An autonomous Hovercraft with Minimum Energy

Consumption
T.A. Anter Y. H. Hossamel-din S. M. Abdrabbo
Mechatronics Eng. Department Mechatronics Eng. Department Vice Dean, Faculty of Engineering
Higher Technological Institute (HTI) Future University in Egypt (FUE) Sinai University
Tenth of Ramadan city, Egypt Cairo, Egypt El_Arish, Egypt
[email protected] [email protected]
[email protected]

Abstract—The objective of this work is to introduce an

autonomous Hovercraft with minimum Energy consumption. To
achieve this goal, a computer simulation in addition to practical
testing of a simple hovercraft model have been carried out. A
mathematical model for the hovercraft is considered and
simulated in the MATLAB/SIMULINK environment. A
navigation system with Global Positioning System (GPS)
integrated with an Inertial Measurement Unit (IMU) sensors are
used to monitor the speed, position, and the direction of the
hovercraft for the autonomous operation. Two motors with
propellers are used for the lifting and thrust systems, while a
third one is used for the rudder movement that controls the Fig. 1. Schematic of a hovercraft [1]
direction. Specific resistance (Ɛ) index is used to test the
hovercraft performance. The results of the practical experiments The previous studies conducted by several researchers
of moving the hovercraft between two and three points are focus on different types of hovercraft ranging from the human
compared with that of the computer simulation. The distance driven hovercraft to the remotely controlled ones.
error from the actual target has been found to be in the range of Some of these researchers focus on the dynamical
about 18%, which proves the idea. It should be noted that this
modeling [1, 2, 3, 4, 5]. Others consider testing the stability
significant error is due to the fact that the GPS considered is a
low price commercial one and using more accurate one will result and controllability of the systems [6, 3, 5, 7].
in more satisfactory results. Based on the computer Proportional Integral Derivative (PID) controller is
simulation and practical testing results, a look up table selected to control the hovercraft [6, 2].
has been prepared to help in deriving an algorithm for Design and development of a hovercraft prototype with
controlling the RPM of lifting and thrust Fans based on the full basic functions using two or four fans are considered in [6,
minimum optimum specific resistance 8, 9]. Remote control via wireless (R/C) for the position
tracking has been used by [10, 3, 4].
Recently, autonomous hovercraft is studied with a
Keywords—component; Hovercraft; Navigation; Dynamic navigation system for its trajectory control [1, 10].
Modeling; GPS; IMU; PID; Specific Resistance. In reference [11], the specific resistance has been
determined and reported for various transportation means but
I. INTRODUCTION not including the hovercraft vehicles.
An autonomous hovercraft with three actuator inputs In this work, a computer simulation in addition to
controlled by a microcontroller with the necessary information practical testing of a simple hovercraft model have been
from a GPS and an IMU sensors has been designed, carried out. In order to study the specific resistance of the
implemented, and tested. The considered hovercraft comprises hovercraft vehicles; different combinations of thrust and
two fans (propellers): One is used to inflate the "skirt" under lifting forces are considered and the corresponding specific
the vehicle, while the other provides the thrust force. The resistances are determined and reported. This may help in
inflating propeller is responsible for lifting the hovercraft by designing an optimum control system for speeds of both
forcing air (increasing pressure) under the hovercraft causing it thrust and lift fans of the hovercraft based on minimum
to raise above the surface. Also high air pressure inside the energy consumption.
cushion enables the hovercraft to float and move smoothly on
II. PROPOSED HOVERCRAFT
different land surfaces. The other propeller provides thrust
force, as the air leaves the fan and directed by the rudder A. Mechanical System
enabling the hovercraft to go straight or to turn as required. Referring to Figure. 2 and Fig. 3, a small model of the
Fig. 1 shows a schematic of this hovercraft. hovercraft with two brushless motors integrated with two

978-1-5/$31.00 ©201 IEEE 1

Authorized licensed use limited to: R V College of Engineering. Downloaded on May 12,2023 at 04:51:31 UTC from IEEE Xplore. Restrictions apply.
 Fig. 4. The Hovercraft thrust and direction inputs

Fig. 2.The proposed hover craft

Governing equations:
Referring to Fig. 4 and Considering Newton’s second law,
the following equations can be arrived at:
In the X direction:
୶ ሺͳ ൅ ‘•ሺɁሻሻ ܾ௫ —
—ሶ ൌ  െ ൅ ɘ˜ െ
ʹ 
ଶ
Ɋଵ  ሺ‰ െ ୪ ሻƒ”…–ƒሺͷͲͲͲ—ሻ (1)
஠
In the Y direction:
୶ •‹ሺɁሻ ܾ௬ ˜
˜ሶ ൌ  െ െ —ɘ
ʹ 
ଶ
െɊଶ  ሺ‰ െ ୪ ሻƒ”…–ƒሺͷͲͲͲ˜ሻ ሺʹሻ
஠

Fig. 3. The hovercraft main dimensions. In the Yaw direction (Rotation about Z axis):
(All dimensions are in mm)
୊౮ ୢ ୗ୧୬ሺஔሻ ୠಐன
propellers is used. One is for the thrust force and the other is ɘሶ ൌ  െ  
ଶ୍ ୍
for the lifting force. ଶ௘
െɊଷ  ሺ‰ െ ୪ ሻƒ”…–ƒሺͷͲͲͲɘሻ (3)
An electronic control system for a remotely & a ஠
completely autonomous vehicle has been designed and built. Where;
A servo motor with a rudder is used for the steering control. ܾ௫ … is the translational viscous coefficient of friction in X
direction (kg/s),
B. Control Board and Sensors ܾ௬ … is the translational viscous coefficient of friction in
An Ardu Pilot Mega (APM) 2.5 board is used as the main Y direction (kg/s),
controller for the hovercraft with an IMU sensor built in it and „஘ … is Rotational viscous coefficient of friction (kg
a GPS sensor are being considered as part of the designed m2/sec),
control system. A radio telemetry 915 MHz wireless module † ….. is the distance between the center of gravity of the
is being used with the board for wireless connection with the hovercraft and the point where the rudder is attached (m),
monitoring computer. e … is the difference between geometric axis and
A Radio Controlled (RC) remote control is used for the principle inertia axis (approximately equals to 0.001 m),
manual remote control. Two Electronic Speed Controllers Fl ….. is the lifting force (N),
(ESC) are used for the speed control of the thrust motor and Fx ….. is the thrust force (N),
the lift motor using PWM technique. ‰ ….. is the gravitational acceleration (9.81 m/s2),
….. is the Moment of inertia of the hovercraft around the
vertical axis (kg m2),
III. MATHEMATICAL MODELING
 ….. is the Mass of the hovercraft ( Kg),
Referring to Fig. 4, the hovercraft has one thrust fan Ɋଵ , Ɋଶ and Ɋଷ ….. are the Friction coefficients in u & v
with a rudder to direct the air according to the steering signal directions; respectively, and
from the control unit. The thrust fan force (Fx) with the rudder Ɂ….. is the rudder angle (rad).
angle (δ) provide the inputs for the system.

Authorized licensed use limited to: R V College of Engineering. Downloaded on May 12,2023 at 04:51:31 UTC from IEEE Xplore. Restrictions apply.
 IV. CONTROLLER DESIGN proportional, integral, and derivative gains (KP, KI, & KD) as
PID controller is used as a controlling method in both follows:
simulation and experimental tests considered in this work. KP = 0.6 *Ku
KI = 2*Ku / Tu
A. Linearization : KD = Ku * Tu /8 (5)
x Speed control:
For linearization, the following are assumed:
The simplified transfer function for the speed in X
x The tangent plane (North, East, Down) is taken as the direction is given by:
local navigation frame.
ଵ
x The tangent frame is obtained by attaching a plane to ‫ܩ‬௨ ሺܵሻ ൌ  (6)
଴Ǥ଺ହାଶǤଵௌ
the surface of the Earth at a specific point.
x Coriolis acceleration is neglected considering small Using Ziegler-Nichols’ tuning method, explained above, the
speeds. control gains are as follow:
x There is no contact with the ground and accordingly,
the coulomb friction will be zero as maximum lift KP = 4.8, KI = 16, KD = 1
force is assumed.
୊౮ ୊౮
x F1=
ଶ
ሺͳ ൅ ‘•ሺɁሻሻ and F2 = ଶ
•‹ሺɁሻ

B. State-space model:
x Position control:
The transfer function of X direction position is
Accordingly; the linearized dynamic equations of the described by the following equation:
hovercraft system in state-space form are written as
ଵ
follows: ‫ܩ‬௨ ሺܵሻ ൌ  (7)
ୗሺ଴Ǥ଺ହାଶǤଵௌሻ

െܾ
Ɂ—ሶ ‫ ۍ‬ൗ݉ Ͳ Ͳ ‫ې‬
Ɂ— By the same method, the PID controller parameters
අ Ɂ˜ሶ ඉ = ‫ێ‬ Ͳ െܾ ൗ Ͳ ‫ ۑ‬අ Ɂ˜ ඉ for position control are found to be:
݉
‫ێ‬ െ„፽ൗ ‫ ۑ‬Ɂɘ
Ɂɘሶ
‫Ͳ ۏ‬ Ͳ ‫ے‬ KP = 6, KI = 6, KD = 2.5
ͳൗ Ͳ Fig. 5 shows the step response for the speed in X direction

+൦ Ͳ ͳൗ ൪ ቂ ͳቃ (4) controlled by the gains calculated above.
 ʹ
† Fig. 6 shows the step response for the position in X
Ͳ ൗ direction.

C. Stability and Controllability Checking:

The poles of the above system were found to be:

{-4.4444, -0.3095, -0.3095}; which have negative values and
accordingly it is stable.

By checking the controllability of the system it is found

that the rank of the controllability matrix which is 3X3 equals
to 3 i.e. it is full rank, and accordingly; the system is
controllable. Therefore, the provided system is verified to be
controllable and a controller can be designed to achieve the
given requirements.

D. PID Controller:
The PID gains have been tuned using Ziegler-
Nichols’ tuning method. The integral and the derivative gains
of the PID controller are firstly set to zero and the proportional
gain is increased until the system starts to oscillate getting the
oscillation period (Tu ) and the ultimate gain (Ku). The
obtained values for Tu and Ku are then used to set the Fig. 5. Step response for speed in x direction

Authorized licensed use limited to: R V College of Engineering. Downloaded on May 12,2023 at 04:51:31 UTC from IEEE Xplore. Restrictions apply.
 Fig. 7. Block diagram representation of hovercraft

The system shown in Fig. 7 has been checked both by

practical testing and by simulation.

A. Practical Testing :
The flow of information starts with the user who
Fig. 6. Step response for postion iIn x direction inputs the Coordinates and the waypoints to the computer.
Then the APM board read the sensors (GPS and IMU) values
to begin the autonomous navigation. These values, such as
V. NAVIGATION SYSTEM position and speed, are passed to the control algorithm. The
The navigation is the translation between two coordinates. job of the control algorithm is to take the inertial
So the navigation process starts by determining the targeted measurements and determine what the actuator output should
points and their Cartesian coordinates. be, then it is translated into a PWM signal and passed along to
The considered vehicle is designed to work in two modes:- the ESC’s which directly control the actuators (servo and
- Remotely controlled mode. thrust motor). The lift fan is controlled directly by the RC
controller. The actuator activity causes a physical result to the
- Completely Autonomous mode.
hovercraft which influences the state of the inertial sensors,
starting the information flow process again. At this point, the
In the remotely controlled mode, RC remote control is
APM also updates the ground control station with the sensor
being used directly with the actuators; while in the
output which logs for later use. After the telemetry data for the
autonomous mode, an APM 2.5 board including a GPS and
parameter of interest is compared to the desired set-point a
IMU sensors are used for autonomous navigation.
PID control is used with the control algorithm to minimize the
The motion of the hovercraft requires two stages, turning
overshoot.
and moving forward. In the first stage, the hovercraft turns
until it becomes in line with the point of interest. Once the first B. Simulation:
stage is achieved, the second stage will follow, such that the
Fig. 8 shows the SIMULINK Model of the model given by
hovercraft moves until it becomes within a specific distance
equations 1, 2, & 3 and the designed position controller
from the point of interest. The control algorithm will reduce
following parameters data are considered:-
the amount of error when the hovercraft is turning, as well as it
is moving closer to the point of interest. Fig.7 shows a block Mass of the hovercraft, m, = 2.1 kg; Translational
diagram representation of hovercraft control system. viscous coefficient of friction, b, = 0.65 kg/sec.; Rotational
viscous coefficient of friction, „஘ ǡ = 0.03 kg m2/sec; Moment
of inertia, I, = 0.045 kg m2; and the Friction coefficients are
assumed to be Ɋଵ =0.1, Ɋଶ =0.01 and Ɋଷ = 0.004 [1].

Authorized licensed use limited to: R V College of Engineering. Downloaded on May 12,2023 at 04:51:31 UTC from IEEE Xplore. Restrictions apply.
 Fig. 8. SIMULINK closed loop model
C. Trajectory Tests Results:
Fig.9 & Fig.10 show the simulation and field testing results
of two-point and three-point trajectory assignments. From both
figures, it can be seen that the actual proposed vehicle could
accomplish the targeted assignment in the autonomous mode
with an error of 2 meters (18%) in the distance. This proves
the idea and it should be noted that this significant error is due
to the fact that the GPS considered is a low price commercial
one and using more accurate one will result in more
satisfactory results.

Fig. 10. Three point trajectory

VI. SPECIFIC RESISTANCE

This section illustrates the price to pay in terms of
energy consumed represented by power efficiency for the
hovercraft. In this work, an index is used to analyze the
efficiency of the hovercraft which is the specific resistance (Ɛ).
According to reference [11], the specific resistance of the
vehicle (Ɛ) is given by:

Ɛ = P/WV (8)
Fig. 9. Two point trajectory Where;
W is the total weight of the vehicle (N),
V is the speed of the vehicle (m/s), and
P is the total power (Watt) which is given by:

Authorized licensed use limited to: R V College of Engineering. Downloaded on May 12,2023 at 04:51:31 UTC from IEEE Xplore. Restrictions apply.
 P= ܲ௟ ൅  ܲ௧ (9)
Where;
ܲ௟ Lifting power (Watt), and
ܲ௧ Thrust power (Watt) TABLE 1 OPTIMUM SPECIFIC RESISTANCE ( SANDY ROAD)

Hovercraft Specific
Experimental tests were applied to the proposed hovercraft
Speed (m/s) Resistance Lifting Thrust
with different ground surfaces. The hovercraft was run with
(optimum) Fan RPM Fan RPM
different combinations of thrust & fan RPMs for a constant
hovercraft speed. The specific resistance was calculated for 0.3 0.380754 1400 2800
each and the values of the RPM of thrust and lifting fans that 3266.66
give minimum specific resistance, Ɛmin is considered. This 0.5 0.280163 1400 7
condition corresponds to maximum efficiency of the 0.6 0.255016 1400 4800
hovercraft and accordingly minimum energy consumption. 0.8 0.263402 2750 3500
A. Testing Results: 0.9 0.32634 4700 3100
Fig. 12 shows the results by testing the hovercraft on a 1 0.299246 4800 3200
sandy road with constant speed 1 m/s. As shown in the figure; 1.3 0.242975 4700 3500
as the lifting fan RPM increases the thrust fan RPM decreases,
the specific resistance decreases. This is until the lifting fan 1.6 0.318829 4700 4800
speed reaches 4800 RPM at which the specific resistance 1.7 0.409199 4700 5800
starts to increase again. Accordingly speeds of 4800 RPM and 1.9 0.442512 6000 4800
3200 RPM for lifting and thrust fans give the minimum
2 0.513142 6000 5800
specific resistance which results in maximum efficiency and,
accordingly minimum energy consumption.
Similar tests are repeated for different constant speeds. Fig. 13 shows the minimum specific resistance for
Table 1 gives the testing results for the RPM values for thrust different speeds at different roads. The figure shows that the
and lifting fans that give minimum specific resistance minimum specific resistance values depends not only on speed
corresponding to each speed considered. This can be used as a but also on the road type. More research is needed in this area.
look up table to select the fan speeds for both lifting and thrust B. Specific Resistance for Different Transportaions Means:
forces. The specific resistance was determined in reference [11] for
various transportation means not including the hovercraft
vehicles. Fig. 14 shows the results obtained in present work
plotted over the graph given by [11]. As shown in the Figure,
Hovercraft Speed 1 m/s the specific resistance is different in wide range from
Road(1) minimum of 0.004 to 0.05 (ships) to maximum of 10 to 12 for
hexapod.
Mechanical Method
10000 1.2

1 0.8
8000
Thrust Fan RPM

Specific Resistance

0.6
Specific Resistance

0.8
Optimum

6000
0.6 0.4
4000 0.2
0.4
2000 0
0.2
0 0.5 1 1.5 2 2.5
0 0 Hovercraft Speed (m/s)
0 2000 4000 6000 8000
Road (1) Road (2) Road (3)
Lifting Fan RPM
Road (4) Road (5)
Thrust Fan RPM Specific Resistance

Fig. 13 Specific resistance optimum values

Fig. 12 Hovercraft Speed 1 M/S on Road 1 Test

Authorized licensed use limited to: R V College of Engineering. Downloaded on May 12,2023 at 04:51:31 UTC from IEEE Xplore. Restrictions apply.
 x Using the proposed small real model, the energy
consumed represented by power efficiency for the
hovercraft has been investigated. This is carried out using
an index called specific resistance.
x Based on these investigations, a look up table has been
prepared to help in deriving an algorithm for controlling
the RPM of lifting and thrust Fans based on the minimum
optimism specific resistance. More work is still needed
in this area.
REFERENCES

[1] R. Sanders, “Control of a Model Sized Hovercraft”,

Research report, the University of New South Wales
Australia., 2003.
[2] W. Cheng-long, L. Zhen-ye, F. Ming-yu and B. Xin-qian,
“Amphibious Hovercraft Course Control Based Adaptive
Multiple Model Approach”, presented at the International
Conference on Mechatronics and Automation, Xi'an,
China, 2010, DOI: 10.1109/ICMA.2010.5588410 .
[3] H. Lindsey, “Hovercraft Kinematic Modelling”, Research
report, Center of Applied Mathematics University of St.
Thomas, St Paul, USA, 2005.
[4] Z. Kulis, "Feedback Control Of A Hovercraft Over A
Wireless Link (calibration)", Master of Science thesis,
Fig. 14 Specific resistance for different transportations Department of Electrical and Computer Engineering,
means [11] University of Maryland, , Maryland, USA, 2006.
[5] A. Lewis, & D. Tynery, "Controllability of a hovercraft
The present work study of the considered hovercraft model (and two general results)", Proceedings of the 43rd
IEEE Conference on Decision and Control, Atlanta,
vehicle shows a full range of specific resistance from 0.2 to 2004, pp. 1204-1209 .
1.2 while that for the optimum (minimum) ranges from 0.2 to [6] A. L. Marconett, “A Study and Implementation of an
0.73. Autonomous Control System for a Vehicle in the Zero
Drag Environment of Space”, University Of California
Davis, California, USA, 2003.
VII. CONCLUSIONS [7] T. Denery, "Multi-Domain Modeling of the Dynamics of
A small model of a hovercraft has been designed, a Hovercraft for Controller Development"”, Modeling
and simulation technolgies conference and exhibit,
implemented, and tested using three actuators for: lifting, American Institute of Aeronautics and Astronautics, -
thrust and steering. A navigation control system has been San Francisco, California, USA 2016.
implemented in two different modes: Remotely and [8] B. Okafor, "Development of a Hovercraft Prototype",
completely autonomous modes using GPS and IMU sensors International Journal of Engineering and Technology,
and an electronic control unit to get location information and Vol 3 No. 3, March, 2013. pp. 276-281.
give the correct decisions for the actuators. Also a theoretical [9] S. Rao,& V. Prakash, "Development of a Integrated Air
model has been developed and simulated in Cushioned Vehicle (Hovercraft)", International Journal
Of Modern Engineering Research (IJMER), Vol. 4, May.
MATLAB/Simulink environment. For the considered model 2014.pp. 21-28.
with data given in section (IV) and from simulation runs and [10] P. Aguiar, L. Cremean and J. Hespanha, “Position
experimental tests the following can be concluded:- Tracking for a Nonlinear Underactuate Hovercraft:
x The actual proposed vehicle could accomplish the targeted Controller Design and Experimental Results”, Decision
assignment in the autonomous mode with an error of 2 and Control, 2003. Proceedings. 42nd IEEE Conference, ,
Hawaii, USA ,vol 4, pp. 3858-3863 ,2003.
meters (18%) in the distance. This proves the idea and it [11] G. Gabrielli and T. Kármán, "What price speed? Specific
should be noted that this significant error is due to the fact power required for propulsion of vehicles", Mechanical
Engineering , ASME, vol.72, 1950. pp. 775-781
that the GPS considered is a low price commercial one.
Using more accurate one will result in more satisfactory
results.

Authorized licensed use limited to: R V College of Engineering. Downloaded on May 12,2023 at 04:51:31 UTC from IEEE Xplore. Restrictions apply.