Upload
letram
View
217
Download
2
Embed Size (px)
Citation preview
Experimental Study on Geo-Pointing Control System
for Quadcopter
Do yoon Kim1 , Sung Kyung Hong*
1
Future Aerospace Technology Team, Korea Aerospace Research Institute(KARI), Daejeon,
Korea ([email protected])
*Department of Aerospace Engineering, Sejong University, Seoul, Korea
([email protected]), Corresponding Author
Abstract. The This paper presents a novel Geo-Pointing algorithm consists of
three phases; 1) estimating coordinates with line of sight(LOS) from the
quadcopter to the target objects, and then 2) tracking the stationary objects for
orienting the line of sight toward the object (Point of Interest, POI), and 3) the
quadcopter automatically around an object (Circle of Interest, COI). With
application of these functions that enables the desired objects always on screen,
all that is required is to fly above the object and mark it in flight via Ground
Control System (GCS). The performance and its usefulness are verified through
flight experiments.
Keywords: Quadcopter, Camera Line of Sight (LOS), Geo-Pointing, Geo-
Location, Circle of Interest (COI), Point of Interest (POI),
1 Introduction
Following the appearance of inertial measurement units with MEMS technologies and
high efficiency small batteries included along with the striking development of
avionics-related technologies for unmanned aerial vehicles (UAV), UAVs have been
rapidly miniaturized and popularized. Among them, quadcopter platforms have
recently been in the limelight, because they are structurally simple yet highly
stable.[1-3] In the present study, geo-pointing algorithms that combine a quadcopter
with a gimbal system to link missions such as reconnaissance, surveillance, and
searches with autonomous flight are proposed. The proposed algorithm consists of
three stages: 1) estimating the target coordinates through the line of sight (LOS)
pointing angle from the quadcopter to the target object(geo-location); 2) tracking the
stationary target by updating that the LOS pointing angle always points at the
target(POI); and 3) perform automatic flight along the circle of interest (COI)
centered on the target. That is, when operating a quadcopter, the controller is not
required to control camera firsthand to obtain continuous and stereoscopic image
information for the designated target but is just required to set the target
Advanced Science and Technology Letters Vol.118 (Electrical and Electronic Engineering 2015), pp.14-20
http://dx.doi.org/10.14257/astl.2015.118.04
ISSN: 2287-1233 ASTL Copyright © 2015 SERSC
Fig. 1. Configuration of Experimental Setup
2 Configuration of flight system
For experimental verification of the quadcopter geo-pointing algorithm, not only a
quadcopter platform but also a controller to drive a gimbal for pointing at the POI and
linking the gimbal with autonomous flight is necessary. A conceptual diagram of the
entire system configured as such is shown in Figure 1, and the details of individual
components are as follows.
2.1 Quadcopter platform
The quadcopter platform used in the present study was a Pelican model from
Ascending Technology Co. in Germany, made for research [4]. The main processor
for flight control was composed of two processors, a low-level processor (LLP) and a
high-level processor (HLP). Meanwhile, to design the outer control loops (geo-
pointing control in the present study) to be installed on the HLP, a mathematical
model that integrates the platform as such and the inner control loops of the LLP is
required. The inner control loops of individual axes are composed of roll ( ), pitch
( ), and yaw angular velocity ( ) control loops. To identify the mathematics for these
control loops, a linear black box model was assumed, and a prediction error method
algorithm [5] was used.
2.2 Camera gimbal system
The camera gimbal installed on the quadcopter was a direct BLDC motor-driven
gimbal with two small axes (roll/pitch angles), installed with a. It is embedded with
its own IMU to implement a basic image-stabilizing algorithm that compensates for
flight vehicles’ dynamic states, and it is driven by roll and pitch angle commands
from the HLP of the quadcopter, installed with a geo-pointing algorithm to move the
LOS vector to target
Advanced Science and Technology Letters Vol.118 (Electrical and Electronic Engineering 2015)
Copyright © 2015 SERSC 15
Figure. 2. Relative vector of target with respect to quadcopter in ENU frame coordinates
3 Geo-Pointing control system of Quadcopter
3.1 Estimation of the target coordinate(geo-location)
The estimation of the target coordinate (geo-location) corresponds to the first stage
in the geo-pointing system. As can be seen in Figure 2, the target position vector ( ) in the ENU coordinate system can be defined as in equation (1) using the quadcopter
position vector ( ) received from the GPS and its relative vector ( )
= + (1)
Here, the relative vector ( ) is produced as follows through the geometric
relationship between the quadcopter height ( ) and the LOS pointing angle
(quadcopter yaw angle : , gimbal pitch angle: ) toward the target
Here, the relative distance (R) is defined as =
3.2 LOS pointing angle control for POI mode
The POI mode is aimed to continuously point at the fixed target [equation (2)]
extracted independently from the movement of the flight vehicle, as described in
section 3.1 of this chapter. To this end, the LOS pointing angles (quadcopter yaw
angle : , gimbal pitch angle: ) intended to position the LOS vector point at the
fixed target based on the moving quadcopter positions are automatically generated
according to the sampling period, and the system is controlled to follow these LOS
pointing angles
= [ 𝑠𝑖𝑛 𝑐𝑜𝑠 ] (2)
Advanced Science and Technology Letters Vol.116 (Electrical and Electronic Engineering 2015)
16 Copyright © 2015 SERSC
1) LOS pointing angle update
To update the LOS pointing angles to those that compensate for quadcopter
maneuvering, the relative position coordinate ( 𝑖) of the fixed target ( ), whose
origin point is the changed position ( 𝑖) of the quadcopter obtained from GPS
information according to the sampling period, should be renewed first. That is, the
coordinate ( ) of the target obtained at the time of lock-on, based on equation (1), is
fixed as . At this time, the new relative position coordinate ( 𝑖) is updated
through equation (3).
𝑖 = − 𝑖 (3)
Through the component ( 𝐸𝑖 , 𝑁𝑖 , 𝑈𝑖) of the renewed relative position vector, the
LOS pointing angle is renewed to the compensated LOS pointing angle (quadcopter
yaw angle : , gimbal pitch angle: ) as follows.
= 𝑎𝑡𝑎𝑛2 𝐸𝑖 , 𝑁𝑖 , = 𝑎𝑡𝑎𝑛2 𝑈𝑖 , 𝑅𝑖
(4)
Where, 𝑅𝑖 = √ 𝐸𝑖2 + 𝑁𝑖
2 .
2) LOS pointing angle control loop
Since the direction axes of the Pelican platform from Ascending Technology Co. used
in the present study are composed only of the LLP angular velocity control loops it
was necessary to design an yaw angle control loop to follow the quadcopter yaw
angle ( ) of the LOS pointing angles updated, as mentioned above. A proportion-
differential (PD) controller that used the yaw angle angular velocity control loop
derived earlier as an inner loop was implemented and installed on the HLP,
3.3 COI control mode
The circle of interest control mode refers to the automatic flight along a circular
trajectory around the target with a certain radius ( ), maintaining a certain speed ( )
In this case, the LOS pointing angle control for the POI mode described in section 3.2
needed to be conducted in combination. The controller maintaining the radius (R) of
the circular trajectory was configured as multiple loops that created pitch commands
( ) using the X axis (forward) velocity ) control loop as an inner loop, and the
circular trajectory circling speed was set to be controlled by roll commands ( )
generated through the speed ( ) control loop in the Y axis direction on the body frame
3.4 Simulation
In the present study, simulations of the quadcopter geo-pointing control system
presented above were implemented. The noise characteristics of various sensors
actually used and GPS error characteristics were measured and added to the
Advanced Science and Technology Letters Vol.118 (Electrical and Electronic Engineering 2015)
Copyright © 2015 SERSC 17
simulation environment. The performance of geo-location and of the POI control loop
was verified together by implementing the POI mode, reverse estimating the position
pointed to by the LOS pointing angle ( , ) generated through the foregoing, and
comparing the initially set target and the estimated target position. (Figure 3)
Fig. 3. Geo-location Simulation Fig. 4. COI mode Simulation
4 Flight Test
4.1 POI mode test
To verify the LOS pointing angle control for the POI, flight tests were conducted
under the following scenario. First, camera images obtained during flight were
monitored in real time on the ground while locking on a certain target, and the
position and height of the quadcopter were manually changed arbitrarily
Figure. 5. POI mode Flight Test Figure. 6. Geo-location Flight Test
4.2 COI mode Test
The flight scenario used to verify the COI mode is as follows. First, a certain POI was
set and the POI mode was implemented first to maintain the LOS pointing angle.
Thereafter, the radius ( ) and speed ( ) input values at the center of the POI were
-20 -15 -10 -5 0 5 10-20
-15
-10
-5
0
5
10
COI(Circle of interest) Test
E (meter)
N (
mete
r)
Quadcopter Path
Target
Advanced Science and Technology Letters Vol.116 (Electrical and Electronic Engineering 2015)
18 Copyright © 2015 SERSC
applied to implement the COI mode. The initial radius ( ) and speed ( ) were 10 m
and 0 m/s, respectively, and the COI mode was applied at around 15 sec to follow
commands.
Fig. 7. Radius Response in COI mode
Fig. 8. Velocity Response in COI mode Fig. 9. Flight Path in COI mode
5 Conclusion
The present study proposed geo-pointing algorithms to estimate target position (geo-
location) and tracking the target coordinate(POI) through LOS pointing anglethat
point at the target from the quadcopter and perform automatic flight along the
circle(COI). To design controllers for the proposed algorithms, modeling was
conducted based on flight data, and model parameters were estimated through the
prediction error method. Based on the model, the effects of diverse disturbance
conditions, including position error elements on the designed geo-pointing algorithms,
were analyzed. The usefulness of the algorithms was demonstrated through flight tests
applied to actual hardware, and performance indicators were derived.
References
1. Luukkonen, Teppo. "Modelling and control of Quadcopter." Independent research project
in applied mathematics, Espoo, 2011
Advanced Science and Technology Letters Vol.118 (Electrical and Electronic Engineering 2015)
Copyright © 2015 SERSC 19
2. Bouabdallah, Samir, Pierpaolo Murrieri, and Roland Siegwart. "Design and control of an
indoor micro quadrotor." Robotics and Automation, 2004. Proceedings. ICRA'04. 2004
IEEE International Conference on. Vol. 5. IEEE, 2004.
3. Gurdan, D., Stumpf, J., Achtelik, M.,Doth, K.M., Hirzinger, G. , Rus, D.,“Energy-efficient
autonomous four-rotor flying robot controlled at 1 kHz” , IEEE International Conference
on Robotics and Automation (ICRA), pp. 361–.366. Roma, Italy (2007)
4. AscTec Hummingbird with AutoPilot User’s Manual, Ascending Technologies GmbH
5. Abed-Meraim, Karim, Eric Moulines, and Philippe Loubaton. "Prediction error method for
second-order blind identification." Signal Processing, IEEE Transactions on 45.3 (1997):
694-705.
Advanced Science and Technology Letters Vol.116 (Electrical and Electronic Engineering 2015)
20 Copyright © 2015 SERSC