def run(agent_class,
agent_config_file_path: Path,
carla_config_file_path: Path,
num_laps: int = 10) -> Tuple[float, int, int]:
"""
Run the agent along the track and produce a score based on certain metrics
Args:
num_laps: int number of laps that the agent should run
agent_class: the participant's agent
agent_config_file_path: agent configuration path
carla_config_file_path: carla configuration path
Returns:
float between 0 - 1 representing scores
"""
agent_config: AgentConfig = AgentConfig.parse_file(agent_config_file_path)
carla_config = CarlaConfig.parse_file(carla_config_file_path)
# hard code agent config such that it reflect competition requirements
agent_config.num_laps = num_laps
carla_runner = CarlaRunner(carla_settings=carla_config,
agent_settings=agent_config,
npc_agent_class=PurePursuitAgent,
competition_mode=True,
lap_count=num_laps)
try:
my_vehicle = carla_runner.set_carla_world()
agent = agent_class(vehicle=my_vehicle, agent_settings=agent_config)
carla_runner.start_game_loop(agent=agent, use_manual_control=False)
return compute_score(carla_runner)
except Exception as e:
print(f"something bad happened during initialization: {e}")
carla_runner.on_finish()
logging.error(f"{e}. Might be a good idea to restart Server")
return 0, 0, False
def main(output_folder_path: Path):
# Set gym-carla environment
agent_config = AgentConfig.parse_file(
Path("configurations/agent_configuration.json"))
carla_config = CarlaConfig.parse_file(
Path("configurations/carla_configuration.json"))
params = {
"agent_config": agent_config,
"carla_config": carla_config,
"ego_agent_class": RLOccuMapE2ETrainingAgent,
"max_collision": 5,
}
env = gym.make('roar-occu-map-e2e-v0', params=params)
env.reset()
model_params: dict = {
"verbose": 1,
"env": env,
"render": True,
"tensorboard_log": (output_folder_path / "tensorboard").as_posix(),
"buffer_size": 10000,
"nb_rollout_steps": 100,
# "batch_size": 16,
"nb_eval_steps": 50
}
model, callbacks = setup(model_params, output_folder_path)
model = model.learn(total_timesteps=int(1e6),
callback=callbacks,
reset_num_timesteps=False)
def run(agent_class, agent_config_file_path: Path,
carla_config_file_path: Path) -> Tuple[float, int, bool]:
"""
Run the agent along the track and produce a score based on certain metrics
Args:
agent_class: the participant's agent
agent_config_file_path: agent configuration path
carla_config_file_path: carla configuration path
Returns:
float between 0 - 1 representing scores
"""
agent_config = AgentConfig.parse_file(agent_config_file_path)
carla_config = CarlaConfig.parse_file(carla_config_file_path)
carla_runner = CarlaRunner(carla_settings=carla_config,
agent_settings=agent_config,
npc_agent_class=PurePursuitAgent,
competition_mode=True,
max_collision=3)
try:
my_vehicle = carla_runner.set_carla_world()
agent = agent_class(vehicle=my_vehicle, agent_settings=agent_config)
carla_runner.start_game_loop(agent=agent, use_manual_control=True)
return compute_score(carla_runner)
except Exception as e:
print(f"something bad happened during initialization: {e}")
carla_runner.on_finish()
logging.error(f"{e}. Might be a good idea to restart Server")
return 0, 0, False
def main():
agent_config = AgentConfig.parse_file(
Path("./ROAR_Sim/configurations/agent_configuration.json"))
carla_config = CarlaConfig.parse_file(
Path("./ROAR_Sim/configurations/configuration.json"))
carla_runner = CarlaRunner(carla_settings=carla_config,
agent_settings=agent_config,
npc_agent_class=PurePursuitAgent)
''' Data collection code. Currently unnecessary
# make csv file to store some data in
# we have current position x, y, z, current velocity x, y, z, next waypoint position x, y, z,
# next waypoint direction relative to the current position of the car x, y, z, steering, and throttle
csvNotes = "{}\n{}\n".format("we have current car position x, y, z, current car velocity x, y, z, next waypoint position x, y, z,",
"next waypoint direction relative to the current position of the car x, y, z, steering, and throttle")
csvHeader = "px, py, pz, vx, vy, vz, wpx, wpy, wpz, wvx, wvy, wvz, steering, throttle\n"
with open("tmp/pid_data.csv", "w") as f:
f.write(csvNotes)
f.write(csvHeader)
'''
try:
my_vehicle = carla_runner.set_carla_world()
# agent = PIDAgent(vehicle=my_vehicle, agent_settings=agent_config)
agent = LQRAgent(vehicle=my_vehicle, agent_settings=agent_config)
carla_runner.start_game_loop(agent=agent, use_manual_control=False)
except Exception as e:
logging.error(f"Something bad happened during initialization: {e}")
carla_runner.on_finish()
logging.error(f"{e}. Might be a good idea to restart Server")
def main(output_folder_path:Path):
# Set gym-carla environment
agent_config = AgentConfig.parse_file(Path("configurations/agent_configuration.json"))
carla_config = CarlaConfig.parse_file(Path("configurations/carla_configuration.json"))
params = {
"agent_config": agent_config,
"carla_config": carla_config,
"ego_agent_class": RLPIDAgent,
"max_collision": 5
}
env = gym.make('roar-pid-v0', params=params)
env.reset()
model_params: dict = {
"verbose": 1,
"render": True,
"tensorboard_log": (output_folder_path / "tensorboard").as_posix()
}
latest_model_path = find_latest_model(output_folder_path)
if latest_model_path is None:
model = DDPG(LnMlpPolicy, env=env, **model_params) # full tensorboard log can take up space quickly
else:
model = DDPG.load(latest_model_path, env=env, **model_params)
model.render = True
model.tensorboard_log = (output_folder_path / "tensorboard").as_posix()
logging_callback = LoggingCallback(model=model)
checkpoint_callback = CheckpointCallback(save_freq=1000, verbose=2, save_path=(output_folder_path / "checkpoints").as_posix())
event_callback = EveryNTimesteps(n_steps=100, callback=checkpoint_callback)
callbacks = CallbackList([checkpoint_callback, event_callback, logging_callback])
model = model.learn(total_timesteps=int(1e10), callback=callbacks, reset_num_timesteps=False)
model.save(f"pid_ddpg_{datetime.now()}")
def __init__(self):
# Set gym-carla environment
agent_config = AgentConfig.parse_file(
Path("configurations/agent_configuration.json"))
carla_config = CarlaConfig.parse_file(
Path("configurations/carla_configuration.json"))
params = {
"agent_config": agent_config,
"carla_config": carla_config,
"ego_agent_class": RLOccuMapE2ETrainingAgent,
"max_collision": 5,
}
super().__init__(params)
# action space = next waypoint
self.view_size = 200
self.max_steering_angle = 1
self.action_space = gym.spaces.Box(
low=np.array([0.4, -self.max_steering_angle]),
high=np.array([1, self.max_steering_angle]),
dtype=np.float32) # throttle, steering
self.observation_space = gym.spaces.Box(low=0,
high=1,
shape=(self.view_size,
self.view_size, 1),
dtype=np.uint8)
self.debug_info: OrderedDict = OrderedDict()
self.prev_location: Optional[Location] = None
self.prev_next_waypoint: Optional[Location] = None
self.dist_diff = 0
def main(output_folder_path: Path):
# Set gym-carla environment
agent_config = AgentConfig.parse_file(
Path("configurations/agent_configuration.json"))
carla_config = CarlaConfig.parse_file(
Path("configurations/carla_configuration.json"))
params = {
"agent_config": agent_config,
"carla_config": carla_config,
"ego_agent_class": RLLocalPlannerAgent,
"max_collision": 5,
}
env = gym.make('roar-local-planner-v0', params=params)
env.reset()
model_params: dict = {
"verbose": 1,
"render": True,
"env": env,
"n_cpu_tf_sess": None,
"buffer_size": 1000,
"nb_train_steps": 50,
"nb_rollout_steps": 100,
# "nb_eval_steps": 50,
"batch_size": 32,
}
latest_model_path = find_latest_model(Path(output_folder_path))
if latest_model_path is None:
model = DDPG(CnnPolicy, **model_params)
else:
model = DDPG.load(latest_model_path, **model_params)
tensorboard_dir = (output_folder_path / "tensorboard")
ckpt_dir = (output_folder_path / "checkpoints")
tensorboard_dir.mkdir(parents=True, exist_ok=True)
ckpt_dir.mkdir(parents=True, exist_ok=True)
model.tensorboard_log = tensorboard_dir.as_posix()
model.render = True
logging_callback = LoggingCallback(model=model)
checkpoint_callback = CheckpointCallback(save_freq=1000,
verbose=2,
save_path=ckpt_dir.as_posix())
event_callback = EveryNTimesteps(n_steps=100, callback=checkpoint_callback)
callbacks = CallbackList(
[checkpoint_callback, event_callback, logging_callback])
model = model.learn(total_timesteps=int(1e10),
callback=callbacks,
reset_num_timesteps=False)
model.save(f"local_planner_ddpg_{datetime.now()}")
def main():
agent_config = AgentConfig.parse_file(Path("../ROAR/configurations/carla_configuration.json"))
carla_config = CarlaConfig.parse_file(Path("../ROAR_Sim/configurations/carla_configuration.json"))
carla_runner = CarlaRunner(carla_settings=carla_config, agent_settings=agent_config,
npc_agent_class=PurePursuitAgent)
try:
my_vehicle = carla_runner.set_carla_world()
agent = PointCloudMapRecordingAgent(vehicle=my_vehicle, agent_settings=agent_config)
carla_runner.start_game_loop(agent=agent, use_manual_control=False)
except Exception as e:
print("Ending abnormally: ", e)
carla_runner.on_finish()
logging.error(f"Hint: Might be a good idea to restart Server. ")
def main(output_folder_path: Path):
# Set gym-carla environment
agent_config = AgentConfig.parse_file(
Path("configurations/agent_configuration.json"))
carla_config = CarlaConfig.parse_file(
Path("configurations/carla_configuration.json"))
params = {
"agent_config": agent_config,
"carla_config": carla_config,
"ego_agent_class": RLLocalPlannerAgent,
"max_collision": 5,
}
env = gym.make('roar-local-planner-v1', params=params)
env.reset()
tensorboard_dir, ckpt_dir = prep_dir(output_folder_path)
model_params: dict = {
"verbose": 1,
"render": True,
"env": env,
"n_cpu_tf_sess": 2,
"buffer_size": 10,
"random_exploration": 0.1,
"tensorboard_log": tensorboard_dir.as_posix(),
}
latest_model_path = find_latest_model(Path(output_folder_path))
if latest_model_path is None:
model = DDPG(
LnMlpPolicy,
**model_params) # full tensorboard log can take up space quickly
else:
model = DDPG.load(latest_model_path, **model_params)
logging_callback = LoggingCallback(model=model)
checkpoint_callback = CheckpointCallback(save_freq=1000,
verbose=2,
save_path=ckpt_dir.as_posix())
event_callback = EveryNTimesteps(n_steps=100, callback=checkpoint_callback)
callbacks = CallbackList(
[checkpoint_callback, event_callback, logging_callback])
model = model.learn(total_timesteps=int(1e10),
callback=callbacks,
reset_num_timesteps=False)
model.save(f"local_planner_v1_ddpg_{datetime.now()}")
def main():
agent_config = AgentConfig.parse_file(
Path("./ROAR_Sim/configurations/agent_configuration.json"))
carla_config = CarlaConfig.parse_file(
Path("./ROAR_Sim/configurations/configuration.json"))
carla_runner = CarlaRunner(carla_settings=carla_config,
agent_settings=agent_config,
npc_agent_class=PurePursuitAgent)
try:
my_vehicle = carla_runner.set_carla_world()
agent = PIDAgent(vehicle=my_vehicle, agent_settings=agent_config)
carla_runner.start_game_loop(agent=agent, use_manual_control=False)
except Exception as e:
logging.error(f"Something bad happened during initialization: {e}")
carla_runner.on_finish()
logging.error(f"{e}. Might be a good idea to restart Server")
def main():
"""Starts game loop"""
agent_config = AgentConfig.parse_file(
Path("./ROAR_Sim/configurations/agent_configuration.json"))
carla_config = CarlaConfig.parse_file(
Path("./ROAR_Sim/configurations/configuration.json"))
carla_runner = CarlaRunner(carla_settings=carla_config,
agent_settings=agent_config,
npc_agent_class=PurePursuitAgent)
try:
my_vehicle = carla_runner.set_carla_world()
#agent = PIDAgent(vehicle=my_vehicle, agent_settings=agent_config)
#agent = OccupancyMapAgent(vehicle=my_vehicle, agent_settings=agent_config)
#agent = PurePursuitAgent(vehicle=my_vehicle, agent_settings=agent_config)
#agent = JAM1Agent_old(vehicle=my_vehicle, agent_settings=agent_config) # *** roll controller
agent = PIDRollAgent(
vehicle=my_vehicle,
agent_settings=agent_config) # *** roll controller
#agent = JAM1Agent(vehicle=my_vehicle, agent_settings=agent_config)
#agent = JAM2Agent(vehicle=my_vehicle, agent_settings=agent_config)
#agent = JAM3Agent_old(vehicle=my_vehicle, agent_settings=agent_config) # *** bstanley
#agent = JAM3Agent(vehicle=my_vehicle, agent_settings=agent_config)
# *** use to record new waypoints ***
# waypointrecord = agent.bstanley_controller.blat_stanley_controller.waypointrecord
# np.save("James_waypoints", np.array(waypointrecord))
#agent = RecordingAgent(vehicle=my_vehicle, agent_settings=agent_config)
#carla_runner.start_game_loop(agent=agent, use_manual_control=True)#*******True for manual control, False auto
carla_runner.start_game_loop(
agent=agent, use_manual_control=False
) # *******True for manual control, False auto
except Exception as e:
logging.error(f"Something bad happened during initialization: {e}")
carla_runner.on_finish()
logging.error(f"{e}. Might be a good idea to restart Server")
def __init__(self, app: QtWidgets.QApplication,
sim_json_config_file_path: Path, **kwargs):
super().__init__(app, Ui_SimulationConfigWindow, **kwargs)
self.setting_list = []
self.setting_dict = dict()
self.dialogs = list()
self.simulation_config = SimulationConfig()
self.json_config_file_path: Path = sim_json_config_file_path
self.model_info: Dict[str,
PydanticModelEntry] = self.fill_config_list()
# =====================================================================================================
# create save button and add to menu
self.ui.actionSave = QtWidgets.QAction(self)
self.ui.actionSave.setObjectName("actionSave")
self.ui.actionSave.setText(
QtCore.QCoreApplication.translate("SimulationConfigWindow",
"Save"))
self.ui.menuFile.addAction(self.ui.actionSave)
self.ui.actionSave.triggered.connect(self.save_config)
class SimConfigWindow(BaseWindow):
def __init__(self, app: QtWidgets.QApplication,
sim_json_config_file_path: Path, **kwargs):
super().__init__(app, Ui_SimulationConfigWindow, **kwargs)
self.setting_list = []
self.setting_dict = dict()
self.dialogs = list()
self.simulation_config = SimulationConfig()
self.json_config_file_path: Path = sim_json_config_file_path
self.model_info: Dict[str,
PydanticModelEntry] = self.fill_config_list()
# =====================================================================================================
# create save button and add to menu
self.ui.actionSave = QtWidgets.QAction(self)
self.ui.actionSave.setObjectName("actionSave")
self.ui.actionSave.setText(
QtCore.QCoreApplication.translate("SimulationConfigWindow",
"Save"))
self.ui.menuFile.addAction(self.ui.actionSave)
self.ui.actionSave.triggered.connect(self.save_config)
# =====================================================================================================
# self.menubar.addAction(self.menuFile.menuAction())
# self.text_change()
#####
# self.currentTextChanged.connect(self.onCurrentTextChanged)
# def onCurrentTextChanged(self, text):
# print("\n text changed \n")
def fill_config_list(self) -> Dict[str, Any]:
model_info: Dict[str, Any] = dict()
for key_name, entry in self.simulation_config.schema(
)['properties'].items():
if "type" not in entry:
continue
model_info[key_name] = entry[
'title'] # PydanticModelEntry.parse_obj(entry)
# pprint(entry)
json_dict: dict = json.load(self.json_config_file_path.open('r'))
for key_name, entry in json_dict.items():
pass
model_info[key_name] = entry
# TODO update model_info, completed
# print(key_name, entry)
model_values = self.simulation_config.dict()
# print("len: ",len(model_info.items()))
##print(self.test_input_list)
# print(self.setting_dict)
for name, entry in model_values.items():
# TODO do not populate if it is not of type [int, float, string, bool]
if name in model_info.keys():
entry = model_info[name]
# type check -> present if type in ['int','float','string','bool']
try:
if isinstance(eval(str(entry)), dict):
continue
except:
pass
self.add_entry_to_settings_gui(name=name, value=entry)
# self.test_input_list.append([name,str(curr_value)])
self.setting_dict[name] = entry
# self.logger.debug()
return model_info
def add_entry_to_settings_gui(self, name: str, value: Union[str, int,
float, bool]):
label = QtWidgets.QLabel()
label.setText(name)
input_field = QtWidgets.QLineEdit() # QTextEdit()
input_field.setText(str(value))
input_field.setObjectName(name)
self.setting_list.append(input_field)
# input_field.textChanged.connect(self.on_change)
# self.test_input_list.append(str(value))
self.ui.formLayout.addRow(label, input_field)
def set_listener(self):
super(SimConfigWindow, self).set_listener()
self.ui.pushButton_confirm.clicked.connect(self.pushButton_confirm)
# def on_change(self,text):
# print("text changed: ", text)
# print(self.setting_list[1].text())
# print(self.setting_list[1].objectName())
# print(self.setting_dict)
def isfloat(self, value):
try:
float(value)
return True
except ValueError:
return False
def pushButton_confirm(self):
self.auto_wire_window(ControlPanelWindow)
# self.save_config()
def save_config(self):
"""
save config to file
"""
sim_config_json = {}
for widget in self.setting_list:
# if isinstance(widget, QtWidgets.QLineEdit):
# print(widget.objectName(),":", widget.text())
if widget.text().isnumeric():
sim_config_json[widget.objectName()] = int(widget.text())
elif self.isfloat(widget.text()):
sim_config_json[widget.objectName()] = float(widget.text())
elif widget.text().lower() == "true":
sim_config_json[widget.objectName()] = True
elif widget.text().lower() == "false":
sim_config_json[widget.objectName()] = False
else:
sim_config_json[widget.objectName()] = widget.text()
json_dict: dict = json.load(self.json_config_file_path.open('r'))
json_object = sim_config_json
for key, entry in json_object.items():
try:
json_dict[key] = eval(entry)
except:
json_dict[key] = entry
# current C:\Users\Zetian\Desktop\project\ROAR\ROAR_Desktop\ROAR_GUI
# need C:\Users\Zetian\Desktop\project\ROAR\ROAR_Sim\configurations
# pathlib
with open("../../ROAR_Sim/configurations/configuration.json",
"w") as outfile:
outfile.write(json.dumps(json_dict, indent=2))
def auto_wire_window(self, target_window):
target_app = target_window(self.app)
self.dialogs.append(target_app)
target_app.show()
self.hide()
target_app.show()
target_app.closeEvent = self.app_close_event
def app_close_event(self, close_event):
self.show()
def main():
"""Starts game loop"""
carla_config = CarlaConfig.parse_file(Path("./ROAR_Sim/configurations/configuration.json"))
agent_config = AgentConfig.parse_file(Path("./ROAR_Sim/configurations/agent_configuration.json"))
"""
Pit Stop:
Use different kinds of 'set' functions at PitStop to tune/fix your own car!
"""
pitstop = PitStop(carla_config, agent_config)
pitstop.set_carla_version(version = "0.9.9")
pitstop.set_carla_sync_mode(False)
pitstop.set_autopilot_mode(True)
pitstop.set_car_color(CarlaCarColor(r = 255,g = 200,b = 00,a = 255))
pitstop.set_num_laps(num=1)
pitstop.set_output_data_folder_path("./data/output")
pitstop.set_output_data_file_name(time.strftime("%Y%m%d-%H%M%S-") + "map-waypoints")
pitstop.set_max_speed(speed = 200)
pitstop.set_target_speed(speed = 30)
print(agent_config.target_speed, " target speed")
pitstop.set_steering_boundary(boundary = (-1.0, 1.0))
pitstop.set_throttle_boundary(boundary = (0, 0.5))
pitstop.set_waypoints_look_ahead_values(values={
"60": 5,
"80": 10,
"120": 20,
"180": 50})
pid_value = {
"longitudinal_controller": {
"40": {
"Kp": 0.8,
"Kd": 0.2,
"Ki": 0
},
"60": {
"Kp": 0.5,
"Kd": 0.2,
"Ki": 0
},
"150": {
"Kp": 0.2,
"Kd": 0.1,
"Ki": 0.1
}
},
"latitudinal_controller": {
"60": {
"Kp": 0.8,
"Kd": 0.1,
"Ki": 0.1
},
"100": {
"Kp": 0.6,
"Kd": 0.2,
"Ki": 0.1
},
"150": {
"Kp": 0.5,
"Kd": 0.2,
"Ki": 0.1
}
}
}
pitstop.set_pid_values(pid_value)
"""Passing configurations to Carla and Agent"""
carla_runner = CarlaRunner(carla_settings=carla_config, # ROAR Academy: fine
agent_settings=agent_config, # ROAR Academy: fine
npc_agent_class=PurePursuitAgent)
try:
my_vehicle = carla_runner.set_carla_world()
agent = PIDAgent(vehicle=my_vehicle, agent_settings=agent_config)
# agent = WaypointGeneratigAgent(vehicle=my_vehicle, agent_settings=agent_config)
carla_runner.start_game_loop(agent=agent, use_manual_control=False) # for PIDAgent
# carla_runner.start_game_loop(agent=agent, use_manual_control=True) # for WaypointGeneratingAgent
except Exception as e:
logging.error(f"Something bad happened during initialization: {e}")
carla_runner.on_finish()
logging.error(f"{e}. Might be a good idea to restart Server")
