Sie sind auf Seite 1von 5

Design and Implementation of

Autopilot System and Ground Station on Autodrive System


Fajar Andy Setyawan1,a , Arief Syaichu Rohman2,b dan Ary Setijadi Prihatmanto3,c
Electrical Engineering, School of Electrical Engineering and Informatics, Institut Teknologi Bandung, Bandung, 40132, Indonesia.
1
2,3
Control and Computer System Laboratory, School of Electrical Engineering and Informatics, Institut Teknologi Bandung, Bandung, 40132,
Indonesia.
Email: setyawan@students.itb.ac.id
Email: arief@stei.itb.ac.id
Email: asetijadi@lskk.ee.itb.ac.id

Abstract – Autodrive system is a system that ease The topic discussed in this document is autopilot
human on transportation. Autodrive system need an subsystem of autodrive system. The desired specification
autopilot module to run its vehicle automatically.On of autodrive system are written on Table 1.
this final project, autopilot module equiped with Table 1 Desired Parameter
IMU, GPS, and telemetry. Desired specification of
autodrive system are tolerance of route of 1 meter, Parameter Desired Explanation
could run on straight route with yaw rate less than
4,52° , and slip angle of 5°.Implementation on
autopilot module are implementation of equation to Yaw during straight 4,52° Regresion result of yaw
route (yaw rate rate tolerance from paper
determine AHRS and steer rate. Besides, parameter tolerance) of Alexander Neukum,
configuration and waypoint configuration for testing. Eric Ufer, Jorn Paulig, and
From the testing, yaw rate for straight route of Hans-Peter Kruger
(1,794±0,385)° , slip angle of 3,92°, and tolerance of “Controllability of
superposition steering
desired route and actual route of system failures”
(2,293392857±2,060191762 ) meter. From the testing
its recommended to increase the accuracy of GPS so Slip angle 5° Sirui Song paper “Towards
the tolerance of desired route and actual route could Autonomous Driving at the
achive below 1 meter. Limit of Friction”[8]
Keywords: Autodrive System, GPS, IMU, Ardupilot Mega
2.6. Accuracy of desired 1 meter Taken from Autonomous
route and actual route Vehicle Policy of Korea.
[9]
I. PRELIMINARY
Traffic accident is something that avoided by human
during transportation activities. According Polantas
dalam Angka Tahun 2013, traffic accident in Indonesia II. DESIGN AND IMPLEMENTATION
88% caused by human factor, alongside with natural
factor, road factor, and vehicle factor. Human factor such A Block Diagram and Flowchart
as careless, exhausted, sleepy, illness, psychological
pressure, inobedient act, medicine factor, alcohol factor, The big picture of autopilot subsystem are explained
and speed limit abuse[3]. Autodrive is a solution to solve with block diagram on Figure 2 and flowchart on Figure
problem caused by human factor 3.
Ground Station
Data Flow Diagram Level 1 of autodrive system are
Autopilot System
shown on Figure 1.
Autodrive Data Link Ground Station
Tranceiver/Receiver Tranceiver/Receiver

Waypoint input
Navigation system Autopilot Controller

Navigation Sensor
Channel Output

Throttle Steering

Figure 2 Block Diagram of Autopilot subsystem


Figure 1 DFD Level 1 of Autodrive System

1
Flight Plan is waypoints data that saved on EEPROM B Attitude and Heading Reference System(AHRS)
of autopilot subsystem. Waypoints data are transmitted
from the ground station with radio telemetry. Next, AHRS is a subsystem on autopilot module to
waypoint data processed on navigation loop on determine attitude(mostly orientation) of autodrive
subsystem navigation and become tracking loop that sent system. Orientation of autodrive system determined by
digital data to steer and throttle on Arduino. accelerometer and magnetometer that calculated from
Autopilot subsystem read waypoint data and than Direction Consine Matrix.
decided the bearing and distance data then doing
Direction Cosine Matrix ( DCM ) used are.[7]
calculation the data for throttle and steer. This function
repeated as long there are waypoint need to pass by
1−2( q22 +q 23) 2(q1 q2−q0 q3 ) 2(q 0 q2 +q 1 q 3)
autodrive vehicle.

start
[
C= 2( q0 q 3+ q1 q2 ) 1−2( q21 +q 23) 2(q 2 q3 +q 0 q 1)
2(q1 q3−q0 q2 ) 2( q0 q1 +q 2 q3 ) 1−2(q 21+ q22 )
From this DCM amount of roll and yaw could be
]
calculated.[4]

−1 C33
Membaca waypoint φ=tan
dari EEPROM C 32
1−2 ( q21 +q 22 )
¿ tan−1 (1)
2 ( q0 q 1+ q2 q3 )

Ada waypoint?
C 21
ψ=tan−1
C 11
Update waypoint 2(q 0 q3 +q 1 q 2)
tujuan ¿ tan−1 2 2 (2)
1−2(q 2+ q3 )
Roll used to observe the orientation change rate on x-
Get bearing axis and yaw used to observe orientation change rate on
& dist z-axis. [6]

Kalkulasi Steer & tidak


Throttle
tidak

Data Digital
Steer dan
Throttle
tidak

Mencapai
Figure 4 Illustration of roll dan yaw
waypoint?

Besides, on AHRS it also calculate groundspeed from


ya
vehicle from GPS with pythagoras equation based on
ground speed vehicle on longitude and latitude.
Waypoint
terakhir? v=√ v x + v y (3)

C Steer and Throttle System


ya Autodrive system have 2 control, which are steer
control and throttle control that its output represented on
11 bit digital data. Autodrive designed doing automatic
Return navigation. Steering could be calculated by using desired
damping (parameter used on implementation of 0,75
Figure 3 Autopilot Subsystem Flowchart
2
based on APM 2.6 standard) to calculate the laterall
start
acceleration (moslty known as sentripetal acceleration).

L 1=4 ∙Ϛ 2 (4)

L 1∙ v 2 Kecepatan
a s= yang
s ∙ sin ⁡( Nu) diinginkan &
(5) nilai P awal

with s variable could be calculated from


Ϛ∙T ∙v
s= (6) Error = Kecepatan
π diinginkan –
Groundspeed kondisi
T is a variable of period for tracking loop that sekarang
configured by user(standard parameter are 20 of APM
2.6). Nu value could be calculated from arctangen
between vehicle velocity perpendicular to the desired
route with the vehicle velocity that alongside with
desired route. Next, lateral acceleration converted into G Pakhir = Pawal *
force value. Then the steer rate could be calculated as Error
follows.[5]
as
steer rate= (7)
v
The steer directio wether its to right or left are Pawal = Pakhir
decided by bearing angle dividing vehicle orientation on
map. If the result more than 1 then steer will rotated to
right, while it the result less than 1 the steer will rotated
to the left. The steer rate and direction o steer will be
converted into 11-bit digital data. This data outputed
from roll channel on CH-1. The impact of steer could be Return
observed from roll rate of vehicle.
Figure 5 P autotuning on APM 2.6
On throttle, speed controled with PID. PID feauture
on APM 2.6 could do automatics tuning with existing
algoritm on APM library. Autotuning of APM with
initial P = 0,7 while I dan D = 0. [2] Automatic tuning could be observed from P value of P
on the graph constantly changed overtime while I and D
value are 0.

Figure 6 Autotuning Graph on APM 2.6

D Hardware implementation and Ground Station


interaction

Hardware implementation on autopilot system by placing


board orientation and vehicle orientation on the same
direction. This also applied on GPS.

3
Parameter Mean Deviation
Standard
CH-11 1385 125,6

Roll(°) -0,009144 0,04442

The steer output influenced by target bearing value.


The output data of give a several amount of critical
value (1100) but the orientation of autodrive system
doesn’t change drastically. This could be shown by
compare steer output and target bearing. Target bearing
is an angle that given between location of vehicle and
waypoint target relatif to North direction.

Figure 7 Board and GPS placement


Figure 11 Comparison between target bearing steer output

The impact of target bearing on steer data are the


output if the bearing angle more than 0° then ideal steeri
output around 1100 to 1499. This factor could be
observed from data comparison of mean target bearing to
the mean steer output .

Table 2 Mean and Deviation Standar of Target Bearing and CH-1


Output
Parameter Mean Deviation Standard
Target Bearing(°) 162,4 25,48
Figure 8 Communication between APM and Ground Station using
Mission Planner software CH-1 1385 125,6
User interact with board by Mission Planner software
that could connect through radio telemetry. Radio
telemetry used board and Ground Station is 915MHz
3DR telemetry.

III. TESTING B. Curve Route Testing


A. Straight Route Testing

Titik start

Figure 9 Yaw rate on straight route Titik finish

Straight route get by setting desired waypoint on the


Titik untuk berbelok
same direction. Yaw rate mean 1,794 ° with deviation
standar 0,385 °. If compared with the desired
Figure 12 Track from curve route
specification, this parameter has been fulfilled.
Yaw rate of this accuracy could be achieven on
On the navigation the desired angle during steering
autopilot system althought 11-bit digital data sent to the
are 93,55°. But the actual route gives the curve angle of
steer are high, but the roll rate of vehicle doesn’t reach
97,47 °. From that it could be calculated that error of the
0,3° for 1 second on curved route). This could explained
slip angle are 3,92°. Compared with the desired
by following Figure.
specification (5°), then the specification has been
fulfilled.

Figure 10 Output of CH-1 and Roll


Figure 13 Target bearing, Roll, and CH-1 output
Table 2 Mean and Deviation Standar of CH-1 and Roll
4
[2] http://wiki.ardupilot.org , accessed on 7 Mei 2017
Turning manuver could be observed from graph on on 10.00 WIB.
Figure 13. Roll rate 0,3° show that the vehicle doing [3] Korlantas Polri(2013).Polantas Dalam Angka
turning manuver. Then the target bearing from the initial 2013. http://korlantas.polri.go.id/polantas-dalam-
point show value under 0° increase over 0° because the angka-2013/. Accessed on 5 September 2016.
waypoint on turning point has reached and change the
mission to the next waypoint. [4] Leo Dorst, Daniel Fontijne, Stephen Mann.(2007)
While target bearing increased, the steer decreased under Geometric Algebra for Computer Science: An
1500. Object-Oriented Approach to Geometry (The
Morgan Kaufmann Series in Computer Graphics).
47-54.
[5] Masato Abe. (2009) Vehicle Handling Dynamics
Theory and Application. Oxford.34-37.
[6] Reza J. Nazar. Theory of Applied Robotics:
Kinematics, Dynamics, and Control (2nd Edition).
New York : Springer.2010.
[7] Sebastian O.H. Madgwick(2010).An efficient
orientation filter for inertial andinertial/magnetic
sensor arrays. 18.
[8] Sirui Song(2014). Towards Autonomous Driving
at the Limit of Friction.45.
Figure 14 Sampling of error of desired route and actual route
[9] South Korea Ministry of Land, Infrastructure, and
Error of desired route and actual route are Transportation.(2016). Autonomous Vehicle
calculated by sampling from the curve route testing Policy of Korea.
and we measure the difference each 3 meter of http://www.ircobi.org/wordpress/downloads/ad-
desired route (56 sample). The mean result session-seoul.pdf. Acceseed ib 11 Mei 2017 on
01.30 WIB.
2,293392857 meter and deviation standard of
2,060191762 meter.

Error of desired route vs actual route


8
6
4
2
0
1 5 9 13 17 21 25 29 33 37 41 45 49 53

Figure 15 Error of desired route vs actual route

IV. CONCLUSION
Autopilot subsistem could do straight route
navigation with yaw rate (1,794±0,385)°,.Then during
curve route the slip angle 3,92°. These result has fulfilled
the desired specification. While the error of desired route
compared to actual route (2,293392857±2,060191762 )
meter. With the result achieved during testing it could be
concluded that navigation system has work but need to
increase the accuracy of desired route and actual route by
increase GPS accuracy.

REFERENCE
[1] Alexandra Neukum, Eric Ufer, Jorn Paulig dan
Hans-Peter Kruger(2008). Controllability of
Superposition Steering System Failures.Cntr of
Traf Sci, University of Wuerzburg. 11.
5

Das könnte Ihnen auch gefallen