ENG5009 Advanced Control 5

Overview

This lab exercise introduces the Fuzzy Logic Toolbox within MATLAB, introduces the robot model and simulation to be used in the assignment and allows you to create and test a simple Neural Network.

You will be required to work your way through the examples provided and keep track of your progress using the answer grid provided on Moodle. This answer grid is to be included as an appendix in the final assignment report and is worth 20% of the assignment mark.

Laboratory 1

Introduction

The main purpose of this first lab is to provide you with an introduction to the Fuzzy Logic Toolbox within MATLAB. This is achieved through:

•    The MATLAB Fuzzy Logic Toolbox Tutorial,

•    Implementing the Fuzzy Logic system you developed into the command line,

•    Create a second Fuzzy Logic system.

Following this process will give you the necessary skills to create a Fuzzy Logic system.

1. Introduction to Fuzzy Logic Toolbox

The first stage in this tutorial is to learn more about the fuzzy logic toolbox in MATLAB.

The easiest way of doing this is to follow the Mathworks Tutorial, which can be found here:

https://uk.mathworks.com/help/fuzzy/building-systems-with-fuzzy-logic-toolbox-software.html

This tutorial introduces you to the basics of using the toolbox. The skills you learn developing this system will support you with the main assignment – if you choose to use a Fuzzy logic controller.

Using the standard settings, provide the tipping level for the following input values in Table 1.

Table 1: Service and Food Values

Note these values in the Laboratory answer sheet.

Change the defuzzificiation method to the Mean of Maximum and using the values in Table 1 note the values in the Laboratory answer sheet. To change the defuzzification method return to the front page of the Fuzzy Logic toolbox and change via the drop down menu on the right hand side of the screen.

Change the defuzzificiation method to the Bisector and using the values in Table 1 note the values in the Laboratory answer sheet.

2. Integrating the Fuzzy system into the command line

To use the Fuzzy system developed you need to be able to call it from within a MATLAB script.  Ensure you have reset the tipper FIS back to Centroid defuzzification before saving your values.

To use tipper in the command line, do this:

tipper = readfis('tipper'); % create a variable name and assign your fis file to it

You then evaluate the inputs using the following:

Using the system you have developed, note the values in the Laboratory Answer Sheet based on the inputs in Table 6.

Table 6: Inputs for the AC system

This completes Laboratory 1.

Laboratory 2

For the assignment you will be using a model of a four wheeled robot. This model will be available from the ENG5009 Moodle page under the Lab and Assignment section.

This laboratory provides a demonstration of how to use the robot model and simulation that is to be used and how to integrate a controller within it.

The model is called:  DynamicalModel.m and is fully described in

ENG5009_ModelDescription_2024.

In addition, there is a simple simulation file, RunModel.m, that can be used to run the model.

Further files that maybe of interest are:

drawrobot.m: this file draws a representation of the robot on to a given figure.

WallGeneration.m: supports the development of obstacles.

Sensor.m: A function that provides a forward-looking sensor that returns the nearest objects on the right and left handside of the robot.

1. Basic Operation

The first stage is to learn how to use the model and the simulation file.

Apply the voltages in Table 7 to voltage_left and voltage_right.

Table 7: Values to be applied to left and right Motors

For each of these insert the output plot figure 1 into the Laboratory Answer Sheet.

2. Basic Manoeuvre

You are required to move the robot:

•    a short way forward,

•    turn left,

•    a short way forward,

•    turn right manoeuvre,

•    a short way forward.

The output should be like the path shown in Figure 1 below. When completed insert output Figure 2 into the Laboratory Answer Sheet.

To achieve this, you need to create a simple time-based controller that alters the voltage_left and voltage_right values as and when required. For example, you could use several if statements to alter the variables at relevant time steps. For example, the code below will alter the voltage_left and voltage_right after 1s.

if (time(timeStep) >= 1)

voltage_left = -6;

voltage_right = 6;

end

3.    Running the Model with a Fuzzy Controller

Develop a fuzzy controller that allows the robot to avoid obstacles. This should be based on the example developed during Lectures 5 and 6 of the course. You should use the Matlab Fuzzy Logic Toolbox to do this.

For this you will be required to add in obstacle detection. To do this add the following line:

[sensorOutput] = Sensor(robot_x, robot_y, headingAngle, obstacleMatrix)

You will also be required to draw walls on the plot. To do this, set drawWalls to 1.

Insert figure 2 into the Laboratory Answer Sheet.

4.    Running the Model with a Neural Controller

In lectures 7 and 8 we discussed avery simple Neural network controller that should be able to drive the robot forward while avoiding obstacles.