def test_turn_location_position(self):
# Human assisted test
# Ensure that the car returns to the same location if we drive around in circles.
import visualizer
pose_estimator = PoseEstimator((0, 0, 0), 0, 0, 0, 0)
d = 4
steering_angle_radians = np.pi / d
outer_turn_radius_meters = pose_estimator.calc_outer_turn_radius(
steering_angle_radians)
front_wheel_turn_circumference = 2 * np.pi * config.FRONT_WHEEL_RADIUS_METERS
turn_circle_circumference = 2 * np.pi * outer_turn_radius_meters
ticks_required = config.ENCODER_RESOLUTION_FRONT_WHEEL * turn_circle_circumference / front_wheel_turn_circumference
result_loc = []
ticks = 0
for i in range(2 * d):
result_loc.append(
pose_estimator.estimate(time=i,
steering_angle=steering_angle_radians,
encoder_ticks=ticks,
angular_velocity=0))
ticks = ticks + ticks_required / (2 * d)
for loc in result_loc:
visualizer.draw_car(loc[0], loc[1], loc[2])
visualizer.show()
def main():
# Reads from the data file and runs estimate for each row
# Then plots the trajectory
data_array = util.read_csv(config.DATASET_ABSOLUTE_PATH)
row = data_array[0]
time, encoder, angular_velocity, steering_angle = np.ravel(row)
resulting_pos_heading = []
pose_estimator = PoseEstimator((0, 0, 0), time, encoder, angular_velocity,
steering_angle)
i = 1
while i < len(data_array):
row = data_array[i]
time, encoder, angular_velocity, steering_angle = np.ravel(row)
x, y, heading = pose_estimator.estimate(
time=time,
steering_angle=steering_angle,
encoder_ticks=encoder,
angular_velocity=angular_velocity)
resulting_pos_heading.append([x, y, heading])
i = i + 1
visualizer.plot_points(
np.asarray(resulting_pos_heading)[:, 0],
np.asarray(resulting_pos_heading)[:, 1])
visualizer.show()
for _ in range(10):
a = time.time()
sess.run([
net.get_output(
name=last_layer.format(stage=args.stage_level, aux=1)),
net.get_output(
name=last_layer.format(stage=args.stage_level, aux=2))
],
feed_dict={'image:0': [image]})
logging.info('inference- elapsed_time={}'.format(time.time() - a))
avg += time.time() - a
logging.info('prediction avg= %f' % (avg / 10))
logging.info('pose+')
a = time.time()
humans = PoseEstimator.estimate(heatMat, pafMat)
logging.info('pose- elapsed_time={}'.format(time.time() - a))
for human in humans:
res = write_coco_json(human, args.input_width, args.input_height)
print(res)
logging.info('image={} heatMap={} pafMat={}'.format(
image.shape, heatMat.shape, pafMat.shape))
process_img = CocoPose.display_image(image,
heatMat,
pafMat,
as_numpy=True)
# display
image = cv2.imread(args.imgpath)
image_h, image_w = image.shape[:2]
def test_straight_location_estimation(self):
# Testing the position for straight line motion
pose_estimator = PoseEstimator((0, 0, 0), 0, 0, 0, 0)
pose_estimator.estimate(0, 0, 0, 0)
self.assertEqual((2 * np.pi * config.FRONT_WHEEL_RADIUS_METERS, 0, 0),
pose_estimator.estimate(1, 0, 512, 10))