You are on page 1of 21

Executive Summary

Environmental disasters, whether man made or natural, have


devastating effects on social, economic and environmental
facets of the community. In the event of a disaster, it is
essential to implement efficient and effective disaster
management procedures to restore normality and quality of
life within a community. In the event of a water related
disaster, whether it is a flood, oil spill or shipping disaster, a
quick clean up is even more critical to reduce the risk of
spreading disease through water and disturbing the delicate
ecosystems. Recognising this need, a consortium including the
Queensland Government Department of Maritime Safety, the
Australian Parks and Wildlife Service and the Defence
Materials Organisation have issued a call for tenders for the
design and development of an automated watercraft to enter
areas hazardous to humans in a disaster (UQ 2011). The
prototype of this craft will be designed to collect ping pong
balls from a tub with a 2m diameter and will have sufficient
power to last 10 minutes. The boat features a catamaran hull
design with two 6V motors powering propellers through
gearboxes. The craft implements a spiral search pattern
algorithm using infrared distance and tactile bump sensors.
1.0 Introduction
1.1 Background
Waterways are vital to our way of life. Communities depend on waterways
for public transport, commercial shipping, industrial manufacturing,
agriculture, fishing and recreation. As a result, there are many social,
economic and environmental factors that must be considered when
making responsible decisions that affect our waterways. When disaster
strikes waterways can become polluted with debris, oil or even radioactive
contaminants. In recent months there have been several events which
have highlighted the vulnerability of waterways. The 2011 Queensland
and Brisbane floods and the earthquake and tsunami in Japan are some of
the more recent examples. Environmental disasters can cause significant
damage. The table below outlines the social, environmental and economic
effects of two recent waterway disasters.

Table 1. Social, Environmental, Economic Effects of Queensland Floods


& Oil Spill
Queensland Floods Gulf of Mexico Oil Spill
Social More than 75% of state Health problems
affected. This includes associated with disease
flooded homes, workplaces to people along the Gulf
and isolation of towns. of Mexico (Marcus E Dr
(Salvation Army 2011) 2011)
Environmental Ecosystems disrupted, Five species of sea turtles
riverbank erosion, soils and in Gulf, ecosystems
waterways contaminated destroyed, underwater
(Wildlife Queensland 2011) oxygen supply limited
(Judd A 2010)
Economic Extent of economic cost Has so far cost BP US$41
predicted at US$10 billion as billion (Trotman A 2011)
of early February
(McNamara M 2011)

To assist in future disaster management, a consortium including the


Queensland Government Department of Maritime Safety, the Australian
National Parks and Wildlife Service and the Defence Materials
Organisation have issued a call for tenders for an autonomous watercraft
to assist in the recovery of debris and clean up in areas hazardous to
humans (UQ 2011).

1.2 Project Scope


The aim of this report is to outline the design of an automated watercraft
to assist in the clean up of marine disasters. The table below outlines the
aspects that are considered in scope of the project and those that are
considered negligible.

Table 2. Project scope


In Scope Out of Scope
Analysis of: Analysis of:
• Drag due to water • Aerodynamic drag due to
• Buoyancy air resistance
• Stability • Effects of vessel's bow
wave
• Effects of uneven water
surface
Design of: Avoidance of:
• Propulsion mechanism • Underwater hazards
• Electrical circuit • Overhead hazards (eg.
connecting all sensors, Bridges)
actuators, microprocessors
and power supplies
• Hull considering strength, Launching of vessel into the
size, weight, water
waterproofing, stability,
rigidity and Rusting of the hull
manoeuvrability
• Debris collecting
mechanism Disposal of debris and land
• Software to control craft, collection
motors and actuators to
collect debris Varying weather conditions
• Software to gather and
collect data from sensors
• Logic to decide where to
move the given sensor
data

1.2.1 Design Objectives


The requirements of the craft are as follows;

• The prototype must fit in a box with the dimensions; 30cmx30cmx30cm


however, it is desirable for it to fit within a box of 15cmx20cmx20cm.
• The watercraft must be fully autonomous and must operate without
touching the bottom of the tank when fully loaded.

• The craft should pick up a maximum amount of debris.

• It is desirable for the craft to have a draft of less than 5cm when fully
loaded and have a freeboard of at least 1cm below the bottom of the
plastic container.

• The watercraft must also have a power supply that enables it to last 10
minutes in total.

• The craft must work reliably with minimal maintenance.

• The craft must comply with a $50 budget but must also be relatively cheap
at in the full scale design.

1.2.2 Assumptions
During the process of design, a number of assumptions will be made. These are
as follows;

• There will be no current in the water as the prototype will be tested in an


enclosed tank

• There will be no air resistance or high winds

• No sensors will be designed or assembled; they will be purchased as


standalone parts.

• Software will be written for an Arduino UNO microcontroller board.

• No microprocessors will be designed or fabricated.

• Motors and gearboxes for the propulsion system will be supplied.

• The density of water will be taken at 1000kg/m^3.

• The force acting upon the craft due to surface tension will be taken as 0.

• As the craft will be travelling at low speeds, the drag acting upon the craft
will be assumed to be negligible.

• The craft will be deployed in the water.

1.2.3 Limitations and Constraints


• The watercraft must run on a portable power supply, that is, it cannot be
attached to a power pack or usb

• Materials must be light enough for the watercraft to float


2.2 Mechanical Systems
There are a wide variety of hull designs that could be used for the construction of
an automated watercraft. A flat bottomed hull has straight or rounded edges. Its
features are outlined below:
• Large surface area results in shallow
draught
• Simple to construct

• Excellent stability at low speeds

• Low design complexity

• Low manoeuvrability
Figure 1. Flat Bottomed Hull

A pontoon has two separate hulls that are connected via the deck. Its features
are outlined below:
• Provides exceptional grip in tight turns

• Much more stable than other hull designs

• Deck remains relatively dry due to distance raised above the water

• Easy to manoeuvre with an outboard


motor

• Easier to construct than v or round


bottom hulls

• Can be used effectively with a flat deck

• Space between hulls allows for storage and collection

• Figureconstruction
Simplicity of design and construction results in lower 2. Pontoon Hull
costs

• Lower surface area in contact with water than for other hulls results in
greater draught

• Less compact as it is wider than boats with different hulls

The pontoon hull is more suitable for the automated watercraft due to the
available storage space and greater manoeuvrability than a flat bottomed hull.
2.3 Propulsion Systems
Table 4. Propulsion options
Propulsion Description Pros Cons Suitability
System

Propeller Propellers are +Can be -Higher RPM The fact that above water
(Above fans that placed above required / propellers can be placed, above
water convert axial water (No Larger fan the water, and thus have little to
rotation into interference size for equal no effect on the net or draft,
thrust. with draft or thrust. makes them an attractive option.
net)
However, larger diameter
required makes this design
unsuitable due to the
dimensional restraints of the
craft.

Propeller +Water is a -Have to sit Although underwater propellers


(Under more dense in water and interfere may possibly interfere
water) fluid, thus may increase with the draft of the craft as well
smaller draft of craft. as the net used, they are highly
propellers suitable for the design as they
can be used. produce an equal amount of
-Propellers thrust for a much smaller
may interfere propeller diameter. Through
with fan. proper integration into the
design, the problems associated
with underwater propellers could
also be reduced or eliminated.

Paddlewheel Waterwheel +Motors can -Much less Despite the advantages relating
incorporating be placed on efficient than to the placement and
scoops around the top of propeller. attachment of paddlewheels, the
the wheel to deck. (Force has unavailability of such
push through upward and paddlewheels in combination
water downwards with the inefficiency of the
+Gearbox component) paddlewheel design makes it an
can be unattractive option.
directly
connected to -Very draft
paddlewheel sensitive This, in combination with the fact
that paddlewheels are very
sensitive to shallow drafts,
-Not as makes this design highly
readily unsuitable. One of the vital
available as design specifications of the craft
propellers is a small draft. Thus, a trade-off
occurring between speed and
draft is undesirable.
2.4 Electrical Systems
There are two main types of battery that can be used in powering the
watercraft. Primary batteries can produce current immediately whilst
secondary batteries must be charged before use (Thomas and David
2010). Primary batteries cost less per battery and have a higher capacity
and initial voltage that secondary batteries (Castelvecchi 2009) however,
secondary batteries can be reused and are therefore more
environmentally friendly and may work out more cost efficient if the
batteries are needed for an extended period of time (Armand and
Tarascon 2008). Primary batteries will be used in the watercraft design
because of the higher initial voltage and capacity. Size, weight and power
capacity of each battery will need to be considered in the design. The
table below compares voltage, weight, size and service characteristics of
different types of Energizer batteries (one brand was used for
standardisation and simplicity).

Table 5. Battery Properties


Battery Model Voltage Average Size (length x Service Hours
(V) Weight diameter) (mm) (mAh) at 70mA
(g)

AA E91 1.5 23.0 50.50 x 14.50 25

C E93 1.5 66.2 50.0 x 26.20 80

D E95 1.5 148.0 61.50 x 34.20 150

528 6.0 885.0 112.00 x 68.20 170


(width)

522 9.0 45.6 48.50 x 26.50 7


(width)

A23 12.0 8.0 28.50 x 10.30 <1

2.5 Software Program


The most widespread automated cleaning device is the iRobot
Corporation’s Roomba vacuum. The most basic Roomba model sells for
AU$599 in Australia (iRobot, 2011). The device has two drive wheels with
basic suspension. An optical sensor detects if the wheels are turning at
the correct speed so more power can be given to a wheel if it becomes
stuck. The robot turns by driving one wheel faster than the other. The
device uses four sets of infrared detectors and emitters to sense when the
device is on the edge of a “cliff” such as a staircase. During normal
operation the IR emitters send out light and the IR detectors receive
reflected light back yet when the emitter / detector pair is over a void the
detector no longer senses reflected light and the wheels are stopped
(Cravotta, 2006). The robot uses a spatial reasoning algorithm called
Monte Carlo Localization (MCL). This algorithm allows the Roomba to
determine its location in the room given sensor data and known
dimensions of the room. In this method a number of hypothetical possible
locations are scattered randomly across the room. The probability of the
robot being at each location given the sensor data it is receiving is
calculated using Bayes’ Theorem (Thrun et al, 2001). Each time the sensor
data is updated the probabilities are recalculated. When the probability of
a hypothetical location is very low that location is replaced with a new
randomly generated one.

The LawnBott from Paradise robotics is an autonomous consumer robot


that can mow a lawn. In Collin’s (2007) review of the device he noted that
the device requires a boundary wire to be placed around the yard’s
perimeter. However, when this wire is in place the robot can mow the
lawn then autonomously dock and recharge until its next use. The
LawnBott starts at the center of the lawn then moves outward in a circle
until the whole lawn bas been mown. This circular pattern is an effective
way to ensure that an entire surface has been covered and is therefore a
possible search pattern algorithm for the autonomous watercraft.
4.0 Design Justification
The following criteria where established as a benchmark upon which to
judge possible design options:

• Performance: How much debris does the boat collect in a given time
period.

• Cost: The price of the materials and manufacturing at a large scale.

• Size: It is desirable for the boat to fit in a box measuring 150 x 200 x
200mm

• Battery Life: The watercraft must feature a power supply that


enables it to last a total of 10 minutes.

• Buoyancy: The craft must float with a draft of less than 5cm.

• Sustainability: The craft must be able to scale to a full size version in


a manner that can be sustained by the environment over a long
term.

• Reliability: The craft must work reliably with minimal maintenance.

A pair wise comparison as seen in appendix was completed to establish


the following criteria weightings, which have been slightly adjusted as
seen in figure 3.

Figure 3. Relative Criteria Weightings


The design options as summarized in table where then scored on a scale
from 1 - 5 in a decision matrix. This decision matrix and an explanation of
the scoring scale can be seen in appendix . The best options as decided
by the decision matrix are shown in bold with all scores from the decision
matrix shown below.

Table 8. Design Options


Propulsion Fan: The main Paddle Wheel: Propellers:
System issue is that a The Two underwater
Options propeller above disadvantage propellers to
the water of paddle drive the craft
doesn’t provide wheels was due were found to
as much power to the size be the most
as below. requirement, as effective.
they would be
mounted on
the side of the
craft.

Score 2.3 2.91 3.3

Hull Material Pine: A pine Polystyrene: A Balsa Wood:


Options hull was found polystyrene The balsa wood
to be not hull was not as hull was found
buoyant sustainable as to score the
enough for the balsa wood. best on the
5cm draft decision matrix
requirement. criteria.

Score 3.4 3.2 3.5

Sensor Sonar: Sonar IR: An infrared IR and


Options (Note: satisfied range distance sensor Compass: The
These options requirements with a range of IR and
are all in greater than 10 -80cm was Compass
addition to the size of the found to be the option would
bump sensors tank, which is best due to its result in the
which are very unnecessary, low cost. best
simple and and more performance
inexpensive.) expensive but it would go
far beyond the
$50 budget.
Score 2.6 3.1 3.07

Battery Single 9V: A Single 7.2V: A 9V and 6V:


Options single 9V Single 7.2V The
battery may batter may combination of
result in errors result in errors a 9V and 6V
due to motor due to motor battery was
feedback. feedback, the determined by
battery also the matrix to
costs much be the most
more than effective due to
AAAs and 9Vs. low cost and
long life.

Score 2.8 2.9 3.3

Control Spiral: A spiral Random: A


Algorithm algorithm to random
Options cover the algorithm was
whole surface predicted to
area was have lower
determined to performance as
deliver the best there is no
performance. guarantee that
the whole area
would be
covered.

Score 3.1 2.7

Having assessed each design option it was decided that a balsa wood boat
powered by a 9V and 6V battery with two propellers would use infrared
and bump sensors to implement a spiral search pattern algorithm.

5.0 Final Design


5.1 Design Overview
The boat features a catamaran hull design with two 6V motors powering
propellers through gearboxes. The craft implements a spiral search
pattern algorithm using infrared distance and tactile bump sensors. The
design features are summarized in table below:

Table 9. Summary of design features


Design Design Feature
Element

Buoyancy Two L-shaped hulls made of balsa wood displace


4.06 x 10-3 m3 of water.

Propulsion Two 6V motors connected through gearboxes with a


12.7:1 gear ratio turn propellers of diameter 40mm.

Debris Debris is funneled through the front of the craft and


Collection and collected in a floating rope net made from bubble
Storage wrap.

Artificial The craft uses a spiral algorithm to cover the entire


Intelligence tank, and turn when stuck.

Sensors The craft has a Sharp GP2D12 Analog Distance


Sensor and two tactile bump Sensors

Microprocessor The craft uses an Arduino Uno with a 16Mhz


microprocessor to read sensors and control
actuators.

Battery A 9V 300mAh battery is used to power the Arduino


Solution and four 1.5V 1000mAh batteries are used to power
the motors.

Motors Each motor delivers 13200 rpm with a torque of


0.72mN, which is converted by the gearboxes to
1039 rpm with 9.2mN of torque.

5.2 Hull Design


The craft uses a catamaran style hull. This design features two wide L-
shaped hulls, connected via a flat deck on top.

The bottom section of each hull is 70mm wide at the widest point.
However, the hull tapers down to 50mm in width at the front of the craft.
This was done in order to work in conjunction with the funnel at the front
of the craft. A 5mm gap in the inner corner of each hull was included to
house the casing for the propellers. This was done to prevent any
unwanted movement in the propeller shafts.

Wide hulls were chosen in order to increase stability as well as to ensure a


relatively shallow draft. These hulls were spaced 60mm apart in order to
allow ping-pong balls (diameter of 40mm) to travel through freely.

The craft’s weight force is 6.63N. The dimensions of the hull result in a
draft of 25mm, which means the hull displaces 4.06 x 10-3 m3 of water,
see calculations in appendix . The centre of buoyancy, which is the point
that the upward buoyant force acts through was determined to be
12.5mm above the bottom of the craft. While the centre of gravity is 69
mm above the bottom of the craft, see appendix for stability calculations.

The upright component of each L-shaped hull is 40mm high, and 20mm
wide. This means that the overall height of the hulls is 75mm, allowing a
space of 50mm between the deck and the surface of the water. This
spacing was included in order to allow ping pong balls to pass through
freely and ensure all electronics remain dry.

These hulls were made out of balsa wood for optimal buoyancy and
durability. To prevent these hulls from swelling due to contact with water,
each hull was coated in Emerclad, a plastic paint. For further
waterproofing, as well as to prevent splintering and breakage, the hulls
were also covered in a layer cloth tape.

The deck, connecting these two hulls was made from a high density
polyethylene. This material was chosen as it was waterproof, cheap, and
strong even at a relatively small thickness. The deck is 2mm thick by
200mm wide by 200mm long. At the front of the craft, on either side, a
cutout measuring 45mm in length and 25mm in width was made to house
the gearboxes.

To fabricate the hull, a hacksaw was first used to cut the balsa wood to
length. These pieces of balsa were then assembled using hot glue. To cut
the deck material to size, a set of heavy duty scissors were used. Each
pontoon was then positioned on the outer edge of the deck and connected
using hot glue.

5.3 Collection System


The collection system consists of a funnel and floating rope net. The
funnel, situated at the front of the craft is composed of steel mesh
(flyscreen). This material was chosen as it is cheap, malleable and holds
its shape. Mesh was used in order to reduce drag in the water. This mesh
is in two pieces, bent at an angle, connected to the underside of the deck
with hot glue. The spacing between each piece spans 160mm at the front
and 40mm at the back. This was designed to create a one way flow, it is
easy for ping-pong balls to pass through, into the craft, but difficult for
them to float back out. This means that if the craft has to stop or reverse,
the collected balls will not be lost.

The floating rope net at the back is comprised of bubble wrap 460mm in
length. Bubble wrap was chosen as it is cheap and buoyant. This
buoyancy is desirable as it reduces the likelihood that the net will interfere
with the propellers. The net is attached to the craft with cloth tape to
allow for removal incase of damage.

The collection system involves balls being funneled through the front of
the craft, out to the rear of the craft. The balls are then contained by the
floating rope to allow for collection after deployment. The use of the
floating rope net theoretically provides a storage area of approximately
67388mm2 or 42 ping-pong balls.

5.4 Artificial Intelligence


5.4.1 Spiral Algorithm
The watercraft uses a spiral search pattern in order to cover the entire
surface area of the tank. This search pattern is achieved by controlling the
motors such that the distance sensor’s reading approaches a certain
variable, see software flowchart figure . This variable is then gradually
increased. The complete code running the watercraft can be viewed in
appendix.

The code used to determine the speed of the motors is as follows:

Lmotor = int(((float(-x)/maxcm)*(minmotor-maxmotor))+maxmotor);

Rmotor = int(((float(x)/maxcm)*(minmotor-maxmotor))+maxmotor);

Where:
x is the difference between the distance sensor reading and the
desired distance,
maxmotor is the maximum motor speed which is 255,
minmotor is the minimum motor speed,
maxcm defines the x value in cm when the motor should first reach
minimum speed.

Figure 4 graphically demonstrates these motor speeds and how they vary
based on the distance sensor reading.
Figure 4. Motor Speeds

The values of maxcm and minmotors were tested experimentally to find a


value that limited oversteering which was the main problem with the
algorithm while still allowing the craft to get back on track. The values of
minmotor = 95 and maxcm = 20 where determined to be the most
effective, these are the values used to generate the chart above.

5.4.2 Bump Sensor Input Handling


Bump sensor input is handled by reversing with the left motor at 100%
and the right motor at 50% for 8 seconds. It was found experimentally
that 8 seconds is the time taken to turn 90 degrees, which is the desired
turn when the boat is facing the wall.

5.4.3 Issue Detection by Analysis of Historical Sensor Data


Early testing indicated that it was possible for the craft to become stuck
driving into a wall even if its bump sensors were operational. The causes
of this include uneven surfaces, debris trapped between the craft and the
wall and insufficient force to depress the bump sensors.

In order to avoid this, the last 20 distance sensor values are stored in an
array which is updated every half second. If the last 20 values are all more
than 20cm from the desired distance sensor reading the boat realizes that
it is stuck. The boat then turns and continues to turn until the difference of
successive distance sensor readings becomes positive. This means it will
turn until it is just past parallel with the closest wall. It then resumes its
spiral algorithm. See the If (runningStuck) { } code section in the full
code listing in appendix for the code used to perform this maneuver.

5.4.4 Hardware
The algorithm runs on an Arduino Uno, which is a microcontroller board. A
microcontroller is a standalone integrated circuit including a simple CPU,
clock, Input/Output ports and memory. The arduino Uno features a 16MHz
microprocessor, 14 Digital I/O pins, 6 analog inputs, a USB connection and
a power jack.

5.4.5 Sensor Electronics


The main sensor used by the watercraft is a Sharp GP2D12 Analog
Distance Sensor. The sensor measures distances from 10 - 80cm. Since
the sensor is analogue, its output voltage has to be converted into a
distance by the software running on the Arduino board. In order to do this
an expression for distance based on voltage must be found. A regression
analysis of experimentally collected data points as seen in appendix was
used to find the following expression:

d = 26.58v-1.08

Where d is distance in cm and

v is voltage in volts.

The Sharp GP2D12 is comprised of an infrared emitter and sensor. The


emitter sends a pulse of IR light, which bounces off the nearest object as
seen in figure. The light then returns to the sensor where a lens focuses
the light onto an array of CCDs (Charge-coupled devices), which are
photoelectric devices used for sensing light. The angle of incidence is then
determined based on which CCD detects the most light. The triangulation
method then uses the angle of incidence to calculate the distance to the
object. This method is effective as both object colour and ambient light
have very little effect on the sensor’s performance. An electrical
schematic including the sensor’s connection can be seen in figure .

The watercraft also uses two tactile bump sensors mounted on the front of
the craft. These sensors are just switches, one pin is connected to the
arduino 3V output and the other pin is wired to a digital input see figure .
When the sensor is pressed this digital pin is raised to 3V, which the
software registers as a high signal. The
digitalRead(bumpPin) function is used to
detect if the bump sensor is depressed.

Figure 5. Tactile Bump Switch

Figure 6. IR Distance
Sensor
5.5 Power Supply
The watercraft is powered by two separate batteries as recommended by
D Robotics (2011). A standard 9V 300mAh battery is connected to the
Arduino UNO board. This battery delivers power to the Arduino, the
sensors and the motor shield electronics but not the motors. A separate
6V battery pack consisting of four 1.5V 1000mAh AA cells was connected
to the motor shield in order to power the motors. The table below outlines
current requirements for different components of the craft.

Table 10. Current Requirements (See appendix for calculations)


Predicted Actual Theoretic Actual
al Operating
Operation Time
Time

Arudino and 100mA 80mA 3.75h > 2h


Sensors

Motors 800mA 260mA 3.8h > 2h (Tested


in 4x30 min
sections)

As seen in the table above, the 9V battery will theoretically last for 3.7h
and the 6V battery will last for 3.8h, see appendix for calculations. The
L298P motor shield is rated for a maximum of 35V, 2A and 35W (Little Bird
Electronics, 2011). At 6V the motors consume 1.56W and 260mA, which
easily fall within the shield’s requirements. The main battery has a charge
capacity of 1000mAh. However, this is not the energy capacity of the
battery, the energy capacity is 6Wh. This is found by multiplying the
voltage by the charge capacity.

The reason for the separation of the motor battery and Arduino battery is
because of the electrical noise that is characteristic of DC motors. This
electrical noise results in an uneven voltage supply to the Arduino board,
which could result in failure (D Robotics, 2011). Although the noise could
be filtered using a capacitor, a different battery guarantees that the
microcontroller receives a consistent 9V supply.

5.6 Propulsion System


The mechanical drive system is comprised of two motors connected to
propellers via a gearbox and shaft. Each motor is connected directly into
the gearbox, which has a gear ratio of 12.7:1. The output shaft of the
gearbox is then connected with a small piece of rubber directly to the
propeller shaft which runs through the casing down to the propeller, see
figure . The output of each motor is 13200 rpm with a torque of 0.72mN.
The gearbox then converts this rpm into torque by a factor of 12.7. The
rpm becomes 1039rpm with 9.2mN of torque. Theoretically speaking the
propellers should run at these speeds although friction somewhat reduce
the speed in practice. The relationship between torque and current can be
loosely modeled by:

τ = k I1.5

Where τ is torque in N,

k is a constant dependant on the motor, and

I is current in Amps.

(Johns, 2003)

For our given motor, k = 5.4 x 10-3. Figure 7 shows this relationship
between torque and current.

Figure 7. Torque vs Current


6.0 FMEA

7.0 Life Cycle Assessment


7.1 Life Cycle Assessment Flow Chart

7.2 Life Cycle Assessment Table

7.3 Life Cycle Assessment Results Interpretation

8.0 Recommendations
In order to implement the automated watercraft design a number of
improvements must be made. Firstly, the

You might also like