def set_gamepad(control, throttle, breakk):
if control == -1: # stop the car
client.sendMessage(Commands(0.0, 1, control))
return
client.sendMessage(Commands(float(throttle), float(breakk),
float(control)))
def work(self):
"""
Pretend this worker method does work that takes a long time. During this time, the thread's
event loop is blocked, except if the application's processEvents() is called: this gives every
thread (incl. main) a chance to process events, which in this sample means processing signals
received from GUI (such as abort).
"""
thread_name = QThread.currentThread().objectName()
thread_id = int(
QThread.currentThreadId()) # cast to int() is necessary
self.sig_msg.emit('Running worker #{} from thread "{}" (#{})'.format(
self.__id, thread_name, thread_id))
# Creates a new connection to DeepGTAV using the specified ip and port.
# If desired, a dataset path and compression level can be set to store in memory all the data received in a gziped pickle file.
# We don't want to save a dataset in this case
self.client = Client(ip=self.args.host, port=self.args.port)
# self.client = Client(ip="127.0.0.1", port=8000)
# We set the scenario to be in manual driving, and everything else random (time, weather and location).
# See deepgtav/messages.py to see what options are supported
scenario = Scenario(drivingMode=-1) #manual driving
# Send the Start request to DeepGTAV. Dataset is set as default, we only receive frames at 10Hz (320, 160)
self.client.sendMessage(Start(scenario=scenario))
# Dummy agent
model = Model()
# Start listening for messages coming from DeepGTAV. We do it for 80 hours
stoptime = time.time() + 80 * 3600
while (time.time() < stoptime and (not self.__abort)):
# We receive a message as a Python dictionary
app.processEvents()
message = self.client.recvMessage()
# The frame is a numpy array that can we pass through a CNN for example
image = frame2numpy(message['frame'], (320, 160))
commands = model.run(image)
self.sig_step.emit(self.__id, 'step ' + str(time.time()))
self.sig_image.emit(image.tolist())
# We send the commands predicted by the agent back to DeepGTAV to control the vehicle
self.client.sendMessage(
Commands(commands[0], commands[1], commands[2]))
# We tell DeepGTAV to stop
self.client.sendMessage(Stop())
self.client.close()
self.sig_done.emit(self.__id)
# We set the scenario to be in manual driving, and everything else random (time, weather and location).
# See deepgtav/messages.py to see what options are supported
scenario = Scenario(drivingMode=-1) #manual driving
# Send the Start request to DeepGTAV. Dataset is set as default, we only receive frames at 10Hz (320, 160)
client.sendMessage(Start(scenario=scenario))
# Dummy agent
model = Model()
# Start listening for messages coming from DeepGTAV. We do it for 80 hours
stoptime = time.time() + 80 * 3600
while time.time() < stoptime:
try:
# We receive a message as a Python dictionary
message = client.recvMessage()
print(message)
# The frame is a numpy array that can we pass through a CNN for example
image = frame2numpy(message['frame'], (320, 160))
commands = model.run(image)
# We send the commands predicted by the agent back to DeepGTAV to control the vehicle
client.sendMessage(Commands(commands[0], commands[1], commands[2]))
except KeyboardInterrupt:
break
# We tell DeepGTAV to stop
client.sendMessage(Stop())
client.close()
count = 0
print("Starting Loop...")
while True:
try:
# Collect and preprocess image
message = client.recvMessage()
image = frame2numpy(message['frame'], (320, 160))
image = ((image / 255) - .5) * 2
# Corrects for model input shape
model_input = []
model_input.append(image)
# Converts classification to float for steering input
category_prediction = np.argmax(model.predict(np.array(model_input)))
decimal_prediction = (category_prediction - 500) / 500
print('Category: ' + str(category_prediction) + ' Decimal: ' +
str(decimal_prediction))
client.sendMessage(Commands(
0.0, 0.0, decimal_prediction *
3)) # Mutiplication scales decimal prediction for harder turning
count += 1
except Exception as e:
print("Excepted as: " + str(e))
continue
client.sendMessage(Stop()) # Stops DeepGTAV
client.close()
print('lane controler takes: ', time.time() - t0)
t0 = time.time()
vehicle_controler.input(tracked_coord, speed_tracked,
s_tracked_coord, pitch_lidar,
roll_lidar, yaw_rate, steering_angle,
gas_brake, speed_now, acc, target_lane)
command = vehicle_controler.run()
print('sttering: ', steering_angle)
##print('Command: ', command) # in camera coord, turn left is +
##print('Comannd now: ', steering_angle)
print('vehicle controler takes: ', time.time() - t0)
client.sendMessage(
Commands(avoid_dead_zone(command[0]),
avoid_dead_zone(command[1]), -avoid_dead_zone(
command[2]))) # steering is opposite for GTAV
#client.sendMessage(Commands(0.8, 0, 0.3))
all_calculation_time = time.time() - time_now
# monitor
if reference_x is not None:
if 1: # for movement detectoin
marker_ground_ref = GE.ground_marker_2_image(
ground_marker, reference_x, reference_y)
image_monitor = IMG.lidar_result_fusion(
image_2, marker, marker_ground_ref)
#if lanes_3d is not None:
# image_monitor = IMG.test_show_lidar_line(image_monitor, lanes_3d, focus)
#ax + b = y
xx = -(yy - (359 + 1 / steering * 320)) * steering
else:
xx = 320
cv2.line(img, (320, 319), (int(xx), int(yy)), [255, 0, 0], 2)
cv2.imshow('binary_image', img)
cv2.waitKey(1)
pass
except Exception as e:
raise e
if image_source == USE_GTAV:
client.sendMessage(
Commands(throttle, breaker, steering)
) # Mutiplication scales decimal prediction for harder turning
print('loop time: {:.5f}s'.format(time.time() - t_loop_start))
except KeyboardInterrupt:
break
except Exception as e:
print("Excepted as: " + str(e))
#if image_source == USE_GTAV:
# client.sendMessage(Stop()) # Stops DeepGTAV
# client.close()
raise e
continue
if image_source == USE_GTAV:
client.sendMessage(Stop()) # Stops DeepGTAV
count = 0
print("Starting Loop...")
while True:
try:
# Collect and preprocess image
# message = client.recvMessage()
# window title 을 이용해 받아온 window 에서 frame 을 따온다.
# 기존의 기능에서 이를 제대로 지원하지 않는듯 하기 때문에 추가로 작성함.
frame_image = screenshot(hwnd=hwnd_list[0])
# frame_image = normalize(frame_image)
frame_image = ((frame_image / 255) - .5) * 2 # Simple preprocessing
# Corrects for model input shape
# frame_image = image.img_to_array(frame_image)
frame_image = np.reshape(frame_image, (1,) + frame_image.shape)
prediction = model.predict(frame_image)
steering = prediction[0][0]
steering = float(steering)
print("predicted steering : " + str(steering))
client.sendMessage(Commands(0.0, 0.0, steering=steering))
# Mutiplication scales decimal prediction for harder turning
count += 1
except Exception as e:
print("Excepted as: " + str(e))
continue
client.sendMessage(Stop()) # Stops DeepGTAV
client.close()
def send_message(self, throttle, brake, steering):
self.client.sendMessage(Commands(throttle, brake, steering))
pass
# brake_prediction = np.argmax(nn_output[36:85])
# brake_prediction = brake_prediction / 50
duration = time.time() - lasttime
print('Predicted Steering Value :', str(steering_prediction))
print('Prediction Probability :', str(steering_probability))
if steering_prediction > 0.08:
print('Right')
elif steering_prediction < -0.05:
print('Left')
else:
print('Straight')
# print('FPS :', str(1/(duration)))
# Mutiplication scales decimal prediction for harder turning
client.sendMessage(
Commands(0.3, 0.0,
2 / (1 + 2.718**(-5.5 * steering_prediction)) - 1))
# client.sendMessage(Commands(0.3, 0.0, steering_prediction))
count += 1
degrees = (2 / (1 + 2.718**(-5.5 * steering_prediction)) - 1) * 180
smoothed_angle += 0.2 * pow(abs(
(degrees - smoothed_angle)), 2.0 / 3.0) * (
degrees - smoothed_angle) / abs(degrees - smoothed_angle)
M = cv2.getRotationMatrix2D((cols / 2, rows / 2), -smoothed_angle, 1)
dst = cv2.warpAffine(img, M, (cols, rows))
cv2.imshow("steering wheel", dst)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
except Exception as e:
print("Excepted as: " + str(e))
interval = 0
while (sqrt((location[0]-args.des[0])**2+(location[1]-args.des[1])**2)>100) and time.time()<stoptime:
try:
message = client.recvMessage()
start = time.time()
# message is python dict; keys: timestamp, trafficSigns, speed,
# frame
if (message['speed']>1) and (manual == False) and time.time()-labelchange[-1]>5:
manual = True
labelchange.append(time.time())
if message['speed']<3:
interval = 0.5
else:
interval = 0.25
random_steering = np.random.uniform(0.5, 0.7) * ((-1.0) ** (np.random.randint(2)))
client.sendMessage(Commands(throttle=max(message['throttle'],0.3), brake=0.0, steering=random_steering, manual=1))
if manual:
print(message['throttle'], message['brake'], message['steering'])
client.sendMessage(Commands(throttle=max(message['throttle'],0.3), brake=0.0, steering=random_steering, manual=-1))
if time.time()-labelchange[-1]>interval:
manual=False
labelchange.append(time.time())
client.sendMessage(Commands(throttle=1.0, brake=0.0, steering=1, manual=0))
if manual:
print("*"*50)
location = message['location']
if message is None:
stoptime = time.time()
# The frame is a numpy array and can be displayed using OpenCV
# or similar
# image = frame2numpy(message['frame'],
yawRate=args.yawRate,
location=args.if_location,
time=args.dataset_time,
roadinfo=args.roadinfo,
direction=args.des)
scenario = Scenario(vehicle=args.vehicle,
weather=weather,
time=args.time,
location=args.location)
client.sendMessage(Start(dataset=dataset, scenario=scenario))
time.sleep(5) # wait for the scene to be ready
# Setup multi-threading for receiving messages
queue = Queue(maxsize=1)
manager = Manager()
c = BufferClient(('localhost', 8766))
capture_gta_to_queue_only_final(c, queue)
while True:
imgtime, img = queue.get()
imgtime = int(time)
speed, angle = model(Variable(torch.Tensor(img)))
throttle, brake, steering = pid(speed, angle)
client.sendMessage(
Commands(throttle=throttle,
brake=brake,
steering=steering,
manual=1))
del queue
client.close()
def send_message(self, throttle, brake, steering, gear):
self.client.sendMessage(Commands(throttle, brake, steering,
gear)) #yyz
pass
