minGap attribute of each type of vehicle
self.minGap = {}
# list of vehicle ids located in each edge in the network
self._ids_by_edge = dict()
# list of vehicle ids located in each lane in the network
self._ids_by_lane = dict()
# number of vehicles that entered the network for every time-step
self._num_departed = []
self._departed_ids = []
# number of vehicles to exit the network for every time-step
self._num_arrived = []
self._arrived_ids = []
self._arrived_rl_ids = []
# whether or not to automatically color vehicles
try:
self._color_by_speed = sim_params.color_by_speed
self._force_color_update = sim_params.force_color_update
except AttributeError:
self._force_color_update = False
# old speeds used to compute accelerations
self.previous_speeds = {}
def __init__(self, master_kernel, sim_params):
"""See parent class.
Copied from `flow.core.kernel.vehicle.traci`"""
KernelVehicle.__init__(self, master_kernel, sim_params)
self.__ids = [] # ids of all vehicles
self.__human_ids = [] # ids of human-driven vehicles
self.__controlled_ids = [] # ids of flow-controlled vehicles
self.__controlled_lc_ids = [] # ids of flow lc-controlled vehicles
self.__rl_ids = [] # ids of rl-controlled vehicles
self.__observed_ids = [] # ids of the observed vehicles
# vehicles: Key = Vehicle ID, Value = Dictionary describing the vehicle
# Ordered dictionary used to keep neural net inputs in order
self.__vehicles = collections.OrderedDict()
# create a sumo_observations variable that will carry all information
# on the state of the vehicles for a given time step
self.__sumo_obs = {}
# total number of vehicles in the network
self.num_vehicles = 0
# number of rl vehicles in the network
self.num_rl_vehicles = 0
# contains the parameters associated with each type of vehicle
self.type_parameters = {}
# list of vehicle ids located in each edge in the network
self._ids_by_edge = dict()
# number of vehicles that entered the network for every time-step
self._num_departed = []
self._departed_ids = []
# number of vehicles to exit the network for every time-step
self._num_arrived = []
self._arrived_ids = []
class Kernel(object):
"""Kernel for abstract function calling across traffic simulator APIs.
The kernel contains four different subclasses for distinguishing between
the various components of a traffic simulator.
* simulation: controls starting, loading, saving, advancing, and resetting
a simulation in Flow (see flow/core/kernel/simulation/base.py)
* scenario: stores network-specific information (see
flow/core/kernel/scenario/base.py)
* vehicle: stores and regularly updates vehicle-specific information. At
times, this class is optimized to efficiently collect information from
the simulator (see flow/core/kernel/vehicle/base.py).
* traffic_light: stores and regularly updates traffic light-specific
information (see flow/core/kernel/traffic_light/base.py).
The above kernel subclasses are designed specifically to support
simulator-agnostic state information calling. For example, if you would
like to collect the vehicle speed of a specific vehicle, then simply type:
>>> k = Kernel(simulator="...") # a kernel for some simulator type
>>> veh_id = "..." # some vehicle ID
>>> speed = k.vehicle.get_speed(veh_id)
In addition, these subclasses support sending commands to the simulator via
its API. For example, in order to assign a specific vehicle a target
acceleration, type:
>>> k = Kernel(simulator="...") # a kernel for some simulator type
>>> veh_id = "..." # some vehicle ID
>>> k.vehicle.apply_acceleration(veh_id)
These subclasses can be modified and recycled to support various different
traffic simulators, e.g. SUMO, AIMSUN, TruckSim, etc...
"""
def __init__(self, simulator, sim_params):
"""Instantiate a Flow kernel object.
Parameters
----------
simulator : str
simulator type, must be one of {"traci"}
sim_params : flow.core.params.SimParams
simulation-specific parameters
Raises
------
ValueError
if the specified input simulator is not a valid type
"""
self.kernel_api = None
if simulator == "traci":
self.simulation = TraCISimulation(self)
self.scenario = TraCIScenario(self)
self.vehicle = KernelVehicle(self, sim_params)
self.traffic_light = TraCITrafficLight(self)
else:
raise ValueError('Simulator type "{}" is not valid.'.
format(simulator))
def pass_api(self, kernel_api):
"""Pass the kernel API to all kernel subclasses."""
self.kernel_api = kernel_api
self.simulation.pass_api(kernel_api)
self.scenario.pass_api(kernel_api)
self.vehicle.pass_api(kernel_api)
self.traffic_light.pass_api(kernel_api)
def update(self, reset):
"""Update the kernel subclasses after a simulation step.
This is meant to support optimizations in the performance of some
simulators. For example, this step allows the vehicle subclass in the
"traci" simulator uses the ``update`` method to collect and store
subscription information.
Parameters
----------
reset : bool
specifies whether the simulator was reset in the last simulation
step
"""
# self.scenario.update(reset)
self.simulation.update(reset)
# self.vehicle.update(reset)
self.traffic_light.update(reset)
def close(self):
"""Terminate all components within the simulation and scenario."""
self.scenario.close()
self.simulation.close()