Uploaded by Sadaf Taj

Artificial Intelligence System Design - MATLAB Project BRIEF

advertisement
DEPARTMENT OF ELECTRONIC AND ELECTRICAL ENGINEERING
Pro-forma to accompany assignment / coursework 2022/2023
This pro-forma should be the first page to any set assignment / coursework. A full assignment brief should
accompany this pro-forma.
Module Code: EE5627/5657
Module Title: Artificial Intelligence Systems Techniques
Module Leader: Maysam Abbod
Assessor: Maysam Abbod
Assessment Title: Part 1: Artificial Intelligence System Design
Weighting: 20%
Main Objectives of the Assessment:
To familiarise yourself with the operation of a mobile robot by implementing a basic model of the
robot (dynamics and kinematics) using MATLAB/Simulink. Then design intelligent controllers to
manipulate the robot movements in following a defined trajectory and test it performance.
Brief Description of the Assessment:
After completing this lab, you should be able to:
Part 1: AI system Design
• Use MATLAB/Simulink to simulate and test a mobile robot dynamic and kinematics models. The mobile robot
simulation and trajectory blocks are provided.
• Simulate the kinematics model, the design a Fuzzy Logic Controller to replace the inverse kinematics.
• Using the inverse kinematics model, generate training data for designing ANN, Neuro Fuzzy systems to be
used as a controller. Test the controller performance in controlling the robot for different trajectories.
Part 2: Deep Learning
Learning Outcomes for the Assessment:
(A) Knowledge and Understanding
Design and evaluate a neural network classifier.
(B) Cognitive (thinking) Skills
Create a neural network system, train and test it
(C) Other Skills and Attributes
(Practical/Professional/Transferable)
MATLAB computing
Assessment and marking criteria
The work will be assessed based on the following:
Design an AI system (10%)
ANFIS controller 25%
ANN controller 25%
Comparison 25%
Report 15%
Assessment method by which a student can demonstrate learning outcomes:
Design, simulation and evaluation.
Format for the assessment/coursework (Guidelines on the expected format and length of submission):
Written report (10 pages maximum)
Distribution date to
students:
26 October 2023
Submission Deadline:
21 December 2023
Indicative Reading List:
MATLAB documentation, Python programming, and lecture notes
Further information:
Department of Electronic and Electrical Engineering
College of Engineering, Design and Physical Sciences
Brunel University London
Assessed Coursework 2022/2023
Courses:
MEng, MSc
Level:
7
Module Code:
EE5627/EE5657
Module Name:
Artificial Intelligence Systems Techniques
Title of Assessment:
Artificial Intelligence Systems Design / Deep Learning
Set By:
Prof M Abbod and Dr Teodorescu
Contribution to Module Mark: 40%
Part 1: AI System Design (20%)
AIM
The aim of this assignment is to familiarise the student with the operation of a mobile robot by implementing a basic
model of the robot (dynamics and kinematics) using MATLAB/Simulink. Then design intelligent controllers to
manipulate the robot movements in following a defined trajectory and test it performance.
OBJECTIVES
The principal objective is to gain knowledge of the different intelligent system when working on an unrehearsed
problem. The advantages and disadvantages of various methodologies will be evaluated when applied to a simulated
control system. Experience will also be gained in designing a Fuzzy Logic and ANN systems to control the trajectory of
a mobile robot.
REPORT STRUCTURE
The report should not exceed more than 10 pages
WORK DESCRIPTION
1. Introduction
A differential drive mobile robot (DDMR) is a mobile robot equipped with a front castor wheel and a pair of co‐axial
drive wheels at the rear. Each of these wheels is driven by a DC motor.
The DDMR movement is based on the control of these two separately driven wheels. It can change its direction by
varying the relative rate of rotation of its wheels. If both wheels are driven in the same direction and speed, the robot
will go in a straight line. If both wheels are turned with equal speed in opposite directions, as is clear from the diagram
shown in Figure 2, the robot will rotate about the central point of the axis. Since the DDMR cannot drive in the direction
transversal to the wheels’ axis, therefore it is considered as nonholonomic mobile robots.
Figure 1: Differential drive mobile robot (DDMR).
Figure 2: Differential drive.
The modelling of the DDMR consists of deriving the kinematic and dynamic models of the robot in addition to the
models of the actuators (DC motors). The kinematic model describes the geometric relationships that govern the
motion of the system and does not consider forces and torques applied to the DDMR. The dynamic model, on the
other hand, is the study of the motion in which forces, and energies are modelled and included. Actuator modelling is
needed to find the relationship between the control signal and the mechanical system’s input.
The aim of this lab is to develop these mathematical models of the DDMR in MATLAB/Simulink and to design a
trajectory tracking controller for the robot. This lab includes three parts which can be carried out independently.
2. MATLAB/Simulink Implementation of Differential Drive Robot
The simplified dynamic model of this DDMR is (refer to reference [2] for more details):
where u is the linear velocity, w is the angular velocity, x, and y are the coordinates, and theta is the angle of the robot
direction.
Let
Then the inverse of A is:
Then the inverse dynamics is given by:
Figure 3 shows the dynamic and kinematic model of the robot (Simulink models are provided):
Figure 3: Combined kinematic and dynamic models of the DDMR.
Simulate and analyse the trajectory of DDMR using the overall model Thet
(kinematic and dynamic) under different inputs
a
to illustrate straight line drive, left‐turn, right‐turn and rotation conditions as illustrated in Figure 2.
3. Controller Design for Trajectory Tracking
The simplified dynamic model of this DDMR is (refer to reference [2] for more details):
and the inverse kinematic controller is given by:
where a = 0.05, Ix = Iy = 0.1, and kx = ky = 0.1
The developed model based on the kinematic and dynamic models is shown in Figure 4.
Figure 4 Inverse kinematic controller of the DDMR.
Simulate the response of the kinematics and dynamics models with the control law on the model of Figure 4 for two
different trajectory types (1: circle and 2: number Eight shape).
4. Neural Network Based Control
a) Design an inverse-kinematics neural network model for the DDMR using data collected from the Simulink model
of the controlled DDMR. Train ANN controller for the following: circle, Eight shape and combined.
b) Using the same data obtained in part b), design a neuro-fuzzy controller for the same purpose.
c) Implement the proposed controllers in parts a) and b) and compare the results with those obtained using the
inverse kinematic controller.
Note: The controller inputs, x_tilda and y_tilda, are for corrections, you can ignore these on the assumption that the
robot is moving on a smooth surface.
References
[1] R. Dhaouadi and A. Abu Hatab, Dynamic Modelling of Differential‐Drive Mobile Robots using Lagrange and
Newton‐Euler Methodologies: A Unified Framework, Advances in Robotics and Automation, Vol. 2, Issue 2, 2013.
[2] Felipe N. Martins, Wanderley C. Celeste, Ricardo Carelli, Mario Sarcinelli‐Filho, Teodiano F. Bastos‐Filho, An
adaptive dynamic controller for autonomous mobile robot trajectory tracking, Control Engineering Practice
(Elsevier), 16 (2008) 1354– 1363.
[3] P. Suster and A. Jadlovska, Neural tracking trajectory of the mobile robot Khepera II in internal model control
structure, 9th International Conference Process Control, June 7 – 10, 2010, Kouty nad Desnou, Czech Republic.
DEPARTMENT OF ELECTRONIC & ELECTRICAL ENGINEERING
STUDENT FEEDBACK FORM
This pro-forma should be used to provide detailed feedback to students.
Module Code: EE5627/EE5657
Module Leader: Prof M Abbod
Module Title: Artificial Intelligence System Techniques
Assessment Title: Artificial Intelligence System Design
Marker: Prof M Abbod
Student Name:
SPO ID:
Main objectives of the assessment:
Assessment Criteria
Markers Feedback
Grade
Demonstrate selfdirection and
originality in designing
an intelligent controller
to replace the standard
control system. (LOs 1,
3) (10%)
Design and Implements
ANN for controlling the
mobile Robot using
MATLAB toolbox. (LOs
2, 3) (25%)
Design and Implements
NFS for controlling the
mobile Robot using
MATLAB toolbox. (LOs
2, 3) (25%)
Implement and
critically evaluate a AI
system (ANN and NF)
models developed in
MATLAB. (LOs 2, 3)
(25%)
Produce a high-quality
technical report (15%)
Please note: The University penalty system will be applied to any work submitted late.
This interim grade does not reflect any late penalties that may be applied and is subject
to approval by the Panel of Examiners.
INTERIM
GRADE
Please note: If you wish to query any aspect of the marking or require further feedback please contact the assessment marker in
the first instance. You may also contact the module leader if you are unable to contact the assessment marker.
Download