Visual Follow Line

Goal

The goal of this exercise is to perform a PID reactive control capable of following the line painted on the racing circuit.

Racing circuit. First Person. Model.
Gallery

The students program a Formula1 car in a race circuit to follow the red line in the middle of the road.

Installation

Install the General Infrastructure of the JdeRobot Robotics Academy.

How to run your solution?

Navigate to the follow_line directory

cd exercises/follow_line

Launch Gazebo with the f1_simple_circuit world through the command

roslaunch ./launch/simple_line_follower_ros.launch

Then you have to execute the academic application, which will incorporate your code:

python2 ./follow_line.py follow_line_conf.yml

How to perform the exercise?

To carry out the exercise, you have to edit the file MyAlgorithms.py and insert in it your code, which gives intelligence to the autonomous car.

Where to insert the code?

In the MyAlgorithm.py file,

def execute(self):
    #GETTING THE IMAGES
    image = self.getImage()

    # Add your code here
    print "Runing"

    #EXAMPLE OF HOW TO SEND INFORMATION TO THE ROBOT ACTUATORS
    #self.motors.sendV(10)
    #self.motors.sendW(5)

    #SHOW THE FILTERED IMAGE ON THE GUI
    self.set_threshold_image(image)

Application Programming Interface

  • self.getImage() - to get the image
  • self.motors.sendV() - to set the linear speed
  • self.motors.sendW() - to set the angular velocity
  • self.set_threshold_image() - allows you to view a debug image or with relevant information. It must be an image in RGB format (Tip: np.dstack())

Theory

PID Control is the main fundamental behind this exercise. To understand PID Control, let us first understand what is Control in general.

Control System

A system of devices or set of devices, that manages, commands, directs or regulates the behavior of other devices or systems to achieve the desired results. Simply speaking, a system which controls other systems. Control Systems help a robot to execute a set of commands precisely, in the presence of unforeseen errors.

Types of Control System

Open Loop Control System

A control system in which the control action is completeley independent of the output of the system. A manual control system is on Open Loop System.

Closed Loop Control System

A control system in which the output has an effect on the input quantity in such a manner that the input will adjust itself based on the output generated. An open loop system can be converted to a closed one by providing feedback.

PID Control

A control loop mechanism employing feedback. A PID Controller continuously calculates an error value as the difference between desired output and the current output and applies a correction based on proportional, integral and derivative terms(denoted by P, I, D respectively).

  • Proportional

Proportional Controller gives an output which is proportional to the current error. The error is multiplied with a proportionality constant to get the output. And hence, is 0 if the error is 0.

  • Integral

Integral Controller provides a necessary action to eliminate the offset error which is accumulated by the P Controller.It integrates the error over a period of time until the error value reaches to zero.

  • Derivative

Derivative Controller gives an output depending upon the rate of change or error with respect to time. It gives the kick start for the output thereby increasing system response.

Control Systems Types of Control Systems PID
Control Systems and PID

Tuning Methods

In order for the PID equation to work, we need to determine the constants of the equation. There are 3 constants called the gains of the equation. We have 2 main tuning methods for this.

  • Trial and Error

It is a simple method of PID controller tuning. While system or controller is working, we can tune the controller. In this method, first we have to set Ki and Kd values to zero and increase proportional term (Kp) until system reaches to oscillating behavior. Once it is oscillating, adjust Ki (Integral term) so that oscillations stops and finally adjust D to get fast response.

  • Zeigler Nichols method

Zeigler-Nichols proposed closed loop methods for tuning the PID controller. Those are continuous cycling method and damped oscillation method. Procedures for both methods are same but oscillation behavior is different. In this, first we have to set the p-controller constant, Kp to a particular value while Ki and Kd values are zero. Proportional gain is increased till system oscillates at constant amplitude.

Real Life Example

1 2

Hints

Simple hints provided to help you solve the follow_line exercise.

Detecting the Line to Follow

The first task of the assignment is to detect the line to be followed. This can be achieved easily by filtering the color of the line from the image and applying basic image processing to find the point or line to follow, or in Control terms our Set Point. Refer to these links for more information:

1 2 3

Coding the Controller

The Controller can be designed in various configurations. 3 configurations have been described in detail below:

  • P Controller The simplest way to do the assignment is using the P Controller. Just find the error which is the difference between our Set Point (The point where our car should be heading) and the Current Output (Where the car is actually heading). Keep adjusting the value of the constant, till we get a value where there occurs no unstable oscillations and no slow response.

  • PD Controller This is an interesting way to see the effect of Derivative on the Control. For this, we need to calculate the derivative of the output we are receiving. Since, we are dealing with discrete outputs in our case, we simply calculate the difference between our previous error and the present error, then adjust the proportional constant. Adjust this value along with the P gain to get a good result.

  • PID Controller This is the complete implemented controller. Now, to add the I Controller we need to integrate the output from the point where error was zero, to the present output. While dealing with discrete outputs, we can achieve this using accumulated error. Then, comes the task of adjustment of gain constants till we get our desired result.

Illustrations

examples examples
Unstable Oscillations (left) - Slow Response (right)

Demonstrative Video

Contributors

References

1 2 3 4 5 6 7