def calculate_ao_heading_vector(self):
"""
:return: An obstacle avoidance vector in the robot's reference frame
and vectors to detected obstacles in the robot's reference frame
"""
# initialize vector
obstacle_vectors = [[0.0, 0.0]] * len(self.proximity_sensor_placements)
ao_heading_vector = [0.0, 0.0]
# get the distances indicated by the robot's sensor readings
sensor_distances = self.supervisor.proximity_sensor_distances()
# calculate the position of detected obstacles and find an avoidance vector
robot_pos, robot_theta = self.supervisor.estimated_pose().vunpack()
for i in range(len(sensor_distances)):
# calculate the position of the obstacle
sensor_pos, sensor_theta = self.proximity_sensor_placements[
i].vunpack()
vector = [sensor_distances[i], 0.0]
vector = linalg.rotate_and_translate_vector(
vector, sensor_theta, sensor_pos)
obstacle_vectors[
i] = vector # store the obstacle vectors in the robot's reference frame
# accumluate the heading vector within the robot's reference frame
ao_heading_vector = linalg.add(
ao_heading_vector, linalg.scale(vector, self.sensor_gains[i]))
return ao_heading_vector, obstacle_vectors
def _bounding_circle( self ):
v = self._as_vector()
vhalf = linalg.scale( v, 0.5 )
c = linalg.add( self.vertexes[0], vhalf )
r = linalg.mag( v ) * 0.5
return c, r
def _bounding_circle( self ):
v = self._as_vector()
vhalf = linalg.scale( v, 0.5 )
c = linalg.add( self.vertexes[0], vhalf )
r = linalg.mag( v ) * 0.5
return c, r
def _draw_detection_to_frame(self):
target_delta = self.proximity_sensor.target_delta
if target_delta != None:
detector_endpoints = self.proximity_sensor.detector_line.vertexes
detector_vector = linalg.sub(detector_endpoints[1], detector_endpoints[0])
target_vector = linalg.add(detector_endpoints[0], linalg.scale(detector_vector, target_delta))
self.viewer.current_frame.add_circle(pos=target_vector, radius=0.02, color="black", alpha=0.7)
def calculate_fw_heading_vector( self, follow_direction ):
# get the necessary variables for the working set of sensors
# the working set is the sensors on the side we are bearing on, indexed from rearmost to foremost on the robot
# NOTE: uses preexisting knowledge of the how the sensors are stored and indexed
if follow_direction == FWDIR_LEFT:
# if we are following to the left, we bear on the righthand sensors
sensor_placements = self.proximity_sensor_placements[7:3:-1]
sensor_distances = self.supervisor.proximity_sensor_distances()[7:3:-1]
sensor_detections = self.supervisor.proximity_sensor_positive_detections()[7:3:-1]
elif follow_direction == FWDIR_RIGHT:
# if we are following to the right, we bear on the lefthand sensors
sensor_placements = self.proximity_sensor_placements[:4]
sensor_distances = self.supervisor.proximity_sensor_distances()[:4]
sensor_detections = self.supervisor.proximity_sensor_positive_detections()[:4]
else:
raise Exception( "unknown wall-following direction" )
if True not in sensor_detections:
# if there is no wall to track detected, we default to predefined reference points
# NOTE: these points are designed to turn the robot towards the bearing side, which aids with cornering behavior
# the resulting heading vector is also meant to point close to directly aft of the robot
# this helps when determining switching conditions in the supervisor state machine
p1 = [ -0.2, 0.0 ]
if follow_direction == FWDIR_LEFT: p2 = [ -0.2, -0.0001 ]
if follow_direction == FWDIR_RIGHT: p2 = [ -0.2, 0.0001 ]
else:
# sort the sensor distances along with their corresponding indices
sensor_distances, indices = zip( *sorted( zip( # this method ensures two different sensors are always used
sensor_distances, # sensor distances
[0, 1, 2, 3] # corresponding indices
) ) )
# get the smallest sensor distances and their corresponding indices
d1, d2 = sensor_distances[0:2]
i1, i2 = indices[0:2]
# calculate the vectors to the obstacle in the robot's reference frame
sensor1_pos, sensor1_theta = sensor_placements[i1].vunpack()
sensor2_pos, sensor2_theta = sensor_placements[i2].vunpack()
p1, p2 = [ d1, 0.0 ], [ d2, 0.0 ]
p1 = linalg.rotate_and_translate_vector( p1, sensor1_theta, sensor1_pos )
p2 = linalg.rotate_and_translate_vector( p2, sensor2_theta, sensor2_pos )
# ensure p2 is forward of p1
if i2 < i1: p1, p2 = p2, p1
# compute the key vectors and auxiliary data
l_wall_surface = [ p2, p1 ]
l_parallel_component = linalg.sub( p2, p1 )
l_distance_vector = linalg.sub( p1, linalg.proj( p1, l_parallel_component ) )
unit_perp = linalg.unit( l_distance_vector )
distance_desired = linalg.scale( unit_perp, self.follow_distance )
l_perpendicular_component = linalg.sub( l_distance_vector, distance_desired )
l_fw_heading_vector = linalg.add( l_parallel_component, l_perpendicular_component )
return l_fw_heading_vector, l_parallel_component, l_perpendicular_component, l_distance_vector, l_wall_surface
def calculate_fw_heading_vector( self, follow_direction ):
# get the necessary variables for the working set of sensors
# the working set is the sensors on the side we are bearing on, indexed from rearmost to foremost on the robot
# NOTE: uses preexisting knowledge of the how the sensors are stored and indexed
if follow_direction == FWDIR_LEFT:
# if we are following to the left, we bear on the righthand sensors
sensor_placements = self.proximity_sensor_placements[7:3:-1]
sensor_distances = self.supervisor.proximity_sensor_distances()[7:3:-1]
sensor_detections = self.supervisor.proximity_sensor_positive_detections()[7:3:-1]
elif follow_direction == FWDIR_RIGHT:
# if we are following to the right, we bear on the lefthand sensors
sensor_placements = self.proximity_sensor_placements[:4]
sensor_distances = self.supervisor.proximity_sensor_distances()[:4]
sensor_detections = self.supervisor.proximity_sensor_positive_detections()[:4]
else:
raise Exception( "unknown wall-following direction" )
if True not in sensor_detections:
# if there is no wall to track detected, we default to predefined reference points
# NOTE: these points are designed to turn the robot towards the bearing side, which aids with cornering behavior
# the resulting heading vector is also meant to point close to directly aft of the robot
# this helps when determining switching conditions in the supervisor state machine
p1 = [ -0.2, 0.0 ]
if follow_direction == FWDIR_LEFT: p2 = [ -0.2, -0.0001 ]
if follow_direction == FWDIR_RIGHT: p2 = [ -0.2, 0.0001 ]
else:
# sort the sensor distances along with their corresponding indices
sensor_distances, indices = zip( *sorted( zip( # this method ensures two different sensors are always used
sensor_distances, # sensor distances
[0, 1, 2, 3] # corresponding indices
) ) )
# get the smallest sensor distances and their corresponding indices
d1, d2 = sensor_distances[0:2]
i1, i2 = indices[0:2]
# calculate the vectors to the obstacle in the robot's reference frame
sensor1_pos, sensor1_theta = sensor_placements[i1].vunpack()
sensor2_pos, sensor2_theta = sensor_placements[i2].vunpack()
p1, p2 = [ d1, 0.0 ], [ d2, 0.0 ]
p1 = linalg.rotate_and_translate_vector( p1, sensor1_theta, sensor1_pos )
p2 = linalg.rotate_and_translate_vector( p2, sensor2_theta, sensor2_pos )
# ensure p2 is forward of p1
if i2 < i1: p1, p2 = p2, p1
# compute the key vectors and auxiliary data
l_wall_surface = [ p2, p1 ]
l_parallel_component = linalg.sub( p2, p1 )
l_distance_vector = linalg.sub( p1, linalg.proj( p1, l_parallel_component ) )
unit_perp = linalg.unit( l_distance_vector )
distance_desired = linalg.scale( unit_perp, self.follow_distance )
l_perpendicular_component = linalg.sub( l_distance_vector, distance_desired )
l_fw_heading_vector = linalg.add( l_parallel_component, l_perpendicular_component )
return l_fw_heading_vector, l_parallel_component, l_perpendicular_component, l_distance_vector, l_wall_surface
def transform_to( self, reference_pose ):
rel_vect, rel_theta = self.vunpack() # elements of this pose (the relative pose)
ref_vect, ref_theta = reference_pose.vunpack() # elements of the reference pose
# construct the elements of the transformed pose
result_vect_d = linalg.rotate_vector( rel_vect, ref_theta )
result_vect = linalg.add( ref_vect, result_vect_d )
result_theta = ref_theta + rel_theta
return Pose( result_vect, result_theta )
def _draw_detection_to_frame( self ):
target_delta = self.proximity_sensor.target_delta
if target_delta != None:
detector_endpoints = self.proximity_sensor.detector_line.vertexes
detector_vector = linalg.sub( detector_endpoints[1], detector_endpoints[0] )
target_vector = linalg.add( detector_endpoints[0], linalg.scale( detector_vector, target_delta ) )
self.viewer.current_frame.add_circle( pos = target_vector,
radius = 0.02,
color = "black",
alpha = 0.7 )
def transform_to(self, reference_pose):
rel_vect, rel_theta = self.vunpack(
) # elements of this pose (the relative pose)
ref_vect, ref_theta = reference_pose.vunpack(
) # elements of the reference pose
# construct the elements of the transformed pose
result_vect_d = linalg.rotate_vector(rel_vect, ref_theta)
result_vect = linalg.add(ref_vect, result_vect_d)
result_theta = ref_theta + rel_theta
return Pose(result_vect, result_theta)
def update_heading( self ):
# generate and store the vectors generated by the two controller types
self.gtg_heading_vector = self.go_to_goal_controller.calculate_gtg_heading_vector()
self.ao_heading_vector, self.obstacle_vectors = self.avoid_obstacles_controller.calculate_ao_heading_vector()
# normalize the heading vectors
self.gtg_heading_vector = linalg.unit( self.gtg_heading_vector )
self.ao_heading_vector = linalg.unit( self.ao_heading_vector )
# generate the blended vector
self.blended_heading_vector = linalg.add( linalg.scale( self.gtg_heading_vector, self.alpha ),
linalg.scale( self.ao_heading_vector, 1.0 - self.alpha ) )
def update_heading(self):
# generate and store the vectors generated by the two controller types
self.gtg_heading_vector = self.go_to_goal_controller.calculate_gtg_heading_vector()
self.ao_heading_vector, self.obstacle_vectors = self.avoid_obstacles_controller.calculate_ao_heading_vector()
# normalize the heading vectors
self.gtg_heading_vector = linalg.unit(self.gtg_heading_vector)
self.ao_heading_vector = linalg.unit(self.ao_heading_vector)
# generate the blended vector
self.blended_heading_vector = linalg.add(
linalg.scale(self.gtg_heading_vector, self.alpha), linalg.scale(self.ao_heading_vector, 1.0 - self.alpha)
)
def _draw_detection_to_frame(self, frame):
"""
Visualize the detection of the proximity sensor
:param frame: The frame to be used
"""
target_delta = self.proximity_sensor.target_delta
if target_delta is not None:
detector_endpoints = self.proximity_sensor.detector_line.vertexes
detector_vector = linalg.sub(detector_endpoints[1],
detector_endpoints[0])
target_vector = linalg.add(
detector_endpoints[0],
linalg.scale(detector_vector, target_delta))
frame.add_circle(pos=target_vector,
radius=0.02,
color="black",
alpha=0.7)
def line_segment_intersection(line1, line2):
# see http://stackoverflow.com/questions/563198
nointersect_symbol = (False, None, None)
p1, r1 = line1[0], linalg.sub(line1[1], line1[0])
p2, r2 = line2[0], linalg.sub(line2[1], line2[0])
r1xr2 = float(linalg.cross(r1, r2))
if r1xr2 == 0.0: return nointersect_symbol
p2subp1 = linalg.sub(p2, p1)
d1 = linalg.cross(p2subp1, r2) / r1xr2
d2 = linalg.cross(p2subp1, r1) / r1xr2
if d1 >= 0.0 and d1 <= 1.0 and d2 >= 0.0 and d2 <= 1.0:
return True, linalg.add(p1, linalg.scale(r1, d1)), d1
else:
return nointersect_symbol
def line_segment_intersection( line1, line2 ):
# see http://stackoverflow.com/questions/563198
nointersect_symbol = ( False, None, None )
p1, r1 = line1[0], linalg.sub( line1[1], line1[0] )
p2, r2 = line2[0], linalg.sub( line2[1], line2[0] )
r1xr2 = float( linalg.cross( r1, r2 ) )
if r1xr2 == 0.0: return nointersect_symbol
p2subp1 = linalg.sub( p2, p1 )
d1 = linalg.cross( p2subp1, r2 ) / r1xr2
d2 = linalg.cross( p2subp1, r1 ) / r1xr2
if d1 >= 0.0 and d1 <= 1.0 and d2 >= 0.0 and d2 <= 1.0:
return True, linalg.add( p1, linalg.scale( r1, d1 ) ), d1
else:
return nointersect_symbol
def calculate_ao_heading_vector( self ):
# initialize vector
obstacle_vectors = [ [ 0.0, 0.0 ] ] * len( self.proximity_sensor_placements )
ao_heading_vector = [ 0.0, 0.0 ]
# get the distances indicated by the robot's sensor readings
sensor_distances = self.supervisor.proximity_sensor_distances()
# calculate the position of detected obstacles and find an avoidance vector
robot_pos, robot_theta = self.supervisor.estimated_pose().vunpack()
for i in range( len( sensor_distances ) ):
# calculate the position of the obstacle
sensor_pos, sensor_theta = self.proximity_sensor_placements[i].vunpack()
vector = [ sensor_distances[i], 0.0 ]
vector = linalg.rotate_and_translate_vector( vector, sensor_theta, sensor_pos )
obstacle_vectors[i] = vector # store the obstacle vectors in the robot's reference frame
# accumluate the heading vector within the robot's reference frame
ao_heading_vector = linalg.add( ao_heading_vector,
linalg.scale( vector, self.sensor_gains[i] ) )
return ao_heading_vector, obstacle_vectors
