def main():
# This will get number of pixels in each image (they must all be the same!)
imgsize = 0
# Read in images from car, convert to grayscale, scale down, and flatten for use as input
images = []
for k in range(SAFESIZE):
image = loadgray(IMAGEDIR + '/image%03d.png' % k)
imgsize = np.prod(image.shape)
images.append(image)
# All but last image is safe (01 = no-crash; 10 = crash)
targets = []
for k in range(SAFESIZE - 1):
targets.append([0, 1])
targets.append([1, 0])
# with tf.Graph().as_default():
x = tf.compat.v1.placeholder('float', [None, imgsize]) # car FPV images
y = tf.compat.v1.placeholder('float',
[None, 2]) # 01 = no-crash; 10 = crash
output = inference(x, imgsize, 2)
cost = loss(output, y)
global_step = tf.Variable(0, name='global_step', trainable=False)
train_op = training(cost, global_step)
sess = tf.compat.v1.Session()
init_op = tf.compat.v1.global_variables_initializer()
sess.run(init_op)
# Training cycle
for epoch in range(training_epochs):
# Fit training using batch data
sess.run(train_op, feed_dict={x: images, y: targets})
# Compute average loss
avg_cost = sess.run(cost, feed_dict={x: images, y: targets})
# Display logs per epoch step
if epoch % display_step == 0:
print('Epoch:', '%04d' % epoch, 'cost =',
'{:.9f}'.format(avg_cost))
print('Optimization Finished; saving weights to ' + PARAMFILE)
params = [
sess.run(param) for param in tf.compat.v1.trainable_variables()
]
pickle.dump(params, open(PARAMFILE, 'wb'))
# the vehicle will resume moving.
brakingCount = 0
# Loop until we detect a collision
while True:
# Get RGBA camera images from the car
responses = client.simGetImages(
[ImageRequest(1, AirSimImageType.Scene)])
# Save it to a temporary file
image = responses[0].image_data_uint8
AirSimClientBase.write_file(os.path.normpath(TMPFILE), image)
# Read-load the image as a grayscale array
image = loadgray(TMPFILE)
# Run the image through our inference engine.
# Engine returns a softmax output inside a list, so we grab the first
# element of the list (the actual softmax vector), whose second element
# is the absence of an obstacle.
safety = sess.run(output, feed_dict={x: [image]})[0][1]
# Slam on the brakes if it ain't safe!
if safety < 0.5:
if brakingCount > BRAKING_DURATION:
print('BRAKING TO AVOID COLLISSION')
sys.stdout.flush()
break
