This tutorial you will learn how to do tag following using the ardrone. Before you continue this tutorial make sure you have gone through this tutorial.

We already have the tag recognition code working and its all meant to identify image tags. This tutorial helps you to have tag recognition and also make the ardrone follow the tag.

Step 1:Open up a terminal and enter:

roscd nolan3d

Step 2: Go to the bin and open up the file nolan3d.py and replace the code with the following code:

nolan3d.py

#!/usr/bin/env python
import roslib; roslib.load_manifest('nolan')
import rospy
 
from ar_recog.msg import Tag, Tags
from geometry_msgs.msg import Twist
from std_msgs.msg import Empty
from time import time
 
class Delta:
	def __init__(self):
		self.old_value = 0
		self.old_time = 0
		self.old_vel = 0
 
	def get_velocity(self, new_value):
		new_time = time()
 
		# velocity = distance / time
		vel = (new_value - self.old_value) / (new_time - self.old_time)
 
		# smooth velocity
		self.old_vel = self.old_vel + (vel - self.old_vel) * 0.1
 
		# save values for next time
		self.old_time = new_time
		self.old_value = new_value
 
		return self.old_vel
 
distance = Delta()
yRot = Delta()
 
def handleTags(msg):
	global pub
	global lastDir
	global lastSeen
 
	width = msg.image_width
	height = msg.image_height
 
	biggest = Tag()
	for tag in msg.tags:
		if (tag.diameter > biggest.diameter):
			biggest = tag
 
	if biggest.diameter == 0:
		twist = Twist()
		twist.linear.x = 0
		if (time() - lastSeen > .5):
			twist.angular.z = .5*lastDir
		pub.publish(twist)
 
		# reset velocity and distance
		distance.get_velocity(0)
		yRot.get_velocity(0)
		return
 
	lastSeen = time()
 
	cx = 0; cy = 0
	for i in [0,2,4,6]:
		cx = cx + biggest.cwCorners[i]
		cy = cy + biggest.cwCorners[i+1]
	cx = cx / 4. / width
	cy = cy / 4. / height
 
 
	twist = Twist()
 
	# move foward and backward, trying to stop at stopping_dist
	stopping_dist = 2000.
	dist = (biggest.distance - stopping_dist) / stopping_dist
	print biggest.distance
	print dist
	dist_vel = distance.get_velocity(dist)
	if abs(dist) < 0.25:
		# if we are close enough to the stopping distance, just try to stop
		twist.linear.x = dist_vel * 0.25
	else:
		# otherwise try to move within stopping_dist
		twist.linear.x = dist * 0.25
	print twist.linear.x
	twist.linear.x = max(0.03, min(0.05, twist.linear.x))   #change done here for stopping at a optimal distance after seeing the tag by karthik
 
	# try to face perpendicular to the tag
	yRot_velocity = yRot.get_velocity(biggest.yRot)
	if abs(biggest.yRot) < 0.5:
		# if we are mostly facing perpendicular, just try to stay still
		twist.linear.y = yRot_velocity * 0.25
	else:
		# otherwise, rotate towards being in front of the tag
		twist.linear.y = biggest.yRot * 0.25
	twist.linear.y = max(-0.05, min(0.05, twist.linear.y))
 
	# rotate to face the tag
	twist.angular.z = (-(cx - .5)/.5)
 
	if (twist.angular.z < 0):
		lastDir = -1
	else:
		lastDir = 1
	pub.publish(twist)
	print twist
if __name__ == "__main__":
	global pub
	global lastDir 
	global lastSeen
 
	lastSeen = 0
	lastDir = -1
 
	rospy.init_node('nolan', anonymous=True)
	pub = rospy.Publisher('cmd_vel', Twist)
	rospy.Subscriber("tags", Tags, handleTags)	
	takeoff_pub =rospy.Publisher('/ardrone/takeoff', Empty)
	rospy.spin()

Save and close the file.

Make sure the ardrone is on and connected with your computer. Enter the following lines each in a new terminal:

roscore
rosrun ardrone_brown ardrone_driver

Enter the commands in the same terminal:

roscd ar_recog
cd bin
rosparam set aov .67
rosrun ar_recog ar_recog image:=/ardrone/image_raw

Enter the commands ,each in a new terminal :

rosrun drone_teleop drone_teleop.py

Hit the button “t” for take off and close this window

Open up a new terminal and enter:

rosrun nolan3d nolan3d.py

Now the drone would be in air. Make sure you hold the image tag at a optimal distance(not too close). Please have a look at this video.