Chapter 2

Equations of Motion for an Unmanned Aerial

Vehicle

2.1 Rigid Body Equations of Motion

Before going into deeper analyses of the equations of motion, it will be useful to
review the axis systems that will be dealt with in further steps.

2.1.1 Coordinate Systems

2.1.1.1 Aircraft Body Axis System

The origin of the frame is located at the center of gravity of the aircraft. The x-axis
is directed toward the nose of the plane, while the y-axis is defined as coming out
of its right wing. The z-axis completes the system by the right-hand rule as it points
down through the bottom of the aircraft.

2.1.1.2 The Earth Axis System

The z-axis of the system shows the center of the Earth. The system lies on the
local horizontal plane of the Earth, as the orthogonal x- and y-axes are defined in
the directions of north and east, respectively. As a general manner, the Earth axis
system is assumed to be the inertial axis for aircraft-related studies. Since aircraft
rotation rates are relatively high in comparison to the rotation rate of the Earth, this
assumption gives acceptably low error for aircraft problems and can be utilized [1,
2] (Fig. 2.1).

© Springer International Publishing Switzerland 2015 9

C. Hajiyev et al., State Estimation and Control for Low-cost Unmanned Aerial
Vehicles, DOI 10.1007/978-3-319-16417-5_2
10 2 Equations of Motion for a UAV

Fig. 2.1 Body and inertial axis system

2.1.2 Derivation of Rigid Body Equations of Motion

Under the assumptions that the aircraft is a rigid body and its mass remains constant
for a relatively short duration of time, Newton’s second law can be written as:
 !
!
!
 dV c
F Dm (2.1)
dt
i

where m is the mass of the aircraft, Vc is the velocity of the center of mass of
!
the airplane, F is the force acting on the plane, and subscript i indicates that the
derivation is taken with respect to the inertial frame.
If the equation is rewritten in order to fix the axis system to the aircraft body
frame, rotation of the plane with respect to the inertial frame must be taken into
!

consideration. The derivative of Vc , referred to a rotating body frame that has an
angular velocity of !

! b , can be shown by:
 !
!  !
! !

C!

dV c dV c
D ! b x Vc (2.2)
dt dt b
i b

!
Hence, F , the force that acts on the aircraft, becomes:
"  !
! #
!
 
!
F Dm
dV c !
C ! b x Vc (2.3)
dt b
b
2.1 Rigid Body Equations of Motion 11

!
where Vc , the velocity of the aircraft in the body frame, is formed by u, v, and w,
b
velocities in the x, y, and z directions, respectively, as:
!
 !
 !
 !

Vc D u i C v j C w k (2.4)
b

and !
! b , the angular velocity vector in the body frame, consists of p, r, and q, angular
rates about the x, y, and z axes (or roll, pitch, and yaw rates), respectively, as:

!
 !
 !
 !

!b D p i Cr j Cq k (2.5)

As a result,the three force equations for the aircraft in the body axis may be given
as:
2 3 2 3 ˇ ˇ 2 3
uP ˇ! !
 !
 ˇ
uP C qw  rv
Fx ˇ i j k ˇ
4 Fy 5 D m 4 vP 5 C m ˇˇ p q r ˇˇ D m 4 vP C ru  pw 5 (2.6a)
ˇ ˇ
Fz wP ˇ u v wˇ wP C pv  qu

Fx D m .Pu C qw  rv/
Fy D m .vP C ru  pw/ (2.6b)
Fz D m .wP C pv  qu/

On the other hand, in order to derive the three moment equations of the aircraft,
a more complicated methodology may be followed based on examining an arbitrary
small elemental mass, ım, which is at a distance from the center of mass of the
aircraft and rotates about it [2] (Fig. 2.2).

Fig. 2.2 Arbitrary small x

differential mass in the body
frame

r δm

δm

z
12 2 Equations of Motion for a UAV

First, the distance between the differential mass and the center of mass of the
aircraft is given by the following vector:

r D x!
!  !
 !

i Cy j Czk (2.7)
ım

Here, x, y, and z are the distances in the x, y, and z axes. Now, the velocity of that
differential mass with respect to the center of mass should be written as:
!
!
 d!
r
C!

!b x!
r
ım
V ım D ım (2.8)
dt
b

Since the aircraft is assumed to be rigid, !

r is constant in time and, so:
ım

!
d!
r
ım
D0 (2.9)
dt
b

Therefore:
ˇ ˇ
ˇ! !
 !
 ˇ
!
 ˇ i j k ˇ
ˇ ˇ
V ım D !

!b x!
r
ım Dˇ p q rˇ (2.10a)
ˇ ˇ
ˇ x y zˇ
!
 !
 !
 !

) V ım D .qz  ry/ i C .rx  pz/ j C .py  qx/ k (2.10b)

Furthermore, an expression for the angular momentum of differential mass,

!

ı H ım , can be derived as:
!
   
r x ım!
ı H ım D ! V ım (2.11)
ım

Hence, the three terms of the angular momentum can be obtained as:
2 3 ˇˇ !
 !
 !
 ˇ
ˇ
ıHx ˇ i j k ˇ
!
 ˇ ˇ
ı H ım D 4 ıHy 5 D ˇ x y z ˇ (2.12a)
ˇ ˇ
ıHz ˇ ım .qz  ry/ ım .rx  pz/ ım .py  qx/ ˇ
 
ıHx D p y2 C z2  ım  qxy ım  rxz ım
) ıHy D q x2 C z2 ım  ryz ım  pxy ım (2.12b)
 
ıHz D r x2 C y2 ım  pxz ım  qyz ım

After that, the expressions should be integrated over the whole aircraft to
determine the angular momentum equations of plane itself. As long as p, r, and
q are not dependent on mass, then:
2.1 Rigid Body Equations of Motion 13

• • • •
 2 2

Hx D ıHx D p y C z ım  q xy ım  r xz ım
• • • •
 2 
Hy D ıHy D q x C z2 ım  r yz ım  p xy ım (2.13)
• • • •
 2 2

Hz D ıHz D r x C y ım  p xz ım  q yz ım

It is known that moments of inertia are defined as:

•
 2 
Ixx D y C z2 ım (2.14a)

•
 
Iyy D x2 C z2 ım (2.14b)

•
 
Izz D x2 C y2 ım (2.14c)

and the products of inertia are:

•
Ixy D xyım (2.15a)

•
Iyz D yzım (2.15b)

•
Ixz D xzım (2.15c)

Consequently, if these moments of inertia and products of inertia terms are

substituted into Eq. (2.13), then:

Hx D pIxx  qIxy  rIxz

Hy D qIyy  rIyz  pIxy (2.16a)
Hz D rIzz  pIxz  qIyz

or in matrix notation:
2 3 2 32 3
Hx Ixx Ixy Ixz p
4 Hy 5 D 4  Ixy Iyy Iyz 5 4 q 5 (2.16b)
Hz  Ixz Iyz Izz r
14 2 Equations of Motion for a UAV

In general, the x–z plane of the aircraft is assumed to be a plane of symmetry and,
so:

Ixy D Iyz D 0 (2.17)

This simplifies the two parts of Eq. (2.16) and:

Hx D pIxx  rIxz
Hy D qIyy (2.18)
Hz D rIzz  pIxz

As the final stage, the time derivative of the angular momentum vector should
!
be determined. In a similar way to Eq. (2.2), the time derivative of H , referred to a
rotating body frame that has an angular velocity of !
! b , can be expressed as:
!
! !
!
!

C!

dH dH
D !b x Hb (2.19)
dt dt
i b

Since it is assumed that the mass and the mass distribution of the plane are
constant, the time moments of inertias and the products of inertias do not vary with
time:

IPxx D IPyy D IPzz D IPxz D 0 (2.20)

Apart from that point:

2 3
!
! pP Ixx  rP Ixz
dH
D4 qP Iyy 5 (2.21)
dt
b rP Izz  pP Ixz

and it is obvious that:

ˇ ˇ
ˇ !
 !
  ˇ 2
! 3
ˇ i j k ˇ q .rIzz  pIxz /  rqIyy
!
 !
 ˇ ˇ
!b x Hb D ˇ p q r ˇ D 4 r .pIxx  rIxz /  p .rIzz  pIxz / 5
ˇ ˇ
ˇ pIxx  rIxz qIyy rIzz  pIxz ˇ pqIyy  q .pIxx  rIxz /
(2.22)

These yields:
2   3
!
! !
! pP Ixx C qr Izz  Iyy .Pr C pq/ Ixz
!

C!

dH dH
D ! b x H b D 4 qP Iyy  pr .Izz  Ixx / C p2  r2 Ixz 5
dt dt  
i b rP Izz C pq Iyy  Ixx C .qr C pP / Ixz
(2.23)
2.2 Orientation and Position of an Aircraft 15

As a result, the three moment equations of the motion of the aircraft in the body
axis system can be written as:
 
L D pP Ixx C qr Izz  Iyy  .Pr C pq/ Ixz
2 2
 zz  Ixx /C p  r Ixz
M D qP Iyy  pr .I (2.24)
N D rP Izz C pq Iyy  Ixx C .qr C pP / Ixz

Here, L is the moment around the x-axis, M is the moment around the y–axis, and
N is the moment around the z-axis. Also, it is good to emphasize that the first terms
  
 pIxx ; qIyy ; rIzz , are related to the angular
of all three equation,  acceleration, while the
second terms, qr Izz  Iyy ; pr.Izz  Ixx / ; pq Iyy Ixx , denote
 gyro precision, and
  2 2
 
the third expressions, r C pq Ixz ; p  r Ixz ; qr C p Ixz , are coupling terms.
Therefore, in summary, the three force and three moment equations for an aircraft
are:

Fx D m .Pu C qw  rv/
Force Eq: )
Fy D m .vP C ru  pw/
Fz D m .wP C pv  qu/
 
L D pP Ixx C qr Izz  Iyy  .Pr C pq/ Ixz
Moment Eq: ) M D qP Iyy  pr .Izz  Ixx / C p2  r2 Ixz
 
N D rP Izz C pq Iyy  Ixx C .qr C pP / Ixz

2.2 Orientation and Position of an Aircraft

The equations of motion of an aircraft examined thus far are derived for the body
frame of the plane. However, the orientation and position of the aircraft cannot be
defined in a relatively moving frame and, instead, a fixed frame such as the Earth
axis system must be used.
Hence, it is necessary to introduce some methodology to relate these two axis
systems: the fixed inertial frame and the aircraft body frame. To achieve this, the
orientation of the aircraft can be described by three consecutive rotations, and these
angular rotations are called Euler angles [1], a concept developed by Leonhard Euler
to define the orientation of a rigid body in 3D Euclidean space.
For such a problem, the order of rotations is important and the 3-2-1 Euler angle
representation is used. That means:
• First, rotate about the z-axis through the yaw angle
• Then, as the second step, rotate about the y-axis through the pitch angle 
• And, finally, rotate about the z-axis through the roll angle of 
As a consequence, it is possible to associate flight velocity components relative
to the fixed reference frame and the velocity components in the body frame:
16 2 Equations of Motion for a UAV

Fig. 2.3 Euler angles, velocity components, and angular velocities for an unmanned aerial vehicle
(UAV)

2 dx 3 2 32 3
dt
cc ssc  cs csc  ss u
4 dy 5 D 4 c s sss C cc css  sc 5 4 v 5 (2.25)
dt
dz
dt
 s sc cc w

Here, u, v, and w are velocity components in the body frame, dx , dy , and dz

dt dt dt
are
velocity components in the fixed inertial frame, and  , , and  are Euler angles.
c() and s() are cosine and sine functions, respectively.
Per contra, by the use of Euler angles, angular rates (or angular velocities) of p,
q, and r may be related with the Euler rates of ,P P , and P (Fig. 2.3):
2 3 2 32 3
p 1 0 s P
4 q 5 D 4 0 c c s 5 4 P 5 (2.26a)
r 0 s cc P

or, inversely:
2 3 2 32 3
P 1 s tan  c tan  p
4 P 5 D 4 0 c s 5 4 q 5 (2.26b)
P 0 s sec  c sec  r

2.3 Small Perturbation Theory

Small perturbation theory is a technique for linearizing nonlinear equations. By

performing this mathematical method, it is possible to find approximate solutions
to problems that cannot be solved exactly.
2.3 Small Perturbation Theory 17

As long as the six equations of motion of the plane, as derived in the previous
section, are nonlinear and must be linearized for solving behavior problems of
aircraft, small perturbation theory must be used to progress. First of all, in order
to apply the method, it is assumed that the motion of the aircraft consists of small
perturbations about the steady flight condition [1]. Then, as a base for the theory,
small perturbation terms are added to the steady states:

u D u0 C u p D p0 C p ™ D ™0 C ™
v D v0 C v q D q0 C q  D 0 C  (2.27)
w D w0 C w r D r0 C r D 0 C 

If the reference flight condition is assumed to be symmetric, the propulsive forces

are assumed to remain constant and the x-axis is considered, as it coincides with the
direction of the aircraft’s velocity vector, then the perturbed states become:

u D u0 C u p D p ™ D ™0 C ™
v D v q D q  D  (2.28)
w D w r D r D

Here, u, v, and w are velocity components in the x, y, and z directions, p, q, and r

are angular velocities about the x, y, and z axes, and ™, , and are the pitch, roll,
and yaw angles. The subscript 0 indicates the steady states, where  is used for
perturbations.
Now, let’s handle the force equation in the x-direction as an example:

Fx D X D m .Pu C qw  rv/ (2.29)

Replace all of the states with perturbed ones:

X0 C X D m .Pu C qw  rv/ (2.30)

At level flight conditions, there is no acceleration and, so, X0 D 0. On the other

hand, if order of .q/ ; .w/ ; .r/ ; .v/ D ", neglect terms with the order of "2 .
Hence:

X D m .Pu/ (2.31)

If the same method is followed for all six equations of motion, linearized
equations may be obtained as follows:
2 3 2 3
X Pu
14
Y 5 D 4 vP  ru0 5 (2.32)
m
Z wP  qu0
18 2 Equations of Motion for a UAV

2 3 2 3
L Ixx pP C Izz rP
4 M 5 D 4 Iyy qP 5 (2.33)
N Izz rP C Ixz pP

Besides, X can be expressed by means of a Taylor series in terms of the

perturbation variables as:

@X @X @X @X
X D u C w C ıe C ıT (2.34)
@u @w @ıe @ıT

where ı e , ı T are control-related terms and stand for the elevator deflection and
change in thrust, respectively. Then, these two equations can be related as follows:

@X @X @X @X
m .Pu/ D u C w C ıe C ıT (2.35)
@u @w @ıe @ıT

If the same kind of association is built for all six linearized equations and, after
that, necessary simplifications are realized under the light of assumptions, the result
will be the longitudinal and lateral equations of motions for the aircraft, which is
also a consequence of small perturbation theory. To learn more about the technique,
[1] and [2] are good references.

2.4 Linearized Equations of Motion

In this final part of the chapter, first, linearized equations of motion of an airplane
are given in state space form and, then, by taking the specifications of the Zagi
unmanned aerial vehicle (UAV) [3] as a guideline, the equations are rewritten. Zagi
is a flying wing type radio-controlled aircraft that is popular amongst hobbyists.
At this point, it might be useful to indicate that we mainly take references [3–
5] as an essence to obtain scalar quantities of stability derivatives and, therefore,
stability coefficients. Hence, the values are accepted as true and used as they are in
general. Some small modifications are applied, like changing the initial conditions,
if necessary.

2.4.1 Equations in General

Linearized longitudinal equations of motion of the plane can be written in state

space form as:
2.4 Linearized Equations of Motion 19

2 3 2 32 3
Pu Xu Xw 0 g u
6 wP 7 6 07 6 7
6 7 6 Zu Zw u0 7 6 w 7
4 Pq 5 D 4 Mu C MwP Zu Mw C MwP Zw Mq C Mw u0 0 5 4 q 5
P 0 0 1 0 
2 3
Xıe XıT
 
6 Z ZıT 7 ıe
C4 6 ıe 7 (2.36a)
Mıe C MwP Zıe MıT C MwP ZıT 5 ıT
0 0

Following this, lateral equations of motion can be given as:

2 3 2 32 3 2 3
vP Yv Yp  .u0  Yr / g cos .0 / v 0 Yır
 
6 Pp 7 6 Lv 0 7 6 p 7 6 Lıa Lır 7
6 7 6 Lp Lr 76 7C6 7 ıa
4 Pr 5 D 4 Nv Np Nr 0 5 4 r 5 4 Nıa Nır 5 ır
P 0 1 0 0  0 0
(2.36b)

Here, ı e , ı a , and ı r are elevator, aileron, and rudder deflections, respectively,

ı T is the change in thrust, and Xu , Zu : : : Zıe : : : Yv , Lv : : : Lıa : : : are stability
derivatives.

2.4.2 Characteristics of the Chosen Zagi UAV

As an experimental platform, the Zagi UAV is chosen and, in the remainder

of the book, the proposed methods are demonstrated using the dynamics and
characteristics of this UAV (Fig. 2.4).

Fig. 2.4 The Zagi UAV used by the students of Brigham Young University [4]
20 2 Equations of Motion for a UAV

Table 2.1 Specifications of m 1.56 kg

the Zagi unmanned aerial
S 0.2589 m2
vehicle (UAV) [3]
b 1.4224 m
c 0.3302
0 m 1
0:1147 0 0:0015
B C
JD@ 0 0:0576 0 A kg:m2
 0:0015 0 0:1712

Table 2.2 Stability Longitudinal coefficients Lateral coefficients

coefficients of the Zagi UAV
[3] CL0 D 0:28 CY70 D 0
CD0 D 0:03 Cl0 D 0
Cm0 D 0 Cn0 D 0
CL˛ D 3:45 CYˇ D 0:98
CD˛ D 0:30 Clˇ D 0:12
Cm˛ D 0:38 Cnˇ D 0:25
CLq D 0 CYp D 0
CDq D 0 Clp D 0:26
Cmq D 3:6 Cnp D 0:022
CLıe D 0:36 CYr D 0
CDıe D 0 Clr D 0:14
Cmıe D 0:5 Cnr D 0:35
CYıa D 0
Clıa D 0:08
Cnıa D 0:06
CYır D 0:17
Clır D 0:105
Cnır D 0:032

In Table 2.1, m is the mass, S is the wing area, b is the span length, c is the mean
chord length, and J is the inertia matrix of the plane. Notwithstanding, the stability
coefficients of the Zagi UAV, which are used in order to obtain stability derivatives,
may be tabulated as in Table 2.1.

2.4.3 Linearized Equations of Motion for the Zagi UAV

In this last step, stability derivatives calculated for the Zagi UAV, in the light of
Tables 2.1 and 2.2 and related equations from [5], are substituted in the linearized
lateral and longitudinal equations of motion. However, before that process, the
2.4 Linearized Equations of Motion 21

equations are modified in such a way that observing the height of the UAV is also
possible and the sideslip angle is used instead of the starboard velocity v:
2 3 2 32 3
Pu Xu Xw 0 g 0 u
6 w 7 6 0 07 6 7
6 7 6 Z u Z w u0 7 6 w 7
6 7 6 76 7
6 Pq 7 D 6 Mu C MwP Zu Mw C MwP Zw Mq C Mw u0 0 0 7 6 q 7
6 P7 6 76 7
4  5 4 0 0 1 0 0 5 4  5
hP 0 1 0 u0 0 h

2 3
Xıe XıT
6 ZıT 7
6 Zıe 7 
6 7 ıe
C 6 Mıe C MwP Zıe MıT C MwP ZıT 7
6 7 ıT
4 0 0 5
0 0
(2.37a)
2 3 2 3
2 3 Yˇ Yp u0  Yr g cos .0 / 2 3 Yır
ˇP 6 u0  7 ˇ 6
0
6 Pp 7 6 u0 u0 u0 7 6 w 7 6 u0 7 
7 ıa

6 7 6L 0 7 6 7 C 6 Lıa Lır 7
4 Pr 5 D 6 ˇ
Lp Lr 74 5 6 7
4 Nˇ Np Nr 0 5 q 4 Nıa Nır 5 ır
P 0 1 0 0  0 0
(2.37b)

Here, h is the height of the plane and ˇ is the sideslip angle.

Therefore, as the calculated quantities are substituted into those equations,
specialized equations for the Zagi UAV will become:
2 3 2 32 3
Pu 0:3356 1:3181 0 9:80665 0 u
6 wP 7 6  1:7916 3:9003 9:8215 0 07 6 7
6 7 6 7 6 w 7
6 7 6 76 7
6 Pq 7 D 6 0:702 3:5375 11:392 0 0 7 6 q 7
6 P7 6 76 7
4  5 4 0 0 1 0 0 5 4  5
hP 0 1 0 9:8215 0 h

2 3
0:7436 6:8728
6 3:7855 0 7
6 7 
6 7 ıe
C 6 47:917 0 7
6 7 ıT
4 0 0 5
0 0
(2.38a)
22 2 Equations of Motion for a UAV

2 3 2 32 3
ˇP 0:1069 0:1962 1 0:9984 ˇ
6 Pp 7 6  1:2213 1:9155 1:0096 0 7 6 7
6 7 6 7 6 w 7
4 Pr 5 D 4 1:7255 0:0919 1:7198 0 5 4 q 5
P 0 1 0 0 
2 3
0 0:1855
 
6 8:348 0 7
C6 7 ıa (2.38b)
4 4:24 2:1272 5 ır
0 0

Note that it is also possible to show the equations in a joined way, as follows:
2 3 2 3
Pu 0:3356 1:3181 0 9:80665 0 0 0 0 0
6 wP 7 6 0 7
6 7 6  1:7916 3:9003 9:8215 0 0 0 0 0 7
6 Pq 7 6 0:702 3:5375 11:392 0 0 0 0 0 0 7
6 7 6 7
6 7 6 7
6 P 7 6 0 0 1 0 0 0 0 0 0 7
6 7 6 7
6 hP 7 D 6 0 1 0 9:8215 0 0 0 0 0 7
6 7 6 7
6
6 ˇP 7
7 6
6 0 0 0 0 0 0:1069 0:1962 1 0:9984 7
7
6 Pp 7 6 0 0 0 0 0 1:2213 1:9155 1:0096 0 7
6 7 6 7
4 Pr 5 4 0 0 0 0 0 1:7255 0:0919 1:7198 0 5
P 0 0 0 0 0 0 1 0 0
2 3
0:7436 6:8728 0 0
6 3:7855 0 0 0 7
6 7
6 47:917 0 7
6 0 0 7 2 ı 3
6 7 e
6 0 0 0 0 76
6 7 6 ıT 7
7
C6 0 0 0 0 74
6 7 ıa 5
6 0 0 0 0:1855 7
6 7 ır
6 0 0 8:348 0 7
6 7
4 0 0 4:24 2:1272 5
0 0 0 0
(2.39)

By regarding the general representation of the state space form as:


X D AX C BU; (2.40)

where A is the system matrix, B is the control distribution matrix, U is the control
vector, and X is the state space vector, one can write the following:
References 23

2 3
0:3356 1:3181 0 9:80665 0 0 0 0 0
6  1:7916 3:9003 9:8215 0 0 0 0 0 0 7
6 0:702 3:5375 11:392 0 0 0 0 0 0 7
6 7
6 0 0 1 0 0 0 0 0 0 7
6 7
AD6
6 0 1 0 9:8215 0 0 0 0 0 77
6 0 0 0 0 0 0:1069 0:1962 1 0:9984 7
6 7
6 0 0 0 0 0 1:2213 1:9155 1:0096 0 7
4 5
0 0 0 0 0 1:7255 0:0919 1:7198 0
0 0 0 0 0 0 1 0 0
(2.41a)
2 3
0:7436 6:8728 0 0
6 3:7855 0 0 0 7
6 7
6 47:917 0 0 0 7
6 7
6 0 0 0 0 7
6 7
6 7
BD6 0 0 0 0 7 (2.41b)
6 7
6 0 0 0 0:1855 7
6 7
6 0 0 8:348 0 7
6 7
4 0 0 4:24 2:1272 5
0 0 0 0
 T
U D ıe ıT ıa ır (2.41c)

 T
X D u w q  h ˇ p r  (2.41d)

References

1. Nelson RC (1998) Flight stability and automatic control, 2nd edn. WCB/McGraw-Hill, Boston
2. Yechout TR (2003) Introduction to aircraft flight mechanics. AIAA Education Series, Reston
3. Matthews JS (2006) Adaptive control of micro air vehicles. M.Sc. thesis, Department of
Electrical and Computer Engineering, Brigham Young University, Provo, UT, USA
4. Christiansen RS (2004) Design of an autopilot for small unmanned aerial vehicles. M.Sc. thesis,
Brigham Young University, Provo, UT, USA
5. Vural Y (2007) Autopilot system design for a small unmanned aerial vehicle (in Turkish). M.Sc.
thesis, Department of Aeronautical Engineering, Istanbul Technical University, Istanbul, Turkey