2021 International Conference on Emerging Techniques in Computational Intelligence (ICETCI)

Trajectory Tracking of a 2-DOF Helicopter System

using Fuzzy Controller Approach

Abhishek Chaudhary Bharat Bhushan

Electrical Engineering Department Electrical Engineering Department
Delhi Technological University, DTU Delhi Technological University, DTU
New Delhi, India New Delhi, India
2021 International Conference on Emerging Techniques in Computational Intelligence (ICETCI) | 978-1-6654-1559-0/21/$31.00 ©2021 IEEE | DOI: 10.1109/ICETCI51973.2021.9574049

[email protected] [email protected]

Abstract—In this paper, the desired pitch and yaw axis Linear quadratic regulator (LQR) was reformed along with
trajectories of a 2-DOF helicopter system are tracked. For this PID to diminish the tracking error in the pitch and yaw output
purpose, fuzzy system is implemented and its performance is by enhancing the quadratic function using the state feedback law
compared with a conventional control approach, namely an LQR [5]. A gradient descent algorithm was proposed in achieving the
controller. In proposed fuzzy system, in order to obtain an optimal design of robust adaptive fuzzy controller [6]. Various other
fuzzy model for benchmark Unmanned Air Vehicle (2-DOF control techniques like model predictive control [7], adaptive
Helicopter), the parameters of the membership functions at the disturbance rejection controller [8] and adaptive second-order
antecedent part of the fuzzy rules are tuned. sliding mode controller [9] have also been used by different
researchers. During the real time operations of helicopter
Keywords— 2-DOF Helicopter; Unmanned Air Vehicles
(UAVs); Fuzzy Logic Rules; LQR; trajectory control; pitch angle;
system, various failure mechanisms occurs that affects the
yaw angle. performance of system and causes instability. To regulate the
higher order flight control actuator failures, a reorganized FTC
system has been considered [10].
I. INTRODUCTION
Fuzzy logic is a commonly used soft computing method
In recent years, the controllers of unmanned air vehicles
which is applied in various fields such as: pattern recognition,
(UAVs) is one emerging research area. Helicopter is one of the
telecommunication, control etc. Since the introduction of fuzzy
most commonly used vehicle in this area. Helicopters are mostly
sets by Prof. Lotfi A. Zadeh in 1965, fuzzy has been rapidly
used in fields like transportation, traffic control, firefighting, air-
developed to a promising research tool. By using linguistic rules,
sea rescue, etc. [1]. Helicopters are benchmark nonlinear
knowledge and human spirit can be integrated into any fuzzy
systems due to their instability and cross-coupling effects
logic system. With the support of fuzzy set philosophy and
between their yaw and pitch axes, which makes their position
mathematics, a fuzzy logic system can perform thinking
and trajectory control a difficult task to be handled.
according to the labelled linguistic rules. Fuzzy control is a
In the literature, the control of trajectory tracking in warm research topic, which has strained a great deal of attention
helicopter systems is still a significant problem [2]–[4]. In [2], from researchers working from fundamental research to
by using two controllers; a feedforward controller for trajectory industrial applications in the recent time. It has been perceived
tracking and a feedback controller for the stabilization of the that fuzzy control has advanced from model-free approach
trajectory tracking error, the trajectory tracking control of a 3- (Mamdani, 1993) to model-based approach.
DOF (Degrees-of-Freedom) helicopter is realized. The results
A dual fuzzy-neural network control is experimented in an
obtained illustrate the good tracking performance of the
intelligent landing system of an aircraft used in air
controllers when designed controller algorithm is tested during
transportation. The simulation studies of the same shows that the
experiment. In [3], for a small scale unmanned helicopter a
considered approach works quite well under the ambiguous and
heading reference trajectory and an inertial position tracking are
changing weather conditions [11]. In [12], the flight control of
studied. The helicopter dynamics are composed in two
an UAV is examined. For this purpose, three ANFIS (Adaptive
subsystems namely; heading-heave movement and longitudinal-
Neuro Fuzzy Interface System) controllers are deployed for the
lateral of the system and these subsystems are controlled by
three axes; pitch, yaw, and roll in order to control the speed, the
feedback loops for trajectory tracking errors. Performance of
altitude and the heading of the UAV. It is observed that the
this proposed controller is tested by using a realistic simulator,
employment of the method is able to meet the anticipated
namely X-Plane flight simulator and the simulation results show
specifications.
the effectiveness of approach proposed. In [4], a helicopter is
stabilized based on visual simultaneous mapping and The developed approach addressed the issues in configuring
localization. The attitude and the position control of the the control laws of fuzzy set rules. The results demonstrated the
helicopter is achieved by using parallel tracking and mapping boundedness of all the signals in the system and the asymptotical
(PTAM) method by using wireless camera and the results shows convergence of the tracking error of pitch angle and yaw angle
the efficiency of this proposed approached. at minimum value. The proposed approach has a self-organizing

Authorized licensed use limited to: Politecnico di Milano. Downloaded on July 09,2023 at 12:11:13 UTC from IEEE Xplore. Restrictions apply.
978-1-6654-1559-0/21/$31.00 ©2021 IEEE 159
 2021 International Conference on Emerging Techniques in Computational Intelligence (ICETCI)

capability and it is applied in documentation of nonlinear

dynamic systems. The obtained simulation results indicate high
tracking performance and fast response of the control system. In Non-Linear Equation for yaw axis:
this study, the fuzzy logic systems is used. The aim is to design 2
(𝐽𝑒𝑞,𝑦 + 𝑚ℎ𝑒𝑙𝑖 𝑙𝑐𝑚 𝑐𝑜𝑠𝜃 2 )𝛹̈ = 𝐾𝑦𝑝 𝑉𝑚,𝑝 + 𝐾𝑦𝑦 𝑉𝑚,𝑦 – 𝐵𝑦 𝛹̇
a fuzzy system in order to track the desired trajectories of a 2- +2𝑚ℎ𝑒𝑙𝑖 𝑙𝑐𝑚 𝑠𝑖𝑛θ𝑐𝑜𝑠θ𝛹̇ 𝜃̇
2
DOF helicopter. The conventional approaches in the training of
fuzzy systems is to tune the parameters of the antecedent
member- ship functions are used. These operators are very
simple and they do not have the associativity or commutativity
properties [13]–[15]. In [16], an intelligent technique is used to
control the non-linear coupling between two rotor motors of the
2-DOF Helicopter system using PID and Fuzzy controller. The
results obtained are compared with a conventional control
approach LQR.
This study is planned as follows; in section II, the
mathematical portrayal of the 2-DOF helicopter system is given.
In section III, fuzzy logic is discussed. In section IV, the
simulation results are given. In section V, the obtained results
are analyzed.

II. DESCRIPTION OF THE 2-DOF HELICOPTER SYSTEM Fig. 1. Simple free body diagram of 2-DOF helicopter [17]

The thrust forces acting on the pitch axis and the yaw axis
Fuzzy Controller with Triangular and Gaussian Membership from the front motor and back motor are then explained. Using
functions and a conventional control approach, LQR are the Euler-Lagrange formula, the nonlinear equations of motion
deployed on the 2-DOF helicopter shown in Fig. 1. It is a of the 2 DOF Helicopter system are obtained. These equations
extremely nonlinear multi-input-multi-output (MIMO) system are linearized about zero and the linear state-space model to find
with robust cross-couplings between the pitch and the yaw axes the position dynamics of the system. Given the state-space
due to the drive torques performing on the axes. Helicopter is representation
attached on a fixed base and its yaw and pitch (the back and the
front) propellers are energetic by two DC motors. The front 𝑥̇ = Ax + Bu
propeller handles the vertical motion of the 2-DOF helicopter
system about its pitch axis. This angle is well-defined as positive state vector for the 2 DOF Helicopter is:
when the front propeller sources a motion in the upward y = Cx + Du
direction. In case the propeller motors are not excited, i.e. when
the helicopter is at resting position, the pitch angle is about 𝑥 𝑇 = [θ(t),ψ(t)]
−40.50 and its gesture is limited between −40.50 and 40.50 . The output vector is:
back propeller handles the horizontal motion about the yaw axis.
𝑦 𝑇 = [θ(t), ψ(t)]
The 2-DOF helicopter is able to freely rotate all around
3600 in the yaw axis. The yaw angle is well-defined as negative
in the anti-clockwise and positive in the clockwise direction. The
two thrust forces are Fp and Fy shown in Fig.1, produced at the
distances 𝑟𝑝 and 𝑟𝑦 from the pitch axis and the yaw axis
respectively [17]. The voltages provided to the front propeller
and the back propeller motors are the inputs of the given
helicopter system and the pitch (θ) and the yaw (𝛹) angles in
radians are the two outputs of this benchmark 2-DOF helicopter
system. The aim is to propose a controller to track down the
desired trajectories in the pitch axis and the yaw axis of the
system. The dynamic nonlinear equations of the system acquired
by using Lagrangian mechanics are shown for the pitch and the
yaw axes as follows [17]:

Non-Linear Equation for pitch axis:

2
( 𝐽𝑒𝑞,𝑝 + 𝑚ℎ𝑒𝑙𝑖 𝑙𝑐𝑚 )𝜃̈ = 𝐾𝑝𝑝 𝑉𝑚,𝑝 + 𝐾𝑝𝑦 𝑉𝑚,𝑦 – 𝑚ℎ𝑒𝑙𝑖 g𝑙𝑐𝑚 cosθ
– 𝐵𝑝 𝜃̇ − 𝑚ℎ𝑒𝑙𝑖 𝑙𝑐𝑚 𝑠𝑖𝑛θ𝑐𝑜𝑠θ𝛹̇ 2
2
Fig 2 Quanser 2-DOF Helicopter [17]

Authorized licensed use limited to: Politecnico di Milano. Downloaded on July 09,2023 at 12:11:13 UTC from IEEE Xplore. Restrictions apply.
160
 2021 International Conference on Emerging Techniques in Computational Intelligence (ICETCI)

The Laboratory model view of Quanser 2-DOF Helicopter Fuzzy logic controller validates all the parameters as a
which is having all these above mentioned specifications is pleasing tool, and its reaction can diverge depending upon the
shown in Figure2 and the description of all non-linear nature of membership functions of input-output & also on the
parameters is given in Table 1. different set of rules being used in the FLC logic. For helicopter
controlling, T-S fuzzy model is used, and two inputs and one
output is provided for pitch angle error as shown in figure 4 as
TABLE 1 : Non-linear model parameters well as for yaw angle error as shown in fig 5. The one input is
Description Values change in pitch error and the other input is the pitch error.
Jeq,p The total moment of inertia about pitch 0.0384kg.m2 Gaussian type and Triangular type membership functions are
axis chosen in order to have continuous differentiable curve. Seven
lcm
Center of mass length along
0.1855m different linguistic variables have been chosen for every single
2-DOF helicopter body from pitch axis input state variable named as NS, ZE, PS, PM, NB, NM and PB
Jeq,y The total moment of inertia about yaw 0.0431kg.m2 with appropriate rule base. These are same for the pitch and yaw
axis input functions
Thrust torque constant acting on the
Kpp 0.2041N.m/V
pitch axis from pitch propeller
Thrust torque constant acting on the
Kpy 0.0068N.m/V
pitch axis from the yaw propeller
Thrust torque constant acting on the
Kyp 0.0219N.m/V
yaw axis from the pitch propeller
Thrust torque constant acting on
Kyy 0.072N.m/V
yaw axis from yaw propeller
g Gravitational constant 9.81m/s2
Bp Viscous damping about the pitch axis 0.8N/V
By Viscous damping about the yaw axis 0.318N/V
mheli The total moving mass of the helicopter 1.3872kg Fig. 3 Fuzzy Logic Controller Architecture
Vm,p Applied Voltage to the pitch motor ±24V
Vm,y Applied Voltage to the yaw motor ±15V

III. FUZZY LOGIC CONTROLLER

Fuzzy logic controllers are able to outline mankind
knowledge about a system with the help of membership
functions and using IF-THEN type of fuzzy rules. This methods
of making any machine behave like a human in a fuzzy manner
is coined as fuzzy logic controller. Fuzzy deals with issues such
as uncertainty, imprecise and qualitative decision making
problems. In the fuzzy application, we need to always remember
that it is a range to point control or vice versa. By describing the
associated membership functions, output can be extracted by the
fuzzification of both output and input. This whole Fig. 4 Membership functions of pitch control
implementation of fuzzy logic controller is carried out in mainly
three different steps: - fuzzification, Fuzzy interference process
and de-fuzzification, as shown in Figure 3. One measured input
and derivative of this measured input is applied to the fuzzy logic
controller and then fed to the 2-DOF helicopter and finally, the
output returned back to the input using closed loop feedback
process of the same unmanned vehicle (2-DOF helicopter).
For the monitoring of a system, a proper human thinking
base knowledge is obtained by this approach for manipulation,
representation and implementation of fuzzy logic controller. To
obtain an ideal fuzzy control model, the conventional approach
is to tune the membership function of the fuzzy logic controller.
It is one of the arrangements which is more likely to be closer to
that of the human thinking and more natural language than that
of the available traditional logical systems. The linguistic control Fig. 5 Membership functions of yaw control
strategy is converted into automatic control strategy which is
obtained by experiential knowledge.

Authorized licensed use limited to: Politecnico di Milano. Downloaded on July 09,2023 at 12:11:13 UTC from IEEE Xplore. Restrictions apply.
161
 2021 International Conference on Emerging Techniques in Computational Intelligence (ICETCI)

IV. THE SIMULATION ANALYSIS

2-DOF helicopter system is controlled by implementing
fuzzy logic controller to track an anticipated reference trajectory.
The 2-DOF helicopter system, a benchmark control
experimental setup is designed for laboratory researches,
includes one fixed base and two different propellers. The base
provides freedom for movement along its center. The propellers
are being rotated by dc motors which are fixed separately at both
the ends of these propellers. The main objective of these two
propellers is to control the elevation angle and the yaw angle of
2-DOF helicopter in the middle of operation. Yaw angle is being
controlled by one back end propeller while elevation angle is
being controlled by front end propeller. These propellers helps
in providing stable side motion.
The controlling challenge here is to deal with the cross
coupling effect which is formed by pitch and yaw motion
altogether. The block diagram of this Fuzzy system is given in
Fig. 6. The 2-DOF helicopter is a benchmark MIMO nonlinear
system and two controllers are designed, one for the pitch axis
and other for the yaw axis. The error 𝑒𝑝 (𝑡) is considered as the
difference between the anticipated pitch angle and the actual
pitch angle in radians. This error 𝑒𝑝 (𝑡) and the derivative of this
error ∆𝑒𝑝 (𝑡) are given as the tow inputs to the fuzzy logic
controller for the pitch axis of 2-DOF Helicopter system.
Similarly, the error 𝑒𝑦 (𝑡) that is considered as the difference
between the desired and the actual yaw angle in radians. The
error 𝑒𝑦 (𝑡) derivative of this error ∆𝑒𝑦 (𝑡) are given as the two
inputs to the fuzzy logic controller for the yaw axis of 2-DOF
Fig.7 Sinusoidal reference trajectory for pitch axis and step reference
Helicopter system. Now in our implementation, each controller trajectory for yaw axis.
has two inputs and each input is fuzzified by using different
seven Gaussian membership functions. These membership
functions are distributed in the closed intervals and have been
depicted in Fig. 4 and Fig.5 for fuzzy logic controllers.

Fig.6 Block Diagram of the proposed fuzzy System

The consequential part of the IF-THEN logic type fuzzy

rules is tuned to obtain the optimal fuzzy logic control model.
The membership functions at the antecedents are tuned in this
process. Several number of simulation results are carried out.
The results obtained for different reference trajectories, during
simulation, are given in Figs.7-9. In Fig. 7, a sinusoidal reference
trajectory is given to the pitch axis and a step reference trajectory
is given to the yaw axis. In Fig. 8, a step reference trajectory
given to the pitch axis and a sinusoidal reference trajectory given
to the yaw axis. Fig.8 Step trajectory for pitch axis and Sinusoidal reference trajectory for
yaw axis.

Authorized licensed use limited to: Politecnico di Milano. Downloaded on July 09,2023 at 12:11:13 UTC from IEEE Xplore. Restrictions apply.
162
 2021 International Conference on Emerging Techniques in Computational Intelligence (ICETCI)

V. RESULT ANALYSIS
The study of helicopter model has been done by
implementing conventional LQR controller and fuzzy
controller using Simulink/MATLAB. The sinusoidal and step
input trajectories are considered. Table2 represents time
response analysis in order to have the performance verification
of the implemented controllers. The desired pitch and yaw axis
trajectories of a 2-DOF helicopter system are tracked and
compared in this section.

TABLE 2 : Time Response Results

Settling Time Peak Over-shoot
(milli seconds) (in %)
LQR 4.72 8.99
FUZZY 0.95 0.5

The response after implementing LQR and Fuzzy

controllers on 2-DOF helicopter system are discussed in Fig. 9
and Fig. 10 respectively.

Fig.10 Response after implementing fuzzy Controller on 2-DOF

Helicopter System.

VI. CONCLUSION
This paper describes the implementation of Fuzzy
controller on helicopter system. The conventional LQR results
are compared with intelligent fuzzy control. Analysis of
implemented control techniques have been done for yaw and
pitch angles. Fuzzy is employed to provide desired trajectory
and eliminates cross-coupling between the rotors. The
performance of the LQR & fuzzy is assessed by using
Simulink/MATLAB. For the comparison of the controller’s
results the time response analysis has been done. Fuzzy
controller shows its excellence when introduced to steady-state
error and settling time with respect to LQR control.

REFERENCES

[1] M. Hernandez-Gonzalez, A.Y. Alanis, and E.A. Hernandez-Vargas,

“Decentralized discrete-time neural control for a Quanser 2-DOF
helicopter,” Applied Soft Computing, vol. 12, no. 8, pp. 2462-2469, 2012.
[2] T. Kiefer, K. Graichen, and A. Kugi, “Trajectory Tracking of a 3DOF
Fig.9 Response after implementing LQR Controller on 2-DOF Helicopter Laboratory Helicopter Under Input and State Constraints,” IEEE Trans.
System. on Control Systems Technology, vol. 18, no. 4, pp. 944-952, 2010.
[3] I.A. Raptis, K.P. Valavanis, and G.J. Vachtsevanos, “Linear Tracking
Control for Small-Scale Unmanned Helicopters,” IEEE Trans. on Control
Systems Technology, vol. 20, no. 4, pp. 995-1010, 2012.

Authorized licensed use limited to: Politecnico di Milano. Downloaded on July 09,2023 at 12:11:13 UTC from IEEE Xplore. Restrictions apply.
163
 2021 International Conference on Emerging Techniques in Computational Intelligence (ICETCI)

[4] K. Tanaka, H. Ohtake, M. Tanaka, and H.O. Wang, “Wireless Vision- [11] K. Xu, G. Zhang, and Y. Xu, “Dual Fuzzy Neural Network Control in
Based Stabilization of Indoor Microhelicopter,” IEEE/ASME Trans. on Civil Aviation Intelligent Landing System,” AICI 2011, Part II, LNAI
Mechatronics, vol. 17, no. 3, pp. 519-524, 2012. 7003, pp. 166-174, Springer-Verlag Berlin Heidelberg, 2011.
[5] C. Liu, J. Pan, and Y. Chang, “PID and LQR trajectory tracking control [12] S. Kurnaz, O. Cetin, and O. Kaynak, “Adaptive neuro-fuzzy inference
for an unmanned quadrotor helicopter: Experimental studies,” in Proc. system based autonomous flight control of unmanned air vehicles,”
35th Chin. Control Conf. (CCC), Jul. 2016, pp. 10845–10850. Expert Systems with Applications, vol. 37, no. 2, pp. 1229-1234, 2010.
[6] M. Jahed and M. Farrokhi, “Robust adaptive fuzzy control of twin rotor [13] I. Batyrshin and O. Kaynak, “Parametric classes of generalized con-
MIMO system,” Soft Comput., vol. 17, no. 10, pp. 1847–1860, Oct. 2013. junction and disjunction operations for fuzzy modeling,” IEEE Trans. on
[7] J. Gao, P. Wu, T. Li, and A. Proctor, “Optimization-based model Fuzzy Systems, vol. 7, no. 5, pp. 586-596, 1999.
reference adaptive control for dynamic positioning of a fully actuated [14] I. Batyrshin, O. Kaynak, and I. Rudas, “Fuzzy modeling based on
underwater vehicle,” Nonlinear Dyn., vol. 87, no. 4, pp. 2611–2623, 2017. generalized conjunction operations,” IEEE Trans. on Fuzzy Systems, vol.
[8] S. Ghabraei, H. Moradi, and G. Vossoughi, “Multivariable robust 10, no. 5, pp. 678-683, 2002.
adaptive sliding mode control of an industrial boiler–turbine in the [15] A.C. Aras, O. Kaynak, and R. Abiyev, “Slip control of a quarter car
presence of modeling imprecisions and external disturbances: A model based on type-1 fuzzy neural system with parameterized
comparison with type-I servo controller,” ISA Trans., vol. 58, pp. 398– conjunctions,” IECON 2012 - 38th Annual Conference on IEEE Industrial
408, Sep. 2015. Electronics Society, pp. 2488-2493, 2012.
[9] S. Mondal and C. Mahanta, “Adaptive second-order sliding mode [16] R. Singh, B. Bhushan, A. Varshney, “Real-Time Simulation and Control
controller for a twin rotor multi-input-multi-output system,” IET Control of Helicopter Systems,” 2019 International Conference on Power
Theory Appl., vol. 6, no. 14, pp. 2157–2167, Sep. 2012 Electronics, Control and Automation (ICPECA).
[10] J. D. Boskovic and R. K. Mehra, “A decentralized fault-tolerant control [17] Quanser Inc., Quanser 2-DOF Helicopter Manual, Tech. rep., Quanser
system for accommodation of failures in higher-order flight control 2010.
actuators,” IEEE Trans. Control Syst. Technol., vol. 18, no. 5, pp. 1103–
1115, Sep. 2010.

Authorized licensed use limited to: Politecnico di Milano. Downloaded on July 09,2023 at 12:11:13 UTC from IEEE Xplore. Restrictions apply.
164