In this tutorial, we will learn about the Bug2 algorithm for robot motion planning. The Bug2 algorithm is used when you have a mobile robot:

With a known starting location
With a known goal location
Inside an unexplored environment
Contains a distance sensor that can detect the distances to objects and walls in the environment (e.g. like an ultrasonic sensor or a laser distance sensor.)
Contains an encoder that the robot can use to estimate how far the robot has traveled from the starting location.

Here is a video of a simulated robot I developed in Gazebo and ROS2 that uses the Bug2 algorithm to move from a starting point to a goal point, avoiding walls along the way.

Table of Contents

Real-World Applications

Imagine an automatic vacuum cleaner like the one below. It vacuums a room and then needs to move autonomously to another room so that it can clean that room. Along the way, the robot must avoid bumping into walls.

Algorithm Description

On a high level, the Bug2 algorithm has two main modes:

Go to Goal Mode: Move from the current location towards the goal (x,y) coordinate.
Wall Following Mode: Move along a wall.

Here is pseudocode for the algorithm:

1. Calculate a start-goal line. The start-goal line is an imaginary line that connects the starting position to the goal position.

2. While Not at the Goal

Move towards the goal along the start-goal line.
- If a wall is encountered:
  - Remember the location where the wall was first encountered. This is the “hit point.”
  - Follow the wall until you encounter the start-goal line. This point is known as the “leave point.”
    - If the leave point is closer to the goal than the hit point, leave the wall, and move towards the goal again.
    - Otherwise, continue following the wall.

That’s the algorithm. In the image below, I have labeled the hit points and leave points.

1-hit-points-leave-points-bug2-algorithmJPG

Python Implementation (ROS2)

The robot you see in the video at the beginning of this tutorial has a laser distance scanner mounted on top of it that enables it to detect distances from 0 degrees (right side of the robot) to 180 degrees (left side of the robot). It is using three Python functions, bug2, follow_wall, and go_to_goal.

Below is my complete code. It runs in ROS2 Foxy Fitzroy and Gazebo and makes use of the Differential Drive plugin.

Since we’re just focusing on the algorithms in this post, I won’t go into the details of how I created the robot, built the simulated maze, and created publisher and subscriber nodes for the goal locations, robot pose, and laser distance sensor information (I might cover this in a future post).

Don’t be intimidated by the code. It is long but well-commented. Go through each piece and function one step at a time. Focus only on the following three methods:

go_to_goal(self) method
follow_wall(self) method
bug2(self) method

Those three methods encapsulate the implementation of Bug2.

I created descriptive variable names so that you can compare the code to the pseudocode I wrote earlier in this tutorial.

# Author: Addison Sears-Collins
# Date: December 17, 2020
# ROS Version: ROS 2 Foxy Fitzroy
 
############## IMPORT LIBRARIES #################
# Python math library
import math 
 
# ROS client library for Python
import rclpy 
 
# Enables pauses in the execution of code
from time import sleep 
 
# Used to create nodes
from rclpy.node import Node
 
# Enables the use of the string message type
from std_msgs.msg import String 
 
# Twist is linear and angular velocity
from geometry_msgs.msg import Twist     
                     
# Handles LaserScan messages to sense distance to obstacles (i.e. walls)        
from sensor_msgs.msg import LaserScan    
 
# Handle Pose messages
from geometry_msgs.msg import Pose 
 
# Handle float64 arrays
from std_msgs.msg import Float64MultiArray
                     
# Handles quality of service for LaserScan data
from rclpy.qos import qos_profile_sensor_data 
 
# Scientific computing library
import numpy as np 
 
class PlaceholderController(Node):
    """
    Create a Placeholder Controller class, which is a subclass of the Node 
    class for ROS2.
    """
 
    def __init__(self):
        """
        Class constructor to set up the node
        """
        ####### INITIALIZE ROS PUBLISHERS AND SUBSCRIBERS##############
        # Initiate the Node class's constructor and give it a name
        super().__init__('PlaceholderController')
         
        # Create a subscriber
        # This node subscribes to messages of type Float64MultiArray  
        # over a topic named: /en613/state_est
        # The message represents the current estimated state:
        #   [x, y, yaw]
        # The callback function is called as soon as a message 
        # is received.
        # The maximum number of queued messages is 10.
        self.subscription = self.create_subscription(
            Float64MultiArray,
            '/en613/state_est',
            self.state_estimate_callback,
            10)
        self.subscription  # prevent unused variable warning
         
        # Create a subscriber
        # This node subscribes to messages of type 
        # sensor_msgs/LaserScan     
        self.scan_subscriber = self.create_subscription(
            LaserScan,
            '/en613/scan',
            self.scan_callback,
            qos_profile=qos_profile_sensor_data)
             
        # Create a subscriber
        # This node subscribes to messages of type geometry_msgs/Pose 
        # over a topic named: /en613/goal
        # The message represents the the goal position.
        # The callback function is called as soon as a message 
        # is received.
        # The maximum number of queued messages is 10.
        self.subscription_goal_pose = self.create_subscription(
            Pose,
            '/en613/goal',
            self.pose_received,
            10)
             
        # Create a publisher
        # This node publishes the desired linear and angular velocity 
        # of the robot (in the robot chassis coordinate frame) to the 
        # /en613/cmd_vel topic. Using the diff_drive
        # plugin enables the basic_robot model to read this 
        # /end613/cmd_vel topic and execute the motion accordingly.
        self.publisher_ = self.create_publisher(
            Twist, 
            '/en613/cmd_vel', 
            10)
         
        # Initialize the LaserScan sensor readings to some large value
        # Values are in meters.
        self.left_dist = 999999.9 # Left
        self.leftfront_dist = 999999.9 # Left-front
        self.front_dist = 999999.9 # Front
        self.rightfront_dist = 999999.9 # Right-front
        self.right_dist = 999999.9 # Right
 
        ################### ROBOT CONTROL PARAMETERS ##################
         
        # Maximum forward speed of the robot in meters per second
        # Any faster than this and the robot risks falling over.
        self.forward_speed = 0.035 
         
        # Current position and orientation of the robot in the global 
        # reference frame
        self.current_x = 0.0
        self.current_y = 0.0
        self.current_yaw = 0.0
         
        # By changing the value of self.robot_mode, you can alter what
        # the robot will do when the program is launched.
        #   "obstacle avoidance mode": Robot will avoid obstacles
        #   "go to goal mode": Robot will head to an x,y coordinate   
        #   "wall following mode": Robot will follow a wall 
        self.robot_mode = "go to goal mode"
         
        ############# OBSTACLE AVOIDANCE MODE PARAMETERS ##############
         
        # Obstacle detection distance threshold
        self.dist_thresh_obs = 0.25 # in meters
         
        # Maximum left-turning speed    
        self.turning_speed = 0.25 # rad/s
 
        ############# GO TO GOAL MODE PARAMETERS ######################
        # Finite states for the go to goal mode
        #   "adjust heading": Orient towards a goal x, y coordinate
        #   "go straight": Go straight towards goal x, y coordinate
        #   "goal achieved": Reached goal x, y coordinate
        self.go_to_goal_state = "adjust heading"
         
        # List the goal destinations
        # We create a list of the (x,y) coordinate goals
        self.goal_x_coordinates = False # [ 0.0, 3.0, 0.0, -1.5, -1.5,  4.5, 0.0]
        self.goal_y_coordinates = False # [-4.0, 1.0, 1.5,  1.0, -3.0, -4.0, 0.0]
         
        # Keep track of which goal we're headed towards
        self.goal_idx = 0
         
        # Keep track of when we've reached the end of the goal list
        self.goal_max_idx =  None # len(self.goal_x_coordinates) - 1 
         
        # +/- 2.0 degrees of precision
        self.yaw_precision = 2.0 * (math.pi / 180) 
         
        # How quickly we need to turn when we need to make a heading
        # adjustment (rad/s)
        self.turning_speed_yaw_adjustment = 0.0625
         
        # Need to get within +/- 0.2 meter (20 cm) of (x,y) goal
        self.dist_precision = 0.2
 
        ############# WALL FOLLOWING MODE PARAMETERS ##################     
        # Finite states for the wall following mode
        #   "turn left": Robot turns towards the left
        #   "search for wall": Robot tries to locate the wall       
        #   "follow wall": Robot moves parallel to the wall
        self.wall_following_state = "turn left"
         
        # Set turning speeds (to the left) in rad/s 
        # These values were determined by trial and error.
        self.turning_speed_wf_fast = 1.0  # Fast turn
        self.turning_speed_wf_slow = 0.125 # Slow turn
         
        # Wall following distance threshold.
        # We want to try to keep within this distance from the wall.
        self.dist_thresh_wf = 0.45 # in meters  
         
        # We don't want to get too close to the wall though.
        self.dist_too_close_to_wall = 0.15 # in meters
         
        ################### BUG2 PARAMETERS ###########################
         
        # Bug2 Algorithm Switch
        # Can turn "ON" or "OFF" depending on if you want to run Bug2
        # Motion Planning Algorithm
        self.bug2_switch = "ON"
         
        # Start-Goal Line Calculated?
        self.start_goal_line_calculated = False
         
        # Start-Goal Line Parameters
        self.start_goal_line_slope_m = 0
        self.start_goal_line_y_intercept = 0
        self.start_goal_line_xstart = 0
        self.start_goal_line_xgoal = 0
        self.start_goal_line_ystart = 0
        self.start_goal_line_ygoal = 0
         
        # Anything less than this distance means we have encountered
        # a wall. Value determined through trial and error.
        self.dist_thresh_bug2 = 0.15
         
        # Leave point must be within +/- 0.1m of the start-goal line
        # in order to go from wall following mode to go to goal mode
        self.distance_to_start_goal_line_precision = 0.1
         
        # Used to record the (x,y) coordinate where the robot hit
        # a wall.
        self.hit_point_x = 0
        self.hit_point_y = 0
         
        # Distance between the hit point and the goal in meters
        self.distance_to_goal_from_hit_point = 0.0
         
        # Used to record the (x,y) coordinate where the robot left
        # a wall.       
        self.leave_point_x = 0
        self.leave_point_y = 0
         
        # Distance between the leave point and the goal in meters
        self.distance_to_goal_from_leave_point = 0.0
         
        # The hit point and leave point must be far enough 
        # apart to change state from wall following to go to goal
        # This value helps prevent the robot from getting stuck and
        # rotating in endless circles.
        # This distance was determined through trial and error.
        self.leave_point_to_hit_point_diff = 0.25 # in meters
         
    def pose_received(self,msg):
        """
        Populate the pose.
        """
        self.goal_x_coordinates = [msg.position.x]
        self.goal_y_coordinates = [msg.position.y]
        self.goal_max_idx = len(self.goal_x_coordinates) - 1       
     
    def scan_callback(self, msg):
        """
        This method gets called every time a LaserScan message is 
        received on the /en613/scan ROS topic   
        """
        # Read the laser scan data that indicates distances
        # to obstacles (e.g. wall) in meters and extract
        # 5 distinct laser readings to work with.
        # Each reading is separated by 45 degrees.
        # Assumes 181 laser readings, separated by 1 degree. 
        # (e.g. -90 degrees to 90 degrees....0 to 180 degrees)
        self.left_dist = msg.ranges[180]
        self.leftfront_dist = msg.ranges[135]
        self.front_dist = msg.ranges[90]
        self.rightfront_dist = msg.ranges[45]
        self.right_dist = msg.ranges[0]
         
        # The total number of laser rays. Used for testing.
        #number_of_laser_rays = str(len(msg.ranges))        
             
        # Print the distance values (in meters) for testing
        #self.get_logger().info('L:%f LF:%f F:%f RF:%f R:%f' % (
        #   self.left_dist,
        #   self.leftfront_dist,
        #   self.front_dist,
        #   self.rightfront_dist,
        #   self.right_dist))
         
        if self.robot_mode == "obstacle avoidance mode":
            self.avoid_obstacles()
             
    def state_estimate_callback(self, msg):
        """
        Extract the position and orientation data. 
        This callback is called each time
        a new message is received on the '/en613/state_est' topic
        """
        # Update the current estimated state in the global reference frame
        curr_state = msg.data
        self.current_x = curr_state[0]
        self.current_y = curr_state[1]
        self.current_yaw = curr_state[2]
         
        # Wait until we have received some goal destinations.
        if self.goal_x_coordinates == False and self.goal_y_coordinates == False:
            return
                 
        # Print the pose of the robot
        # Used for testing
        #self.get_logger().info('X:%f Y:%f YAW:%f' % (
        #   self.current_x,
        #   self.current_y,
        #   np.rad2deg(self.current_yaw)))  # Goes from -pi to pi 
         
        # See if the Bug2 algorithm is activated. If yes, call bug2()
        if self.bug2_switch == "ON":
            self.bug2()
        else:
             
            if self.robot_mode == "go to goal mode":
                self.go_to_goal()
            elif self.robot_mode == "wall following mode":
                self.follow_wall()
            else:
                pass # Do nothing           
         
    def avoid_obstacles(self):
        """
        Wander around the maze and avoid obstacles.
        """
        # Create a Twist message and initialize all the values 
        # for the linear and angular velocities
        msg = Twist()
        msg.linear.x = 0.0
        msg.linear.y = 0.0
        msg.linear.z = 0.0
        msg.angular.x = 0.0
        msg.angular.y = 0.0
        msg.angular.z = 0.0
         
        # Logic for avoiding obstacles (e.g. walls)
        # >d means no obstacle detected by that laser beam
        # <d means an obstacle was detected by that laser beam
        d = self.dist_thresh_obs
        if   self.leftfront_dist > d and self.front_dist > d and self.rightfront_dist > d:
            msg.linear.x = self.forward_speed # Go straight forward
        elif self.leftfront_dist > d and self.front_dist < d and self.rightfront_dist > d:
            msg.angular.z = self.turning_speed  # Turn left
        elif self.leftfront_dist > d and self.front_dist > d and self.rightfront_dist < d:
            msg.angular.z = self.turning_speed  
        elif self.leftfront_dist < d and self.front_dist > d and self.rightfront_dist > d:
            msg.angular.z = -self.turning_speed # Turn right
        elif self.leftfront_dist > d and self.front_dist < d and self.rightfront_dist < d:
            msg.angular.z = self.turning_speed 
        elif self.leftfront_dist < d and self.front_dist < d and self.rightfront_dist > d:
            msg.angular.z = -self.turning_speed 
        elif self.leftfront_dist < d and self.front_dist < d and self.rightfront_dist < d:
            msg.angular.z = self.turning_speed 
        elif self.leftfront_dist < d and self.front_dist > d and self.rightfront_dist < d:
            msg.linear.x = self.forward_speed
        else:
            pass
             
        # Send the velocity commands to the robot by publishing 
        # to the topic
        self.publisher_.publish(msg)
                             
    def go_to_goal(self):
        """
        This code drives the robot towards to the goal destination
        """
        # Create a geometry_msgs/Twist message
        msg = Twist()
        msg.linear.x = 0.0
        msg.linear.y = 0.0
        msg.linear.z = 0.0
        msg.angular.x = 0.0
        msg.angular.y = 0.0
        msg.angular.z = 0.0
         
        # If Bug2 algorithm is activated
        if self.bug2_switch == "ON":
         
            # If the wall is in the way
            d = self.dist_thresh_bug2
            if (    self.leftfront_dist < d or
                self.front_dist < d or
                self.rightfront_dist < d):
             
                # Change the mode to wall following mode.
                self.robot_mode = "wall following mode"
                 
                # Record the hit point  
                self.hit_point_x = self.current_x
                self.hit_point_y = self.current_y
                 
                # Record the distance to the goal from the 
                # hit point
                self.distance_to_goal_from_hit_point = (
                    math.sqrt((
                    pow(self.goal_x_coordinates[self.goal_idx] - self.hit_point_x, 2)) + (
                    pow(self.goal_y_coordinates[self.goal_idx] - self.hit_point_y, 2))))    
                     
                # Make a hard left to begin following wall
                msg.angular.z = self.turning_speed_wf_fast
                         
                # Send command to the robot
                self.publisher_.publish(msg)
                 
                # Exit this function        
                return
             
        # Fix the heading       
        if (self.go_to_goal_state == "adjust heading"):
             
            # Calculate the desired heading based on the current position 
            # and the desired position
            desired_yaw = math.atan2(
                    self.goal_y_coordinates[self.goal_idx] - self.current_y,
                    self.goal_x_coordinates[self.goal_idx] - self.current_x)
             
            # How far off is the current heading in radians?        
            yaw_error = desired_yaw - self.current_yaw
             
            # Adjust heading if heading is not good enough
            if math.fabs(yaw_error) > self.yaw_precision:
             
                if yaw_error > 0:    
                    # Turn left (counterclockwise)      
                    msg.angular.z = self.turning_speed_yaw_adjustment               
                else:
                    # Turn right (clockwise)
                    msg.angular.z = -self.turning_speed_yaw_adjustment
                 
                # Command the robot to adjust the heading
                self.publisher_.publish(msg)
                 
            # Change the state if the heading is good enough
            else:               
                # Change the state
                self.go_to_goal_state = "go straight"
                 
                # Command the robot to stop turning
                self.publisher_.publish(msg)        
 
        # Go straight                                       
        elif (self.go_to_goal_state == "go straight"):
             
            position_error = math.sqrt(
                        pow(
                        self.goal_x_coordinates[self.goal_idx] - self.current_x, 2)
                        + pow(
                        self.goal_y_coordinates[self.goal_idx] - self.current_y, 2)) 
                         
             
            # If we are still too far away from the goal                        
            if position_error > self.dist_precision:
 
                # Move straight ahead
                msg.linear.x = self.forward_speed
                     
                # Command the robot to move
                self.publisher_.publish(msg)
             
                # Check our heading         
                desired_yaw = math.atan2(
                    self.goal_y_coordinates[self.goal_idx] - self.current_y,
                    self.goal_x_coordinates[self.goal_idx] - self.current_x)
                 
                # How far off is the heading?   
                yaw_error = desired_yaw - self.current_yaw      
         
                # Check the heading and change the state if there is too much heading error
                if math.fabs(yaw_error) > self.yaw_precision:
                     
                    # Change the state
                    self.go_to_goal_state = "adjust heading"
                 
            # We reached our goal. Change the state.
            else:           
                # Change the state
                self.go_to_goal_state = "goal achieved"
                 
                # Command the robot to stop
                self.publisher_.publish(msg)
         
        # Goal achieved         
        elif (self.go_to_goal_state == "goal achieved"):
                 
            self.get_logger().info('Goal achieved! X:%f Y:%f' % (
                self.goal_x_coordinates[self.goal_idx],
                self.goal_y_coordinates[self.goal_idx]))
             
            # Get the next goal
            self.goal_idx = self.goal_idx + 1
         
            # Do we have any more goals left?           
            # If we have no more goals left, just stop
            if (self.goal_idx > self.goal_max_idx):
                self.get_logger().info('Congratulations! All goals have been achieved.')
                while True:
                    pass
 
            # Let's achieve our next goal
            else: 
                # Change the state
                self.go_to_goal_state = "adjust heading"               
 
            # We need to recalculate the start-goal line if Bug2 is running
            self.start_goal_line_calculated = False            
         
        else:
            pass
             
    def follow_wall(self):
        """
        This method causes the robot to follow the boundary of a wall.
        """
        # Create a geometry_msgs/Twist message
        msg = Twist()
        msg.linear.x = 0.0
        msg.linear.y = 0.0
        msg.linear.z = 0.0
        msg.angular.x = 0.0
        msg.angular.y = 0.0
        msg.angular.z = 0.0        
 
        # Special code if Bug2 algorithm is activated
        if self.bug2_switch == "ON":
         
            # Calculate the point on the start-goal 
            # line that is closest to the current position
            x_start_goal_line = self.current_x
            y_start_goal_line = (
                self.start_goal_line_slope_m * (
                x_start_goal_line)) + (
                self.start_goal_line_y_intercept)
                         
            # Calculate the distance between current position 
            # and the start-goal line
            distance_to_start_goal_line = math.sqrt(pow(
                        x_start_goal_line - self.current_x, 2) + pow(
                        y_start_goal_line - self.current_y, 2)) 
                             
            # If we hit the start-goal line again               
            if distance_to_start_goal_line < self.distance_to_start_goal_line_precision:
             
                # Determine if we need to leave the wall and change the mode
                # to 'go to goal'
                # Let this point be the leave point
                self.leave_point_x = self.current_x
                self.leave_point_y = self.current_y
 
                # Record the distance to the goal from the leave point
                self.distance_to_goal_from_leave_point = math.sqrt(
                    pow(self.goal_x_coordinates[self.goal_idx] 
                    - self.leave_point_x, 2)
                    + pow(self.goal_y_coordinates[self.goal_idx]  
                    - self.leave_point_y, 2)) 
             
                # Is the leave point closer to the goal than the hit point?
                # If yes, go to goal. 
                diff = self.distance_to_goal_from_hit_point - self.distance_to_goal_from_leave_point
                if diff > self.leave_point_to_hit_point_diff:
                         
                    # Change the mode. Go to goal.
                    self.robot_mode = "go to goal mode"
 
 
                # Exit this function
                return             
         
        # Logic for following the wall
        # >d means no wall detected by that laser beam
        # <d means an wall was detected by that laser beam
        d = self.dist_thresh_wf
         
        if self.leftfront_dist > d and self.front_dist > d and self.rightfront_dist > d:
            self.wall_following_state = "search for wall"
            msg.linear.x = self.forward_speed
            msg.angular.z = -self.turning_speed_wf_slow # turn right to find wall
             
        elif self.leftfront_dist > d and self.front_dist < d and self.rightfront_dist > d:
            self.wall_following_state = "turn left"
            msg.angular.z = self.turning_speed_wf_fast
             
             
        elif (self.leftfront_dist > d and self.front_dist > d and self.rightfront_dist < d):
            if (self.rightfront_dist < self.dist_too_close_to_wall):
                # Getting too close to the wall
                self.wall_following_state = "turn left"
                msg.linear.x = self.forward_speed
                msg.angular.z = self.turning_speed_wf_fast      
            else:           
                # Go straight ahead
                self.wall_following_state = "follow wall" 
                msg.linear.x = self.forward_speed   
                                     
        elif self.leftfront_dist < d and self.front_dist > d and self.rightfront_dist > d:
            self.wall_following_state = "search for wall"
            msg.linear.x = self.forward_speed
            msg.angular.z = -self.turning_speed_wf_slow # turn right to find wall
             
        elif self.leftfront_dist > d and self.front_dist < d and self.rightfront_dist < d:
            self.wall_following_state = "turn left"
            msg.angular.z = self.turning_speed_wf_fast
             
        elif self.leftfront_dist < d and self.front_dist < d and self.rightfront_dist > d:
            self.wall_following_state = "turn left"
            msg.angular.z = self.turning_speed_wf_fast
             
        elif self.leftfront_dist < d and self.front_dist < d and self.rightfront_dist < d:
            self.wall_following_state = "turn left"
            msg.angular.z = self.turning_speed_wf_fast
             
        elif self.leftfront_dist < d and self.front_dist > d and self.rightfront_dist < d:
            self.wall_following_state = "search for wall"
            msg.linear.x = self.forward_speed
            msg.angular.z = -self.turning_speed_wf_slow # turn right to find wall
             
        else:
            pass
 
        # Send velocity command to the robot
        self.publisher_.publish(msg)    
         
    def bug2(self):
     
        # Each time we start towards a new goal, we need to calculate the start-goal line
        if self.start_goal_line_calculated == False:
         
            # Make sure go to goal mode is set.
            self.robot_mode = "go to goal mode"            
 
            self.start_goal_line_xstart = self.current_x
            self.start_goal_line_xgoal = self.goal_x_coordinates[self.goal_idx]
            self.start_goal_line_ystart = self.current_y
            self.start_goal_line_ygoal = self.goal_y_coordinates[self.goal_idx]
             
            # Calculate the slope of the start-goal line m
            self.start_goal_line_slope_m = (
                (self.start_goal_line_ygoal - self.start_goal_line_ystart) / (
                self.start_goal_line_xgoal - self.start_goal_line_xstart))
             
            # Solve for the intercept b
            self.start_goal_line_y_intercept = self.start_goal_line_ygoal - (
                    self.start_goal_line_slope_m * self.start_goal_line_xgoal) 
                 
            # We have successfully calculated the start-goal line
            self.start_goal_line_calculated = True
             
        if self.robot_mode == "go to goal mode":
            self.go_to_goal()           
        elif self.robot_mode == "wall following mode":
            self.follow_wall()
def main(args=None):
 
    # Initialize rclpy library
    rclpy.init(args=args)
     
    # Create the node
    controller = PlaceholderController()
 
    # Spin the node so the callback function is called
    # Pull messages from any topics this node is subscribed to
    # Publish any pending messages to the topics
    rclpy.spin(controller)
 
    # Destroy the node explicitly
    # (optional - otherwise it will be done automatically
    # when the garbage collector destroys the node object)
    controller.destroy_node()
     
    # Shutdown the ROS client library for Python
    rclpy.shutdown()
 
if __name__ == '__main__':
    main()

Here is the state estimator that goes with the code above. I used an Extended Kalman Filter to filter the odometry measurements from the mobile robot.

# Author: Addison Sears-Collins
# Date: December 8, 2020
# ROS Version: ROS 2 Foxy Fitzroy
 
# Python math library
import math
 
# ROS client library for Python
import rclpy
 
# Used to create nodes
from rclpy.node import Node
 
# Twist is linear and angular velocity
from geometry_msgs.msg import Twist 
 
# Position, orientation, linear velocity, angular velocity
from nav_msgs.msg import Odometry
 
# Handles laser distance scan to detect obstacles
from sensor_msgs.msg import LaserScan
 
# Used for laser scan
from rclpy.qos import qos_profile_sensor_data
 
# Enable use of std_msgs/Float64MultiArray message
from std_msgs.msg import Float64MultiArray 
 
# Scientific computing library for Python
import numpy as np
 
class PlaceholderEstimator(Node):
    """
    Class constructor to set up the node
    """
    def __init__(self):
        ############## INITIALIZE ROS PUBLISHERS AND SUBSCRIBERS ######
     
     
        # Initiate the Node class's constructor and give it a name
        super().__init__('PlaceholderEstimator')
         
        # Create a subscriber 
        # This node subscribes to messages of type 
        # geometry_msgs/Twist.msg
        # The maximum number of queued messages is 10.
        self.velocity_subscriber = self.create_subscription(
            Twist,
            '/en613/cmd_vel',
            self.velocity_callback,
            10)
             
        # Create a subscriber
        # This node subscribes to messages of type
        # nav_msgs/Odometry
        self.odom_subscriber = self.create_subscription(
            Odometry,
            '/en613/odom',
            self.odom_callback,
            10)
         
        # Create a publisher
        # This node publishes the estimated position (x, y, yaw) 
        # The type of message is std_msgs/Float64MultiArray
        self.publisher_state_est = self.create_publisher(
            Float64MultiArray, 
            '/en613/state_est', 
            10)
             
        ############# STATE TRANSITION MODEL PARAMETERS ############### 
         
        # Time step from one time step t-1 to the next time step t
        self.delta_t = 0.002 # seconds
         
        # Keep track of the estimate of the yaw angle
        # for control input vector calculation.
        self.est_yaw = 0.0
         
        # A matrix
        # 3x3 matrix -> number of states x number of states matrix
        # Expresses how the state of the system [x,y,yaw] changes 
        # from t-1 to t when no control command is executed. Typically 
        # a robot on wheels only drives when the wheels are commanded 
        # to turn.
        # For this case, A is the identity matrix.
        # A is sometimes F in the literature.
        self.A_t_minus_1 = np.array([   [1.0,  0,   0],
                        [  0,1.0,   0],
                        [  0,  0, 1.0]])
         
        # The estimated state vector at time t-1 in the global 
        # reference frame
        # [x_t_minus_1, y_t_minus_1, yaw_t_minus_1]
        # [meters, meters, radians]
        self.state_vector_t_minus_1 = np.array([0.0,0.0,0.0])
         
        # The control input vector at time t-1 in the global 
        # reference frame
        # [v,v,yaw_rate]
        # [meters/second, meters/second, radians/second]
        # In the literature, this is commonly u.
        self.control_vector_t_minus_1 = np.array([0.001,0.001,0.001])
         
        # Noise applied to the forward kinematics (calculation
        # of the estimated state at time t from the state transition
        # model of the mobile robot. This is a vector with the
        # number of elements equal to the number of states.
        self.process_noise_v_t_minus_1 = np.array([0.096,0.096,0.032])
 
        ############# MEASUREMENT MODEL PARAMETERS #################### 
         
        # Measurement matrix H_t
        # Used to convert the predicted state estimate at time t=1
        # into predicted sensor measurements at time t=1.
        # In this case, H will be the identity matrix since the 
        # estimated state maps directly to state measurements from the 
        # odometry data [x, y, yaw]
        # H has the same number of rows as sensor measurements
        # and same number of columns as states.
        self.H_t = np.array([   [1.0,  0,   0],
                    [  0,1.0,   0],
                    [  0,  0, 1.0]])
                 
        # Sensor noise. This is a vector with the
        # number of elements as the number of sensor measurements.
        self.sensor_noise_w_t = np.array([0.07,0.07,0.04])
 
        ############# EXTENDED KALMAN FILTER PARAMETERS ###############
         
        # State covariance matrix P_t_minus_1
        # This matrix has the same number of rows (and columns) as the 
        # number of states (i.e. 3x3 matrix). P is sometimes referred
        # to as Sigma in the literature. It represents an estimate of 
        # the accuracy of the state estimate at t=1 made using the
        # state transition matrix. We start off with guessed values.
        self.P_t_minus_1 = np.array([   [0.1,  0,   0],
                        [  0,0.1,   0],
                        [  0,  0, 0.1]])
                         
        # State model noise covariance matrix Q_t
        # When Q is large, the Kalman Filter tracks large changes in 
        # the sensor measurements more closely than for smaller Q.
        # Q is a square matrix that has the same number of rows as 
        # states.
        self.Q_t = np.array([       [1.0,   0,   0],
                        [  0, 1.0,   0],
                        [  0,   0, 1.0]])
                         
        # Sensor measurement noise covariance matrix R_t
        # Has the same number of rows and columns as sensor measurements.
        # If we are sure about the measurements, R will be near zero.
        self.R_t = np.array([       [1.0,   0,    0],
                        [  0, 1.0,    0],
                        [  0,    0, 1.0]])
         
    def velocity_callback(self, msg):
        """
        Listen to the velocity commands (linear forward velocity 
        in the x direction in the robot's reference frame and 
        angular velocity (yaw rate) around the robot's z-axis.
        Convert those velocity commands into a 3-element control
        input vector ... 
        [v,v,yaw_rate]
        [meters/second, meters/second, radians/second]
        """
        # Forward velocity in the robot's reference frame
        v = msg.linear.x
         
        # Angular velocity around the robot's z axis
        yaw_rate = msg.angular.z
         
        # [v,v,yaw_rate]        
        self.control_vector_t_minus_1[0] = v
        self.control_vector_t_minus_1[1] = v
        self.control_vector_t_minus_1[2] = yaw_rate
 
    def odom_callback(self, msg):
        """
        Receive the odometry information containing the position and orientation
        of the robot in the global reference frame. 
        The position is x, y, z.
        The orientation is a x,y,z,w quaternion. 
        """                    
        roll, pitch, yaw = self.euler_from_quaternion(
                    msg.pose.pose.orientation.x,
                    msg.pose.pose.orientation.y,
                    msg.pose.pose.orientation.z,
                    msg.pose.pose.orientation.w)
                 
        obs_state_vector_x_y_yaw = [msg.pose.pose.position.x,msg.pose.pose.position.y,yaw]
         
        # These are the measurements taken by the odometry in Gazebo
        z_t_observation_vector =  np.array([obs_state_vector_x_y_yaw[0],
                        obs_state_vector_x_y_yaw[1],
                        obs_state_vector_x_y_yaw[2]])
         
        # Apply the Extended Kalman Filter
        # This is the updated state estimate after taking the latest
        # sensor (odometry) measurements into account.              
        updated_state_estimate_t = self.ekf(z_t_observation_vector)
         
        # Publish the estimate state
        self.publish_estimated_state(updated_state_estimate_t)
 
    def publish_estimated_state(self, state_vector_x_y_yaw):
        """
        Publish the estimated pose (position and orientation) of the 
        robot to the '/en613/state_est' topic. 
        :param: state_vector_x_y_yaw [x, y, yaw] 
            x is in meters, y is in meters, yaw is in radians
        """
        msg = Float64MultiArray()
        msg.data = state_vector_x_y_yaw
        self.publisher_state_est.publish(msg)
 
    def euler_from_quaternion(self, x, y, z, w):
        """
        Convert a quaternion into euler angles (roll, pitch, yaw)
        roll is rotation around x in radians (counterclockwise)
        pitch is rotation around y in radians (counterclockwise)
        yaw is rotation around z in radians (counterclockwise)
        """
        t0 = +2.0 * (w * x + y * z)
        t1 = +1.0 - 2.0 * (x * x + y * y)
        roll_x = math.atan2(t0, t1)
     
        t2 = +2.0 * (w * y - z * x)
        t2 = +1.0 if t2 > +1.0 else t2
        t2 = -1.0 if t2 < -1.0 else t2
        pitch_y = math.asin(t2)
     
        t3 = +2.0 * (w * z + x * y)
        t4 = +1.0 - 2.0 * (y * y + z * z)
        yaw_z = math.atan2(t3, t4)
     
        return roll_x, pitch_y, yaw_z # in radians
         
    def getB(self,yaw,dt):
        """
        Calculates and returns the B matrix
         
        3x3 matix -> number of states x number of control inputs
        The control inputs are the forward speed and the rotation 
        rate around the z axis from the x-axis in the 
        counterclockwise direction.
        [v,v,yaw_rate]
        Expresses how the state of the system [x,y,yaw] changes 
        from t-1 to t
        due to the control commands (i.e. inputs).
        :param yaw: The yaw (rotation angle around the z axis) in rad 
        :param dt: The change in time from time step t-1 to t in sec
        """
        B = np.array([  [np.cos(yaw) * dt,  0,   0],
                [  0, np.sin(yaw) * dt,   0],
                [  0,  0, dt]])                 
        return B
         
    def ekf(self,z_t_observation_vector):
        """
        Extended Kalman Filter. Fuses noisy sensor measurement to 
        create an optimal estimate of the state of the robotic system.
         
        INPUT
        :param z_t_observation_vector The observation from the Odometry
            3x1 NumPy Array [x,y,yaw] in the global reference frame
            in [meters,meters,radians].
         
        OUTPUT
        :return state_estimate_t optimal state estimate at time t   
            [x,y,yaw]....3x1 list --->
            [meters,meters,radians]
                     
        """
        ######################### Predict #############################
        # Predict the state estimate at time t based on the state 
        # estimate at time t-1 and the control input applied at time
        # t-1.
        state_estimate_t = self.A_t_minus_1 @ (
            self.state_vector_t_minus_1) + (
            self.getB(self.est_yaw,self.delta_t)) @ (
            self.control_vector_t_minus_1) + (
            self.process_noise_v_t_minus_1)
             
        # Predict the state covariance estimate based on the previous
        # covariance and some noise
        P_t = self.A_t_minus_1 @ self.P_t_minus_1 @ self.A_t_minus_1.T + (
            self.Q_t)
         
        ################### Update (Correct) ##########################
        # Calculate the difference between the actual sensor measurements
        # at time t minus what the measurement model predicted 
        # the sensor measurements would be for the current timestep t.
        measurement_residual_y_t = z_t_observation_vector - (
            (self.H_t @ state_estimate_t) + (
            self.sensor_noise_w_t))
             
        # Calculate the measurement residual covariance
        S_t = self.H_t @ P_t @ self.H_t.T + self.R_t
         
        # Calculate the near-optimal Kalman gain
        # We use pseudoinverse since some of the matrices might be
        # non-square or singular.
        K_t = P_t @ self.H_t.T @ np.linalg.pinv(S_t)
         
        # Calculate an updated state estimate for time t
        state_estimate_t = state_estimate_t + (K_t @ measurement_residual_y_t)
         
        # Update the state covariance estimate for time t
        P_t = P_t - (K_t @ self.H_t @ P_t)
         
        #### Update global variables for the next iteration of EKF ####     
        # Update the estimated yaw      
        self.est_yaw = state_estimate_t[2]
         
        # Update the state vector for t-1
        self.state_vector_t_minus_1 = state_estimate_t
         
        # Update the state covariance matrix
        self.P_t_minus_1 = P_t
         
        ######### Return the updated state estimate ###################
        state_estimate_t = state_estimate_t.tolist()
        return state_estimate_t
 
 
def main(args=None):
    """
    Entry point for the progam.
    """
     
    # Initialize rclpy library
    rclpy.init(args=args)
 
    # Create the node
    estimator = PlaceholderEstimator()
 
    # Spin the node so the callback function is called.
    # Pull messages from any topics this node is subscribed to.
    # Publish any pending messages to the topics.
    rclpy.spin(estimator)
 
    # Destroy the node explicitly
    # (optional - otherwise it will be done automatically
    # when the garbage collector destroys the node object)
    estimator.destroy_node()
     
    # Shutdown the ROS client library for Python
    rclpy.shutdown()
 
if __name__ == '__main__':
    main()

Wall Following Robot Mode Demo

Just for fun, I created a switch in the first section of the code (see previous section) that enables you to have the robot do nothing but follow walls. Here is a video demonstration of that.

Obstacle Avoidance Robot Mode Demo

Here is a demo of what the robot looks like when it does nothing but wander around the room, avoiding obstacles and walls.