Académique Documents
Professionnel Documents
Culture Documents
net/publication/307575896
CITATIONS READS
0 166
3 authors:
1 PUBLICATION 0 CITATIONS
The LNM Institute of Information Technology
7 PUBLICATIONS 5 CITATIONS
SEE PROFILE
SEE PROFILE
Abhishek Sharma
The LNM Institute of Information Technology
28 PUBLICATIONS 151 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Real-Time Traffic Light Signal Recognition System for a Self-driving Car View project
All content following this page was uploaded by Narendra Khatri on 15 January 2018.
Abstract— Recently small UAVs are widely used in real would also have a huge impact on the overall design, for
world applications. In the absence of the pilot, Flight control which code parallelization provides faster speed in FPGA and
system (FCS) is the brain of the any aircraft especially in the
UAVs. FCS can use various controllers such as conventional other hardware components. Small UAV is constraint to less
PID Controller, combination of PID controller with Kalman volume and weight. As per current trend in modern UAV
filter, fuzzy logic controller and FPGA /DSP based controller, development, FPGAs severed as the main architecture
etc. This paper presents, brief explanation of four techniques
platform in flight control system to provide quick and
used in the development of the flight control system for small
UAVs namely, purpose of work, flow diagram, advantages and efficient response with respect to proof of concept and time.
disadvantages. Field Programmable Gate Array (FPGA) /DSP
based small UAVs are better one, as it is a low power, fast In fig. 1 Flight control system in the form of an
response, less volume and compatibility to applications. The aim
autopilot, having component such as sensor, Airframe,
of this paper is to achieve a suitable characteristics compatible
with applications in the field of unmanned aviation. controller, Navigation signal and servo etc. UAV flight path,
attitude, altitude, airspeed are measured and compared to
Keywords—UAVs, FCS, FPGA, PID, LUT, RTOS, FCS20 desired state in Automatic closed loop control system.
I. INTRODUCTION Sensors and actuators are important blocks in FCS.
The sensors measure the air vehicle’s attitude, Discretization provides a discrete approximation of
angular rate, airspeed, heading, altitude and other functions the continuous problem in comparison to continuous
are measured and compared to desired states then error signal optimization which are very time consuming. Discretization
is eliminated by control surface. Controller also important provide optimum solution as well as substitution of time
block in FCS, The modification and combining of signals consuming problem so output of PID controller are
from the different axes is accomplished in the controller. In discretized. After discretization the Simulink model is
this paper four types of controllers are discussed as PID deployed in FPGA platform [3]. Parallel PID controller is
controller, PID with KALMAN filter, fuzzy logic controllers faster but the performance is not comparable to cost paid.
and FPGA /DSP based controller. Increasing critical path in PID controller reducing the
throughput of the controller. So critical paths of these
II. PID CONTROLLER controllers are broken down using appealing pipelining and
re-timing techniques to improve throughput of the controller.
Small UAVs are controlled by PID controllers.
Improve throughput of parallel PID controller in small UAVs
Nonlinear [3] model of UAVs are implemented by Aerosim
with having faster performance using proper suitable
blockset in Mat lab Simulink. In fig. 2 the representation of
algorithm. PID controllers have slow response, overshoot,
process flow of the small UAVs by PID controller is shown.
long adjustment also have limitations in optimality and
Longitudinal and lateral (linearization) planes are separated
robustness. PID Controllers are mostly used to control the
in controller after identifying the potential problems [2].
feedback mechanism, and are known for their robustness,
which is the main reason of its usefulness in small UAVs.
Open loop nonlinear model
III. KALMAN AND PID CONTROLLER
Linearization of model
Linear Model
No KALMAN Filter
Desired
Results
Yes
Desired
Discretization Process Results
No
Kalman and PID controller offers short transition, Defuzzification: the fuzzy output variables are decoded into
good stability, anti-disturbance, good control and fulfills the desired values. MFLIPS using the Centre of Gravity
requirement of real-time and accurate control. Using defuzzification method.
combination of Kalman and PID controller, UAV can avoid
FPGA board come with multiple variants of logic
collision accident and accurate control. In fig. 3 the output of elements, user interface, IP generation tools and memory
the PID controller is passed through Kalman filter. Kalman options viz Artix7, Virtex-V and Cyclone II for architecture
filter is added after PID controller. This is the one change development. In general, the UAV organization includes a
from fig.1 to achieve better performance from previous one, 16-bit parallel and pipelined structure implemented using the
in this fig.3 all block work same as previous one. Accuracy integrated development environment provided by vendor.
and Stability are two important factor to achieve a target in For code compilation, linking and debugging adaptation of
tool chains and API’s are necessary. Before implementing the
small UAVs. Stability and accuracy can also be increased by
final design, it is mandatory to analysis the structural,
using different combination of Kalman and PID controller. behavioral and physical model of the system carefully.
Fuzzification method, input variables, output variables, input
IV. UAV CONTROLLER BASED ON FUZZY LOGIC membership functions amount, output membership functions
amount, rules, inferencing method, defuzzification method
In focus to UAV operation, Fuzzy [4,5] Logic offers are the design parameter of controller in fuzzy logic
decision making skill, other task provide such diverse area as approach.
fuzzy image processing, data mining, decision support,
trajectory generation and trajectory tracking. Decision Max defuzzification and center-of-gravity can be
making include path planning, collision avoidance and implemented with a tree of disjunctive elements having the
compulsory landings. To solve unexpected condition such as pipelined fuzzy processor. Crisp having input data
weather condition, faults [10] in telecommunication links and fuzzification method was implemented by Look-up Tables as
UAV subsystem failures etc. Fig.4 represents the fuzzy logic LUT is the fastest element on FPGA board. 16-bit crisp input
mechanism [6,12]. data is implemented by the address of the LUT to get the
degree of membership as membership function. Here, FPGA
Knowledge Base as the processor in Fuzzy Logic Controller. Its on-board clock
frequency is 50 MHz with an inner DLL (delay locked loop)
Data base unit, be altered to a 125 MHz system clock [14].
Fig.4: Fuzzy Logic Mechanism In fig 5 the system architecture include FCS20
hardware and low level software based on neural network and
adaptive flight control system having real time operating
Fuzzification process: fuzzification is a first step in the fuzzy system (RTOS) [8], inter-processor communication,
interfacing where certain real data are converted to a fuzzy navigation sensors and hardware integration. It is also
amounts of membership functions. platform independent so run of this high level software [6].
This hardware optimize the performance to size ratio enabled
Fuzzy Inferencing: the fuzzy rules are applied to decide the by Micro Electro Mechanical Systems (MEMS) sensor and
output fuzzy variables. embedded processing technologies.