st1 st3

IoT Application in Smart Agriculture
Modular, Autonomous Unmanned Aerial Vehicle For Precision Agriculture
MODULAR, AUTONOMOUS UNMANNED AERIAL

VEHICLE FOR PRECISION AGRICULTURE
Authors Akshay G R, Akshay Hiregoudar, Alvin P Cheriyan and Harish P Jain
ABSTRACT: This paper covers the autonomous

trajectory generation of the UAV and identifies Yield
Analysis, Topographical Mapping Techniques,
Assistive Methods by Cloud Based Portals, Modular
Solutions and Swarm UAV exploration.
I.INTRODUCTION
Farmers in India are faced with umpteen
number of problems. There is uncertainty in crop
success rate, excess or under-use of pesticides leads
to quality deterioration, lack of proper resources and Figure 1 Layout of Quad-Rotor. For Complete Motion of Quad-rotor
instruction and external factors like weather, water access: http://bmscemodularuav.weebly.com/basic-motion
deficiency and pest attacks lead to crop damage. We
are agog to tackle this problem by conceptualising a 1. Stop quickly decrease stopping distance (by
Cloud Based Autonomous Unmanned Aerial increasing linear acceleration)
Vehicle (UAV). Inspired by Google Ara smartphone 2. Turn quickly minimize turning radius (by
project, Modular drones empowers farmers to have increasing angular acceleration).
custom-made UAVs which aids in staying up-to-
date with technological advancements. To achieve this
Fixed-wing UAVs though efficient, are 1. for a 2d robot model: taking acceleration
unable to hover and Avian Based Crafts requires along vertical direction and pitch (phi).
perception of complex fluid mechanics and aero-
elasticity. Rotor based UAVs are the most suitable

platform as they have required manoeuvrability and
( )
efficiency. [Ratti. J and Vachtsevanos G, 2011] [1]
II.FORCES AND MOTION ( )
A quad rotor has 4 Degrees of Freedom and

can manoeuvre in 3D Space. Two rotors revolve in
clockwise direction about z axis (Figure 1) and the
other two in anti-clockwise direction.
The short and stubby blades of the quad rotor
dont flap due to gyroscopic moments produced
during yaw. Hence to decrease stopping distance, thrust
(T) to weight (mg) ratio should be increased and to
III.OBSTACLE EVASION decrease turning radius, moment (M) to moment of
Robot has to be very agile to avoid inertia (Ixx) ratio has to be increased.
unexpected obstacles like birds etc. therefore the
goal is to maximize agility:
Akshay G .R akshaygorur1@gmail.com
Akshay Hiregoudar aky.hiregoudar@gmail.com
Alvin P Cheriyan alvin22071996@gmail.com BMS College of Engineering Bangalore
Harish P Jain harishjain41@gmail.com
2. Based on Froude Scaling (For low speed (Simultaneous Localization and Mapping)
applications) of size of quad rotor, [Wolowicz. et al, technique.
1979] [2] linear acceleration remains nearly constant
and angular acceleration varies inversely with SLAM (Figure 3) - UAV at location 1 finds
distance between centre of mass and axis of rotation. out the distance and orientation of fixed obstacles
Hence the quad rotor has to be compact. from it. Then it moves to location 2 where it finds
updated distance and orientation with respect to the
IV.COMPONENT SELECTION obstacles. With this data, UAV calculates the change
in its position and orientation.
With modularity, different types of cameras
(Laser Focused or Beacon based or Thermal or
Figure 2 Motor Selection. Source: Aerial Robotics (Coursera)

Figure 3 SLAM
The hover state is considered as the design

state (fig). For the minimum thrust needed to Colour-Depth or Multi-Spectral), varying accuracy
balance the weight, the speed of the motor is known; of field maps are obtained, with laser based cameras
And the corresponding drag moment acting on the providing highest accuracy.
propeller is also plotted. The motor is chosen such
VI.FORCE AND MOMENT EQUATIONS
that, it produces a torque able to balance the drag
moment. Based on this result, the thrust to weight The input that drives the motor to the desired height,
ratio can also be adjusted. Both the drag moment u: (Here u=acceleration required)
and the thrust are proportional to the square of the
RPM. [William Hanna, 2014] [3] [ ]
Based on the current energy density of Li-Po
The data received from IMU is in Body Fixed Co-
batteries in the market, 25% to 40% of the mass of
ordinates of UAV. For analysis, Earth fixed Co-
the UAV, hence battery selection is an important
ordinate values are required. Co-ordinate
design consideration. Compromises have to be made
Conversion can be achieved in following ways:
on the flight time of UAV.
1. Rotation Matrices
V.STATE ESTIMATION
2. Euler Angles
For a sustained autonomous flight, 3. Axis Angle Representation
continuous data about robots position and 4. Quaternions
orientation has be measured. This involves myriad 5. Exponential Co-ordinates
of sensors and camera unit.
For Force equation, Rotation matrices(R) is
Inertial Measurement Unit (IMU) comprising multiplied
of gyroscope (orientation data) and accelerometer
(specific force data) can be used for SLAM
A. Trajectory Tracking (Assuming a linear

[ ] [ ] model such that are small):
For moment equation, Euler angles are used in the The data from IMU is sent to position
Eulers equation of motion [F. olc*, 2010] [Teppo controller (generates thrust) and attitude controller
Luukkonen] [4] [5] (generates moment for orienting), which drives the
motor controller (controls motors as per output from
[ ] [ ] [ ] [ ] position and attitude controller). Hence the
commanded position is generated by PID Controller.
Where p, q and r are angular velocities and L is the
characteristic length of UAV. .
Using PID controller throughout for all three
VII. PID CONTROLLER directions given i=1, 2, 3, the general form would
be:
Any system, when it has to move from a ( )
state to another, a delay is experienced because of
the transitions that are needed. So, we define an
error function e (t) which should converge , is the commanded thrust sent to motor controller by
exponentially to zero. A general form of the position controller.
equation can be: The pitch ( ) and the roll angle ( ) are also
Here, is the desired acceleration, calculated based on a linear model approximations.
is the derivative gain and is the proportional [5]
gain. Modelling the equation for the disturbances
(including payload and wind), we get a third order [ ]
differential equation. [5]
[ ]

[ ]
Fine tuning the values of these constants can
allow us to make the system springy or damped. , is the commanded moment sent to motor controller
But this equation would be limited by the maximum by attitude controller.
thrust the motors can produce. B. Generating trajectories for an agricultural
[ ] drone with smooth waypoints (Minimum Jerk
Trajectory):
Using the concept of Calculus of Variations
VIII.TRAJECTORY GENERATION to obtain the shortest distance geometry that can
The quad rotor in three dimension has 4 allow us to develop an nth order trajectory. Since we
Degrees of freedom. So, the terms that can be have considered a minimum jerk trajectory, the
explicitly be defined for desired position are system is 3rd order. [Shaojie Shen, 2014] [6] [5]
[ ]. ( )
The other two variables are dependent on
these values.
Where for shortest path,
( ) ( )
The above minimum functions can be solved velocity, acceleration, jerk and snap are same when
by the Euler-Lagrange equation for entering and leaving a waypoint (and zero at initial
explicitly: and final position). Such a method is chosen to
reduce jerk conditions to minimum. The above
( ) ( ) equation is solved using a calculus solver.

[ ( )] IX.SWARM EXPLORATION
The flight time of the UAV ranges from 10
minutes to 20 minutes based on module used or
Solving for n=3, we get, Upon payload mass. Large areas cannot be covered by a
Integrating this, the solution would be: single UAV. Such situations can be tackled by
Swarm Distribution with mutual co-ordination to
Most navigation (especially 2nd order and cover the Areas of Interest (AoI).
higher systems) require smooth trajectories. And this
requires that the quad rotor transits across the The swarm motion planning problem
waypoints without sudden variation in the velocity consists of two main challenges: 1) how to find a
and acceleration. suitable swarm distribution that covers all AoIs, and
2) how to find feasible trajectories of UAVs (the
swarm deployment) from a depot to the final
X Coordinates of Waypoint positions, which are described by the swarm
Navigation distribution while avoiding obstacles in near
3 proximity. The solution also involves the UAVs to
mutually share the data.
2.5
X.MODULARITY
X-Coordinate
2
1.5
1
0.5
0
0 0.5 1 1.5 2
Time
Figure 4 Minimum Jerk Waypoint Navigation
Considering a trajectory as shown in figure 4

with three paths between the intermediate points,
Solving for,
{ }
Considering the terms up to snap to be equal

at the waypoints, which is for smooth motion:
Akshay G .R akshaygorur1@gmail.com Figure 5 Modular UAVs

Farmers in India are faced with umpteen

number of problems. There is uncertainty in crop
success rate, excess or under-use of pesticides leads We generated a code to count oranges using
to quality deterioration, lack of proper resources and Matlab Image Processing Toolbox (figure 6). The
instruction and external factors like weather, water following steps were used:
deficiency and pest attacks lead to crop damage. i. Degenerating blue colour (160 level) to
Modular UAVs serve as solution for the above differentiate bright leaves and oranges.
problems at an economical cost. The farmer can ii. Thresholding R, G and B components up
identify the problems which he/she faces the most to 0.6, 1, 0.1 levels respectively.
and he/she can account for it by choosing one of iii. Producing a binary black & white image.
many configurations possible due to modularity. iv. Filling minor holes.
Upgradation or change in features is easy by snap to v. Identifying and plotting the boundaries
change modules. Same body of UAV can be used
for different tasks by changing components in plug The actual number of oranges in the original
and play fashion easily by farmers. Various image were 36 and detected oranges were 38. Three
configurations covering a wide range of price range false detections were two bright leaves at the lower
are available to farmers. Various configurations have right corner and one bright leaf at upper centre and
been conceptualised as shown in Figure 5 one orange at left central was missed. However, this
technique is still efficient in figuring out rough
XI. IMAGE AND LIVE VIDEO ANALYSIS number of oranges and can be used to segregate
Different modules delivers differing mapping oranges based on size and colour. The Matlab
techniques. Laser Cameras in expensive modules Code that we generated is available at our website.
can be used for High-Res terrain mapping and http://bmscemodularuav.weebly.com
accurate yield analysis. Yield mapping can be done XII.CLOUD BASED SOLUTIONS
spontaneously and high yielding plants can be
identified such that in following harvest cycle, the
crops planted are from seeds of those plants.
Thermal cameras are able to detect cooler,
well-watered field regions as well as dry hot patches.
Farmers can use this data to adjust field irrigation
and avoid wasting excess water or identify weaker
crops and provide them with extra nourishment.
Live Video analysis can help identify
trespassers or pests and alarm the farmers through
SMS or Calls.
Figure 7 Flowchart of Components
Figure 6 Fruit Counting using MATLAB
Figure 7 is a flow process of sharing and We will further work on employing Radon
visualizing data using cloud based platforms. The Transforms, Gamma Correction, morphological
data from the ground controller is uploaded to cloud operations such as anti-dilation in our image
services like Amazon AWS. This data can then processing to improve accuracy of Yield map
be analyzed and interpreted. This data can be shared analysis.
to national servers where nationwide analysis is Regular updates will be posted on
possible without visiting these farms. From these http://bmscemodularuav.weebly.com/
results, its easier for government to roll out schemes
and policies for boosting agricultural revenue.
Farmers can remotely access all details of their farm.
Regular updates on farm conditions and alerts are XIV. ACKNOWLEGEMENTS
instantly notified to farmer even if he/she is not
present at the farm. UAVs can be deployed and 1. Google Ara Modular phone project.
controlled by farmers from any corner of the world. 2. Coursera University of Pennsylvania
Instant service calls in case of malfunction can be Aerial Robotics
given without approaching a 3rd party Service 3. Matlab Image Processing Toolbox
provider. Updates from weather satellites about 4. SOLIDWORKS
incoming storms reach faster than even Television or 5. Amazon AWS
newspapers.
XII. CONCLUSION
XV.REFERENCES
UAV industry is expected to grow to $127
billion by 2020, currently worth $2 billion. The [1]. Ratti J, Vachtsevanos G. Towards energy
Internet of Things (IoT) is emerging as the most efficiency in micro hovering air vehicles IEEE
disruptive technology of early 21st century. Tapping Aerospace Conference Proceedings.
and combining resources from both these industries [2]. C.H. Wolowicz, J.S. Bowman, W.P.
would be revolutionary. Gilbert, Similitude requirements and scaling
Inspired by Google Ara smartphone project, relationships as applied to model testing. Technical
modular UAVs help farmers stay technologically up report, NASA, 1979
to date by updating only the components they wish [3]. William Hanna, Modelling and Control of
to. With evolution of Additive manufacturing an Unmanned Aerial Vehicle A Thesis submitted in
industry, cost of manufacturing using these partial fulfilment of the requirements for the degree
techniques will reduce and can be factory-less by 3- of Bachelor of Engineering in Electrical and
d printers present in Drone-shops in each taluk Electronics Engineering. (2014)
reducing the infrastructure cost drastically. [4].F. olc, Modelling and Control of a
Moreover, elimination of cast and die components Quadrocopter. Advances in Military Technology
will enable easy creation of custom-made UAV parts Vol. 5, No. 2. (2010)
as per liking of farmers. This is not possible in mass [5] Teppo Luukkonen,Modelling and control of
manufacturing factories where possible quad copter. Independent research project in
configurations are limited. applied mathematics.
GPS usage in IMU is under control of US [6] Shaojie Shen, AUTONOMOUS
Military so use of IRNSS (Indian Regional NAVIGATION IN COMPLEX INDOOR AND
Navigation Satellite System) will enable complete OUTDOOR
control by Indian government. ENVIRONMENTS WITH MICRO AERIAL
With a swarm of Super-abled drones, large VEHICLES (2014).
farms can be efficiently managed. [7]. Mellinger, Daniel Warren, "Trajectory
Generation and Control for Quadrotors". Publicly
Accessible Penn Dissertations. Paper 547. (2012).

st1 st3

Загружено:

Сведения о документе

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

st1 st3

Загружено:

Авторское право:

Доступные форматы

IoT Application in Smart Agriculture

Modular, Autonomous Unmanned Aerial Vehicle For Precision Agriculture

MODULAR, AUTONOMOUS UNMANNED AERIAL

ABSTRACT: This paper covers the autonomous

A quad rotor has 4 Degrees of Freedom and

Figure 2 Motor Selection. Source: Aerial Robotics (Coursera)

The hover state is considered as the design

A. Trajectory Tracking (Assuming a linear

Figure 4 Minimum Jerk Waypoint Navigation

Considering a trajectory as shown in figure 4

Considering the terms up to snap to be equal

Akshay G .R akshaygorur1@gmail.com Figure 5 Modular UAVs

Farmers in India are faced with umpteen

Figure 7 Flowchart of Components

Figure 6 Fruit Counting using MATLAB

Вам также может понравиться