49329 Group Projects for 2023

Projects form 30% of the assessment for the subject. There are 14 potential project problems listed
in this document.

You need to form a group of 4 or 3 or 2 and submit your three preferences (choose 3 problems
from the list) as a group. You will be assigned a project based on your skills and your preference.

The group can be formed in Canvas—People—Student Group, with names like 4A, 4B (for 4
people) and 3A, 3B (for 3 people) etc. We will then create a group with the same name in Canvas—
People—Project Group (and delete the original group you created). If you prefer to work
individually, please also create a group with name 1A, 1B etc.

For submitting your preference, please do that within the Canvas group as an announcement. Then
we will reply your message and assign one problem to you.

You are encouraged to find another project at similar level (state space control for linear or
nonlinear systems) that is suitable for you (If this is the case, write a description of the project and
send it to [email protected]).

After the problem assignment, each group of students will negotiate with the subject coordinator to
finalise the scope and the deliverables, and allocate tasks to the group members.

Schedule:

12 September 2023 --- Project list distributed with short descriptions.

• Students form groups, and submit preferences in Canvas

19 September 2023 – project allocation done

11am on 3 October 2023 --- proposal due

• Proposal: Each group submit a document on the scope, deliverables, marking criteria,
proposed approaches, related references of the project (3 to 5 pages). Email the proposal to
[email protected] and cc to all the other group members.

11am, 24 October 2023 – Project presentation, Project report due

Presentation:

• Evaluated by the lecturer and other students

o The difficulty of the project
o Amount of work involved
o Quality of the work
o Presentation skills

Report:

The report will contain the details of the problem statement, the project goals and assumptions, the
modelling, the control method and algorithms, the results, the code used, and the references.

The mark of each student will be based on the performance of the group as well as their individual
contributions, according to the marking criteria.
Brief description of the projects
Short descriptions of the projects offered are given below.

Project 1: Mobile robot formation control

When a team of robots want to keep the relative position among them (e.g. a triangle) during their
motion, each robot needs to control its motion according to its relative position with other robots.

This project aims to achieve certain formation through feedback control. The final outcome will be
a demonstration in MATLAB simulation to achieve the desired formation.

The group working on this project has the flexibility of defining the scenarios, the robot motion
model (the model should contain some uncertainties), and the project goals.

Below are some youtube videos about robot formation:

https://www.youtube.com/watch?v=mR3cwn-r67A

https://www.youtube.com/watch?v=h4ypcDwTZpI
Project 2: One vehicle following another vehicle

The aim of this project is to develop an algorithm such that a vehicle (e.g. a robot or a wheelchair)
can automatically follow another vehicle or person. The vehicle behind is equipped with sensors
(such as camera, laser, kinect) such that it can monitor the motion of the vehicle in front of it and
obtain the way points. Then the vehicle behind needs to follow the way points of the previous
vehicle. The vehicle model could be assumed as

The measurement noises of the sensors used for estimating the way points needs to be taken into
account. The algorithm will be developed in MATLAB with necessary visualisations.
Project 3: Vehicle following a desired complicated path

The aim of this project is to develop a control algorithm such that a vehicle can automatically
follow a desired path in an indoor environment. The desired trajectory is designed based on the
environment structure. The vehicle is equipped with sensors (such as laser scanner or a camera)
such that it can localise itself within the environment. Then the vehicle needs to follow the desired
trajectory.

The uncertainty of the estimate on the robot location needs to be taken into account. The algorithm
will be developed in MATLAB with necessary visualisations.

The vehicle model could be assumed as

Project 4: Control of inverted pendulum systems

Design the feedback controller for inverted pendulum systems (use the nonlinear model as in the
notes provided) taking into account the nonlinearity and uncertainty of the model. Then consider the
control of one more complicated pendulum systems such as Furuta pendulum
http://en.wikipedia.org/wiki/Furuta_pendulum

Understand the models and design the controllers. Use MATLAB to simulate and visualise the
designed control systems.
Project 5: Alignment of underground mining vehicle

The mining vehicle needs to move along a straight line when cutting the ore underground. Suppose
the vehicle is equipped with a sensor that can detect the distances and relative angles to two
landmarks with known positions (one in front and one at the back, both on the straight line).
Develop an algorithm to estimate the vehicle position and orientation with respect to the straight
line and then derive the control algorithm such that the vehicle can follow the straight line. The
uncertainty of the sensor readings needs to be taken into account in the estimation and control
algorithm. This will be developed using simulation in MATLAB.
Project 6: Control of underwater robot

This project requires to model and control an underwater robot. The underwater robot is required to
follow a desired trajectory. It is necessary to take into account the possible disturbances such as
water flow. Simulation in MATLAB is needed. Any kind of underwater robot can be considered.

.
Project 7: Trajectory following with large initial errors

Suppose the velocity of the robot is fixed as a constant and the only control is the turning rate.
Suppose the position of the robot is known all the time. For any desired trajectory and any initial
robot position, design the control law of the robot such that the robot will follow the trajectory as
fast/accurate as possible. This problem will be formulated as a nonlinear optimal control problem.
The algorithm will be developed using simulations in MATLAB. Model Predictive Control could
be used as the control technique.

The vehicle model could be assumed as

where v1 is a constant.
Project 8: Control of unmanned aerial vehicle

This project requires to model and control the unmanned aerial vehicle (UAV). The UAV is
required to follow desired trajectories given from the users. It is necessary to take into account the
possible disturbances such as wind etc. Simulation in MATLAB is needed. Any type of UAV can
be used.

One type of UAV model could be assumed as

Project 9: Using Model Predictive Control for Mobile Robot Complex Trajectory following

Apply Model Predictive Control (MPC) for mobile robot trajectory following. First assume the
position of the robot is known all the time. For any desired trajectory and any initial robot position,
design the control law of the robot such that the robot will follow the trajectory. Then assume the
robot position/orientation estimation is noisy and you can only use the noisy estimate to construct
the controller (add a small noise on the true robot position and orientation), then test the algorithm
and evaluate the performance. The algorithm will be developed using simulations in MATLAB.

The vehicle model (without noise) could be assumed as

Project 10: Autonomous parking

Develop autonomous parking control algorithms for a car in a carpark. The car position and
orientation and the position of the parking lot are given by the user and there are possible obstacles
(e.g. other cars) around. A control algorithm (e.g. Model Predictive Control) for controlling the car
to the required parking lot (to a certain level of accuracy, and avoid all the obstacles) needs to be
designed. The algorithm will be developed using simulations in MATLAB with visualisations.

The vehicle model (without noise) could be assumed as

Project 11: Visual servo control using a mobile robot equipped with a camera

Apply visual servo control for a mobile robot to reach a desired location. A camera is attached on
the robot and the relative position between the robot and the camera is given. Assuming there are
four landmarks in the environment with known locations that form a rectangle. The objective of the
control is to move the robot such that the four landmarks in the image of the camera form a desired
rectangle. The algorithm will be developed using simulations in MATLAB.

The vehicle model (without noise) could be assumed as either a ground vehicle or a UAV.

For more details about visual servo control, please refer to:

https://www.youtube.com/watch?v=GbXrjSEhOWM

F. Chaumette and S. Hutchinson, Visual Servo Control, Part I: Basic Approaches, IEEE
Robotics and Automation Magazine, Vol. 13, No. 4, Dec., 2006, pp. 82-90.

http://www-cvr.ai.uiuc.edu//res.php?u=publications#Entry004
=======================================================================
** For the groups which work on the Fetch Robot and/or UR Robot related projects, at least one
group member needs to be familiar with Robotic Operating Systems (ROS) and C++**
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

Group Project 12: Fetch robot following a guider

Using the learnt and state-of-art control algorithm to control the Fetch robot to follow the path of
the guider in front. Both depth images and RGB images can be used to track the guider. Special
designed artificial markers or patterns can be put on the back of the guider.

A map of the environment needs to be built and a localisation algorithm is also required. Open
source code are available for these two parts but the group needs to be able to make them working
on in the simulator (and the real robot).

Target: The Fetch robot follows the guider in front (maintaining a certain distance) in the office and
corridor environment. The project will be first done using a Fetch simulator; once the simulator
works well, the real robot can be used if it is allowed to access the lab.

Necessary skills: ROS, C++ programming, control

Preferable: basic computer vision knowledge

Group Project 13: Fetch robot following the waypoints on a designed path and perform pick
and place tasks

Using the learnt and state-of-art control algorithm to control the Fetch robot to follow the way
points of a predefined path and reach a table to perform pick and place tasks. Some markers could
be used as the way points. Both depth images and RGB images can be used for detecting the
waypoints and the objects to be picked and placed.

A map of the environment needs to be built and a localisation algorithm is also required. Open
source code are available for these two parts but the group needs to be able to make them working
on the simulator (and the real robot).

Target: The Fetch robot follows the waypoints and finally complete the designed path in the office
and corridor environment. The project will be first done using a Fetch simulator; once the simulator
works well, the real robot can be used if it is allowed to access the lab.

Necessary skills: ROS, C++ programming, control

Preferable: basic computer vision knowledge

=======================================================================
** For the groups which work on the Fetch Robot and/or UR Robot related projects, at least one
group member needs to be familiar with Robotic Operating Systems (ROS) and C++**
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

Group Project 14: Control of UR robot to follow complicated trajectory and perform pick
and place tasks

Using the learnt and state-of-art control algorithm to control a UR robot end effector to follow the
complicated trajectory defined by the user. Then use the robot to perform pick and place tasks. Both
depth images and RGB images from a camera on the robot can be used to provide feedback for the
accurate control.

This project will be done on a UR robot simulator first. Some tests will be conducted on a real robot
if the algorithm performs well in simulator.

Necessary skills: ROS, C++ programming, control

Preferable: basic computer vision knowledge