You are on page 1of 18

J Intell Robot Syst (2015) 77:229–246

DOI 10.1007/s10846-014-0077-y

Path Planning for Multi-UAV Formation


YongBo Chen · JianQiao Yu · XiaoLong Su ·
GuanChen Luo

Received: 19 August 2013 / Accepted: 19 June 2014 / Published online: 20 July 2014
© Springer Science+Business Media Dordrecht 2014

Abstract This paper presents an efficient and feasible rationality of the path planning results. The simula-
algorithm for the path planning problem of the multi- tion results show that the artificial potential method
ple unmanned aerial vehicles (multi-UAVs) formation with the additional control force improved by the
in a known and realistic environment. The artificial optimal control method has a good path planning
potential field method updated by the additional con- ability for the single UAV and the all UAVs for-
trol force is used for establishing two models for the mation. At the same time, the path planning results
single UAV, which are the particle dynamic model and are available and the UAVs can basically track the
the path planning optimization model. The additional UAV formation.
control force can be calculated by using the opti-
mal control method. Furthermore, the multi-UAV path Keywords Multi-UAV path planning · Virtual
planning model is established by introducing “vir- velocity rigid body · Virtual target point · Optimal
tual velocity rigid body” and “virtual target point”. control · The artificial potential field · Path following
Then, the motion states of the lead plane and wingmen
are obtained from the path planning model. Finally,
the path following process based on the quadrotor 1 Introduction
helicopter PID controllers is introduced to verify the
The multi-UAV path planning is a process that the
UAVs find their own paths from their starting points to
Y. Chen · J. Yu () · X. Su · G. Luo their destinations cooperatively. It has been a research
School of Aerospace Engineering, Beijing Institute focus in recent years. Of course, it is based on the
of Technology, Beijing 100081, China path planning of the single UAV. Comparing to the sin-
e-mail: jianqiao@bit.edu.cn gle UAV path planning, the multi-UAVs need to deal
Y. Chen with their cooperative relationships. As the formation
e-mail: bit chenyongbo@163.com fly is an important cooperative way in many situa-
X. Su tions, the concept of the formation path planning is
e-mail: suxiaolong@bit.edu.cn presented to solve the multi-UAV path planning prob-
G. Luo lem. Specifically, the formation path planning means
e-mail: gc luo@eyou.com each UAV finds its own collision free path and simul-
Y. Chen · J. Yu · X. Su · G. Luo
taneously tries to keep their formation structure. Thus
Key Laboratory of Dynamics and Control of Flight Vehicle, the key to solve the multi-UAV formation path plan-
Ministry of Education, Beijing 100081, China ning is to solve: the single UAV path planning, the
230 J Intell Robot Syst (2015) 77:229–246

unique selection of the formation structure, the rule of improved by optimal control method is used in the
the formation variation and refactoring process. single UAV path planning and UAV formation config-
The main problems in multi-UAV path planning uration. The multi-UAVs in the formation can obtain
are the single UAV path planning, formation constitu- their own collision free paths and keep their forma-
tion, formation configuration and formation retention. tion structure simultaneously by applying this method.
Some mature methods have been set up for the sin- In order to verify the rationality of the path planning
gle UAV path planning in recent years, such as: A * result, we use the multiple quadrotor helicopters to
method [1], artificial potential field method [2], reced- follow the planning results. And the whole path fol-
ing horizon algorithm [3] and a large number of intel- lowing control system is designed by the PID control
ligent algorithms [4, 5]. Although these methods have method.
solved the single UAV path planning successfully, The structure of this paper is as follows: In the
most of them are difficult to be applied in the multi- second section, the single UAV path planning opti-
UAV formation path planning immediately. There are mization model using the artificial potential field
many researches being put forward to study the for- method with additional control is built on the single
mation constitution, configuration, retention and con- UAV particle dynamic system. In the third section,
trol. Anderson B. D. O. and Yu Changbin did some the multi-UAV formation path planning model is built
research on fixed-formation system configuration [6]. with the introduction of the virtual speed rigid body
Xiaohua Wang successfully solved the problem in and the virtual target point. In the fourth section,
UAV formation constitution combining optimal con- the functional optimization model with constraints
trol and model predictive control [7]. At the aspect of is solved by the optimal control method approxi-
the multi-UAV formation retention, Das A V, Fierro matively. Furthermore, the path planning results are
R. established the formation model based on the com- obtained by substituting the control force into the
bination of bionics and vision [8]. Chang BoonLow UAV’s particle dynamics system. In the fifth section,
solved the multi-UAV path planning based on dynamic the PID controller system for the path following
virtual structure [9]. All the above methods provide process is designed on the quadrotor helicopter plat-
a solid foundation to solve the multi-UAV formation form. In the sixth and seventh section, the path
path planning problems. But their researches focus planning and path following results of the single
on either the single UAV path planning or the for- UAV and multi-UAV formation are obtained and ana-
mation method separately. The combination of them lyzed using the Matlab simulation in the simple and
isn’t considered for the multi-UAV formation path complex planning spaces. In the last section, some
planning. conclusions are drawn and further discussions are
Some researchers also have provided some meth- presented.
ods for the multi-UAV path planning. For example,
C. R. McInnes built the single UAV and multi-UAV
path planning model based on velocity potential field 2 Single UAV Particle Dynamics and Optimization
[10]. Alberto Bemporad and Claudio Rocchi applied Model in the Path Planning Process
the Model Predictive Control (MPC) method on the
multiple rotor UAVs for path planning [11]. However, 2.1 Single UAV Particle Dynamics in the Path
most of the above methods are applied in the two- Planning Process
dimensional plane. And most of them only are based
on the simple planning spaces. Moreover, the results In order to simplify the algorithm, we can use the sim-
of the multi-UAV path planning aren’t optimized by plified particle dynamics to replace the real dynamic
given indexes. Moreover, the concept of the formation model in the path planning process. Then, the single i-
fly is weakened during the path planning process. th UAV particle dynamics is based on the APF method
This paper classifies the multi-UAVs by the lead with the additional control force. The main idea of the
plane and wingmen. The virtual rigid body is estab- APF method is to use repulsive potential fields ema-
lished based on the position and attitude of the lead nating from the obstacles to force the UAV away and
plane. After arranging the virtual target point on the an attractive potential field emanating from the target
virtual rigid body, the artificial potential field method as to attract the UAV [12].
J Intell Robot Syst (2015) 77:229–246 231

The forms of the repulsive field functions and attractive field and is defined as F y = −∇ϕ(pi∗ ) =
attractive field functions are various. Many entirely p∗
−2ky pi∗  , where ky. means the alterable coefficient
different field functions have been used in the ref- i
of the attractive field. For the different planning space,
erences [13]. The repulsive field and attractive field
kφ , lφ and ky need to vary with the size of the planning
are defined out of its simple mathematical form as
space. The resultant force joined by the potential field
follows:
force and control force is used to control the UAV to
The repulsive field φ i (pi ) is caused by the obsta-
avoid collision and reach target.
cles and it has a safe space that the repulsive field
The particle dynamics system of UAV under resul-
can only exist at a safe distance. So the UAV needs
tant force is [12, 15]:
to search the nearest points on the different obsta-  
  t 2
cles to check whether it has entered into the safe I tI I
space. The repulsion caused by the whole obstacles X(k + 1) = X(k) + 2 a(k), (1)
0 I tI
can be simplified into causing by the nearest points.
It is defined as φi (p i ) = kφ e−lφ pi  [14], where kφ
2
where X(k) is the motion state of UAV at time k
and lφ mean the alterable coefficients of the repulsive which includes the speed v(k) and position p(k) of 
p(k)
field (kφ >0, lφ >0). The repulsion is the negative UAV at time k and the expression is X(k) = .
v(k)
gradient of repulsive field and is defined as F c =
a(k) is the resultant force of potential field force q(k)
−∇φi (pi )= 2kφ lφ p i e−lφ pi  , where pi is the vec-
2
(repulsion and attraction), the additional control force
tor that the UAV points to the nearest point on the u(k) and damping force Fz (k) (Fz (k)=kz v(k), where kz
obstacles. The repulsion can help the UAV avoid the means the drag coefficient).
obstacles during the path planning process. Accord-
ing to the form of the repulsion Fc , the change of the 2.2 Optimization Model for the Single UAV Particle
function value is shown in Fig. 1 [12]. Thereinto, the Dynamics
value A and B are based on the alterable coefficients
kφ and lφ . For the repulsion, it generally demands that The optimization model is widely used in the single
the smaller ||pi || causes the bigger repulsion ||Fc || UAV path planning problem. Importantly, the plan-
(Hatched Area). So the value A needs to be much less ning path of the UAV can be optimized by given
than the size of the safe distance. indexes. So the optimization model for the single UAV
The target point generates the attractive field filled particle dynamic can be expressed as [16]:
full planning space whose potential field is ϕ(p∗i ),
where p∗i is the vector that the UAV points to the tar- minu(k) J (X(k), u(k))
get point. The attraction is the negative gradient of the s.t. F ((X(k), u(k)) = 0
G((X(k), u(k)) > 0 (2)

2.2.1 Objective Function

It is known that searching for the best path is often


associated with searching for the shortest path. Obvi-
ously, the shorter path can save more fuel, save more
time and give more security. At the same time, the
smaller additional control force of the UAV is also
important for the UAV path planning process. The
control force is directly related to the energy. So
the objective function of the optimization model for
the single UAV particle dynamics is based on the
weighted indexes of the shortest path Js and the least
Fig. 1 Sketch map of the repulsion energy Je [17]. The indexes try to make the additional
232 J Intell Robot Syst (2015) 77:229–246

control energy consumption as small as possible and Boundary conditions: according to the path plan-
the path as short as possible in the premise of com- ning process, the UAV flies from the starting point S to
pleting the path planning. Although the contradiction the target point T. So the boundary constraints of the
of the two indicators is clear, the weighted model can optimization model are as follow, i.e. p(0)= S, p(n)=
adjust the weight to achieve the purpose of adjusting T.
optimization objective. The expression is:
2.2.3 Single UAV Optimization Particle Dynamics
minu(k) J = αJ s + βJ e
Model

n−1
= (α p(k + 1) − p(k)2 + β u(k)2 ) In summary, the optimization model of the single UAV
k=0
particle dynamics is:
(3)
where α and β are the adjustment coefficients of the
objective function (α >0, β >0). min J = αJ s + βJ e (4)
u(k)
   
I tI t 2
2.2.2 Optimum Conditions I
X(k + 1) = X(k) + 2 a(k)
0 I tI
The path planning results need to be applied in the s.t. u ≤ um .
real UAV automatic flight control system, so the rea- p(0) = S
sonable optimization conditions are the premise of p(n) = T
the accuracy and reasonableness of the optimization
model. Considering the whole UAV flying process,
the optimum conditions in the optimization model for 3 Multiple UAVs Path Planning Model
the single UAV particle dynamic include performance
constraints, spatial constraints, dynamic constraints 3.1 Multiple UAVs Formation Configuration
and boundary conditions.
Performance constraints: there are speed con- The formation structure of multi-UAVs is divided into
straints and overload constraints on the UAV body. the close formation and loose formation according to
Only in the reasonable constraint range of the speed whether there is aerodynamic coupling [18]. Because
and overload, the path planning results are trackable of the complex aerodynamic coupling in close forma-
for the real UAV control system. So the performance tion, the close formation configuration achieves the
constraints which include the speed constrains and goal of saving funnel and expanding combat radius
overload constraints are very necessary for the opti- based on bionics principles. But the mechanism of
mization model. For simplicity, the form of the over- the close formation is complex and changeful. So the
load constraints and speed constraints is similar to the loose formation is also used widely. The loose for-
form of speed constraints u ≤ um in the suitable mation has variable formations considering the com-
artificial field. munication and combat requirements. In this paper,
Spatial constraints: obviously, the UAV needs to the whole formation structure is on the basis of the
avoid the obstacles in the planning space. In other loose formation (lead plane- wingmen) without regard
words, the path planning points are not allowed to to the aerodynamic coupling. Specifically, the loose
appear in the obstacles. The existence of the repul- UAV formation applied in this paper is a herringbone
sion fields keeps the UAV away from the obstacles. So formation and its size is shown in Fig. 2.
the spatial constraints can be satisfied by the repulsive The herringbone formation is similar to the flight
field of the artificial potential field method [12]. mode of the geese flock and is widely used in close
Dynamic constraints: the optimization variable formation and loose formation. The concept of “vir-
u(k) and state variable X(k) are not independent. tual speed rigid body” is introduced to achieve the
Equation 1 shows the differential and integral rela- lead plane- wingmen topological structure of UAVs
tionship among the force, velocity and acceleration of formation model. “Virtual speed rigid body” is the vir-
UAV. tual rigid body model based on the velocity vector of
J Intell Robot Syst (2015) 77:229–246 233

Fig. 2 Schematic diagram which is perpendicular to x  in the vertical plane; z


of the multi-UAV formation wingman axis is obtained according to the right-hand rule. The
conversion relationship between the coordinate in vir-
tual rigid body coordinate system and coordinate in
wingman inertial coordinate system is shown as follows:
⎡ ⎤ ⎡ ⎤
x x
⎣ y  ⎦ = AB ⎣ y ⎦ , (5)
Lead plane z z
120deg
⎡ ⎤
cos θ sin θ 0
A = ⎣ − sin θ cos θ 0 ⎦ , (6)
200m 0 0 1
wingman
⎡ ⎤
cos ψ 0 − sin ψ
B=⎣ 0 1 0 ⎦, (7)
wingman sin ψ 0 cos ψ
where [x y z ]T is the coordinates in the virtual rigid
the lead plane. The rigid body of the whole multi- body coordinate system for the UAVs; [x y z]T is
UAV formation relies on the velocity vector of the lead the coordinates in inertial coordinate system for the
plane. And the ideal position of each wingman in for- UAVs; θ is the inclination angle between the velocity
mation is S i . In other words, S i is the virtual target vector of the lead plane and horizontal plane; ψ is the
point of the i-th wingman. yaw angle between the velocity vector of the leading
The virtual velocity rigid body is shown in Fig. 3. plane and horizontal plane.
The lead plane locates in the “head” of the herring-
bone formation and the attitude angle of the lead plane 3.2 Motion Model for Multi-UAV Formation
is used to determine the attitude angle of the rigid
body. Since this paper uses the UAV particle dynamics The motion model for the multi-UAV formation is
model, the attitude information of the rolling channel based on the single UAV particle dynamic model and
is already out of consideration. The velocity vector of the multi-UAV formation configuration (Fig. 4).
the lead plane is defined as the x  axis of virtual rigid With the introduction of the virtual velocity rigid
body coordinate system; y  axis is defined as the axis body, the path planning model for the lead plane is
independent. Then, the wingmen need to follow the
virtual position on the virtual velocity rigid body to
maintain the UAV formation. So “virtual target point”
is introduced to help the wingmen to keep the for-
mation during their own path planning processes. In
other words, the ideal positions of the wingmen are
determined by the virtual target point and the virtual
velocity rigid body. The virtual target points become
the planning targets of the wingmen. Importantly, the
wingmen use the same path planning method as the
lead plane. So the functional relationship is as follows:
X(k) = AP F (k − 1, T , O)
T ∗i = X(k) + B T AT S i , (8)
X∗i (k) = AP F (k − 1, T ∗i , O)
where AP F is the path planning function of the artifi-
Fig. 3 Schematic diagram of the virtual velocity rigid body cial potential field method with the additional control
234 J Intell Robot Syst (2015) 77:229–246

Fig. 4 Information flow of


the multi-UAV formation
Path planning Path planning

APF Virtual APF


Virtual
Target Lead Wing-
target
point plane rigid body man
point

u u

force; i is the number of UAVs, T is the target point safe radius of the collision-avoidance potential field
of the lead plane; T ∗i is the virtual target point of each is 20m. It means the UAV will be subjected to the
wingman; O are the set of the obstacles; Si is the ideal repulsion when it gets into the other UAVs’ safe radius
coordinate of each wingman in the body-fixed coor- (Fig. 5).
dinate system; X∗i is the actual path planning state of
each wingman; X is the actual path planning state of Remark 1 Sometimes the virtual target points of the
the lead plane. wingmen may locate in the obstacles, according to the
formation structure of Section 3.1. The main reason is
3.3 Collision-avoidance Model for Multi-UAVs the virtual target points of the wingmen are based on
the planning path of the lead-plane. Although the lead-
The UAVs may collide with each other without a plane can avoid the obstacles easily under the help of
proper arrangement in the formation. The repulsion the path planning method, the virtual speed rigid body
field of the artificial potential field method is intro- and virtual target point may enter into the obstacles,
duced to handle this problem. When the distance as shown in Fig. 6 below. So in order to prevent the
between each two UAVs is less than the danger value, wingmen from flying into the obstacle, the repulsion
the repulsion field of UAV works only on the other force is much bigger than the attractive force. In other
UAVs and won’t work on itself. At the same time, words, the collision avoidance is important than the
the path planning of the lead plane won’t be affected formation fly. If the UAV flies into the repulsive force
by the repulsion fields of the wingmen so as to fields of the obstacles and the other UAVs, it will be
keep the dependence of the lead plane path planning. manifest as giving up the formation. So the formation
Therefore, the wingman is subjected to the repul- errors are big when they fly into the repulsive force
sion field of other wingmen and the lead plane. The fields. What’s more, it is not appropriate to judge the

Fig. 5 Repulsive force field


between the multi-UAVs
J Intell Robot Syst (2015) 77:229–246 235

The cost function is rewritten as:


n−1 
n−1
J =α p(k + 1) − p(k)2 + β a(k)2
k=0 k=0
n−1 
   2
 p(k) t 2 
=α  t 2 + u(k)
 2 G(k) tI
v(k) 2 
k=0

n−1
+β u(k)2
k=0
(10)

where G(k) is the simplified transitive matrix that


varies with p(k) and r(k). It fits:
Fig. 6 Special circumstances
r(k) = G(k)p(k). (11)

In order to transform Eq. (10) into the quadratic form,


formation-flight performance by the big errors during
partition the following quadratic form matrixes as:
these situations.
⎡ 4 ⎤
t
G T (k)G(k) t 3 G(k) t 4 G(k)
⎢ 4 t 3 T 2 4
t 3 ⎥
⎣ 2 G (k) t 2 I 2 I ⎦
, (12)
t 4 T t 3 t 4
4 Solution of Optimal Control for the Path 4 G (k) 2 I 4 I
Planning Optimization Model
⎡ ⎤
0 0 0
The main difference between the lead plane and wing- ⎣0 0 0⎦. (13)
men in the path planning process is the target point.
0 0 I
The lead plane is in charge of the overall path while
the wingman is responsible for maintaining the forma- Define:
tion. But all UAVs can use the same method to solve  
their own single path planning problems. Then, the t 4 T t 3
Q (k) = 4 G (k)G(k) 2 G(k) ,
3
optimal control is applied in solving the optimization t T t 2 I
2 G (k)
model.  
0 0
The general form of discrete optimal control system Q1 = , (14)
0 0
is:
−1
kf
minu(k) J = φ x(kf ), kf + ϕ(x(k), u(k), k)    
t 4
4 G(k) 0
k=k0
S(k) = t 3
, S1 = , (15)
x(k + 1) = g(x(k), u(k), k) (9)
2 I
0
s.t. h(x(k), u(k), k) ≥ 0 .
x(k0 ) = x0
t 4
The optimization model of the UAV dynamic model R= I, R1 = I, (16)
4
needs to be converted to the general form of the opti-
mal control problem and it includes the conversion    
2 t 2
of the target functions and conversion of the con- I + t2 G(k) tI
straint conditions. Because this paper focuses on the
(k) = , = 2 .(17)
tG(k) I t
path planning for the multi-UAV formation, this solu-
tion procedure is sententious. The similar detailed With the definition of Eqs. 14–17, the cost function
derivation process is shown in reference [12]. is formed into the quadratic form, and the dynamic
236 J Intell Robot Syst (2015) 77:229–246

constraint is formed as the state equation. By now, the Then[19, 20]:


optimization model is rewritten as:  u (k) u (k) u (k) T
n−1  T u (k) = x y z
1  X T (k)
minu(k) J = u
2 uT (k) min x (k) umin y (k) umin z (k) T

k=0
  umin (k) = ,
2αQ(k) + 2βQ1 2αS(k) + 2βS1 X(k) ⎧
2αS T (k) + 2βS1T 2αR + 2βR1 u(k) ⎨ um (umin • ≥ um )
⎧ u• (k) = umin •(k) (umin (k) ∈ U )

⎪ X(k + 1) =
(k) X(k) + u(k) ⎩

⎨ T −um (umin • ≤ −um )
X(0) = p Ts vT (0)
s.t. T T T (24)

⎪ X(n) = p t v (n)


u(k) ≤ um where the subscript • represents x, y, z and umin (k)
(18) is the optimal control u (k) without the control con-
straint, which is:
Define:
umin (k) = −R̃ −1 (k) T (P −1 (k + 1) (25)
Q2 (k) = 2αQ(k) + 2βQ1 , (19)
 −1 T −1  
+ R̃ (k) )
(k)X (k),

where P(k) can be calculated by the corresponding


S2 (k) = 2αS(k) + 2βS1 , (20)
matrixes
 (k),  , Q̃(k) andR̃(k), via the Riccati
equation:
R2 = 2αR + 2βR1 . (21) P (k) =
 (k)P (k + 1)
 (k) + Q̃(k) −
 (k)P (k + 1) 
T T

 −1 T
R̃ +  P (k + 1)   P (k + 1)
 (k). (26)
T
Go on with the variable substitutions:

Q̃(k) = Q2 (k) − S2 (k)(R2−1 )T S2T (k) (22)


R̃ = R2 Then, the result of optimal control of discrete opti-


(k) =
(k) − 
(R2−1 )T S2T (k). mal control system with constraints: u (k) is obtained
from the former equations Eq. 24 and u(k) is obtained
 =
according to Eq. 22. u(k) can be substituted into the
X (k) = X(k) UAV dynamics equations. Each UAV can get its own
u (k) = (R2−1 )T S2T (k)X(k) + u(k) planning path by the help of its u(k).

All above, the general form of the discrete optimal


control that satisfies Eq. (9) is obtained from the con-
5 Path Following Control
version of the constraint conditions and cost function
of the optimization model Eq. (3).
It’s known that the UAV path planning model is built

1 T  on the artificial potential field method. The rational-
minu(k) J = n−1 k=0 2 X (k)Q̃(k)X (k)
 ity of the path planning result can’t be guaranteed
+ 12 uT (k) R̃u (k) considering some constraints. So after the multi-UAV
⎧ 

⎪ X (k + 1) =
 (k)X (k) +  u (k) , (23) path planning process, the path following control is

⎨ T
X (0) = p Ts vT (0) an important link to verify the rationality of the path
s.t. T planning result. In the path following process, the

⎪ X  (n) = p Tt vT (n)

⎩ real dynamic model of UAVs needs to be applied to
u (k) ∈ U
simulate the real multi-UAV formation flying state.
where U is the allowable interval of the control force. The quadrotor helicopters are used to follow the plan-
By the help of the optimal control method for ning results which include the position information of
the discrete optimal control system, we can get the the lead plane and wingmen. Obviously, the position
approximate result u . information is time-varying [12].
J Intell Robot Syst (2015) 77:229–246 237

5.1 Dynamics Model of the UAVs in Path Following Y b , Z b ; uci (i=1,2,3,4) satisfy the following equations
Process [23]:
uc1 = b(ω12 + ω22 + ω32 + ω42 )
The quadrotor helicopter is an under actuated aircraft
uc2 = b(ω22 − ω42 )
with fixed four pitch angle rotors as shown in Fig. 7 , (28)
uc3 = b(ω12 − ω32 )
[12, 21]. It has several obvious characteristics such as:
uc4 = d(ω12 − ω22 + ω32 − ω42 )
highly nonlinear, strongly coupled and underactuated
system. where b is the scaling factor of the thrust generated by
In the structural diagram of the quadrotor heli- the rotor; d is the scaling factor of the propeller torque
copters, ψ  , θ  , φ  represent the Euler angles between constant generated by the rotors.
the body-axis coordinate system B =(X b , Y b , Z b ) and Above models are based on the following
the inertial coordinate system E=(X be , Y be , Z be ). So assumptions [24]:
via the Newton’s laws and Euler equation, the simpli-
(1) The UAV structure is symmetrical and rigid;
fied dynamic model of the UAV can be obtained as
(2) The center of mass and Ob coincides;
follows [22]:
(3) Thrust and drag are proportional to the square of
the propellers speed;
1 Kx
ẍb = (cos φ  cos ψ  sin θ  + sin φ  sin ψ  )uc1 − ẋb (27) (4) Ignore the rotational resistance of each UAV;
m m
1 Ky (5) Ignore the effect of noise.
ÿb = (cos φ  sin ψ  sin θ  − sin φ  cos ψ  )uc1 − ẏb
m m
1 Kz 5.2 PID Control for Formation Path Following
z̈b = (cos θ  cos φ  )uc1 − żb − g
m m
 
l Iy − Iz The PID controllers are mentioned in many refer-
φ̈  = uc2 + θ̇  ψ̇  ,
Ix Ix ences. The PID controllers are the standard tools in
 
l Iz − Ix
θ̈  = uc3 + φ̇  ψ̇  current industrial automation experience thank to their
Iy Iy
  flexibility.
1 I − Iy
ψ̈  = uc4 + φ̇  θ̇ 
x The PID controller takes many structures but the
Iz Iz
most important one is in the following form [25]:
t
de(t)
where xb , yb , zb are the positions of the UAV;m is ufollow (t) = KP e(t) + KI e(τ )dτ + KD . (29)
dt
the mass of the UAV; g is the acceleration of grav- 0
ity; l is the half length of the quadrotor helicopter;K x , The algorithm of the PD controller is shown as
K y , K z mean the drag coefficients; I x , I y , I z are the follows:
moments of the inertia with respect to the axes X b ,
de(t)
ufollow (t) = KP e(t) + KD , (30)
dt
where uf ollow (t) is the control variable, the track-
ing error e(t) is defined as e=W r -W where W r is
the scheduled output and W is the real output. K P ,
K I and K D are controller gains associated with pro-
portional (P), integral (I), and derivative (D) actions,
respectively.
In Section 4, we obtain the path planning results
of the whole UAV formation which include the
lead plane and wingmen. Those time-varying results
are the input of the whole path following con-
trol system. Since the main research content of
this paper is focused on the multi-UAV path plan-
Fig. 7 Quadrotor helicopter structural diagram ning, the simple traditional PID controller and PD
238 J Intell Robot Syst (2015) 77:229–246

controller are applied to solve the path follow-


ing problem for the whole multi-UAV formation.
The structure of the control system is shown as
followed:
In the structure chart, the horizontal position PD
controller is as follows [23]:

θri (t) = arcsin(Kpx (xpi (t) − xi (t)) + Kdx (ẋpi (t) − ẋi (t)))
,
φri (t) = − arcsin(Kpx (ypi (t) − yi (t)) + Kdx (ẏpi (t) − ẏi (t)))
(31)

The altitude PD controller with the nonlinear compen-


sation is as follows:
Fig. 9 Following result of the simple example
1
uc1i (t) = (K z (zp −z)+Kdz (żp −ż)+mg).
cos φi cos θi p

an acceptable range. The speed of the UAV is as
(32)
follows:
In order to verify the effectiveness of this control sys- Figures 9 and 10 show the validity of the control
tem, a simple example is provided. Concretely, in a system, and these results provide some help about the
small scale example, the UAV takes off from [0, 0, 0] rationality for later simulation work.
to [4, 8, 0] using a predetermined orbit which is shown
in Fig. 9. Moreover, the magnitude of the velocity
remains 2m/s and the direction changes with the pre- 6 Simple Simulation Examples
determined orbit. Finally, the result is shown in Fig. 9.
The red line is the reference trajectory, and the blue 6.1 Planning Space and Constraint Conditions
line means the real path following result which uses
the control system in Fig. 8. The formation simulation is carried out using 5 UAVs
Figure 9 shows the simulation results of the and it is a simple no-obstacle space. This simple
path following of this reference trajectory. It can simulation is provided in order to evaluate formation-
be seen that the following error is small and in flight performance on simple scenario without the

Fig. 8 Control system chart


xbi (t ), ybi (t ), zbi (t )

'i (t ), 'i (t ), 'i (t )


ri (t )
x pi (t ) uc 2i (t )
ri (t )
y pi (t ) uc 3i (t )
ri (t )
uc 4i (t )

z pi (t )

uc1i (t )
'i (t ), 'i (t ), zbi (t )
J Intell Robot Syst (2015) 77:229–246 239

Fig. 10 Flying state of the


simple example

effects of the obstacles. The size of the planning whose shape is always straight. The result is shown
space is as follows: the range of x axis: [1500m, in Fig. 11.
6000m]; the range of y axis: [1500m, 7000m]; the The simulation result is a straight space curve
range of z axis: [-50m, 1200m]. The target point which is in line with expectations. But because the
of the UAV group is [6000m, 7000m, 1000m]. The simulation scenario is too simple, it is not enough
specific circumstances of the starting points are to verify the effectiveness of the single UAV path
as follows: planning method.

6.2 Path Planning of Lead Plane 6.3 Path Planning of Multi-UAV Formation

In the no-obstacle space, the path planning result The path planning of the multi-UAV formation uses
of the lead plane is a very simple space curve the initial parameters which are listed in Table 1. In the
no-obstacle space,the simulation result of the multi-
UAV formation is shown in Fig. 12.
Figure 12 indicates that the UAVs set out from
the different starting points and form the formation
when they are flying. The UAVs fly in the expected
formation and reach the target point at the same
time. In order to describe the formation-flight per-
formance exactly, the formation error results of four
wingmen are shown in Fig. 13. The formation error

Table 1 Table 1 Starting points coordinates of the UAVs

UAV (x, y, z)/m

Lead plane (blue) (1500,2000,500)


Wingman 1 (black) (2000,1500,200)
Wingman 2 (red) (1500,1500,400)
Wingman 3 (yellow) (1500,2000,500)
Wingman 4 (green) (1500,4000,500)
Fig. 11 Single UAV path planning result in a no-obstacle space
240 J Intell Robot Syst (2015) 77:229–246

From these results, the path planning of multi-UAV


formation can be divided into two different stages: In
the beginning of the path planning process, the UAVs
set out from the different starting points. The virtual
target points of the wingmen are faraway from the
different starting points. In the other word, they do
not form the formation yet. So the formation errors
are big.The results are reasonable. After they formed
the multi-UAV formation, the whole multi-UAV for-
mation gets into the stableflying state. Figure 13
shows that the formation errors of the path plan-
ning results are all below 2m (Amplifying part in
the last figure). In conclusion, the formation-flight
performance is very good in the simple no-obstacle
space.

6.4 Path Following of Multi-UAV Formation

Fig. 12 Single UAV path planning result in a no-obstacle space The rationality of the multi-UAV formation path plan-
ning results needs to be verified by the path fol-
lowing process. Naturally, the path planning results
containing the whole formation position information
means that the distance between the path planning become the input of the path following control system.
position of the i-th wingman and its correspond- Because of using the same control system, the effects
ing virtual target point (wingman i corresponds to of the path following are no differences between
distance i ). the lead plane and wingmen in the path following

Fig. 13 Formation error results of four wingmen in a no-obstacle space


J Intell Robot Syst (2015) 77:229–246 241

Fig. 14 Path following


error of multi-UAVs in a
no-obstacle space

process. No matter the lead plane or wingmen, the polation. The obstacles include: alpine terrain, radar
all path following simulations are finished. The path zone, bad weather zone and so on. These obstacles are
following result is shown in Fig. 14. very common in the real battlefield environment. The
Where the red lines mean the real path follow- specific circumstance is as follows (Fig. 15):
ing results, and the blue lines mean the multi-UAV All obstacles including the radar areas and terrain
path planning results. During the whole path follow- in the planning space are regarded as the no-fly zone
ing process, the path following error of each UAV is where the trajectories of UAVs cant be allowed into.
all below 2m, and the velocity and the acceleration The horizontal velocity constraint of the UAV is vzh =
are all below the given upper limits. So the control 10 m/s, the vertical velocity constraint of the UAV is
system can help the multi-UAV formation to achieve vzm = 5 m/s; the horizontal acceleration constraint
the multi-UAV formation path planning successfully of the UAV is azh = 10m/s2 , the vertical accelera-
in simple on-obstacle space. tion constraint of the UAV is azm = 10 m/s2 ; the
flight height constraint of the UAV is z > z (surface),
where z (surface) means the interpolation result of the
7 Complex Simulation Examples planning space.

7.1 Planning Space and Constraint Conditions 7.2 Path Planning of Lead Plane

In the complex environment, the complicated and The flight status of the single UAV is the same as
actual planning space is built by the MATLAB inter- the lead plane in the formation, so the prior choice is

Fig. 15 Planning space in a


complex simulation
242 J Intell Robot Syst (2015) 77:229–246

Fig. 16 Single UAV path


planning result in a complex
simulation

verifying the path planning ability of the single UAV. simulation result of the multi-UAV formation is shown
The optimal control artificial potential field method is in Fig. 17.
applied in the path planning from the starting point Figure 17 indicates that the UAVs set out from the
(lead plane) to the target point. The result is shown in different starting points and form the formation when
Fig. 16. they are flying. The UAVs fly in the expected for-
The simulation result shows that this method has mation and reach the target point at the same time.
a good path planning ability in the given planning In order to describe the formation-flight performance
space and it overcomes the problem of the dead exactly, the formation error results of four wingmen
zone in the path planning procedure. Combining with are shown in Fig. 18.
the simple simulation result, it is enough to verify The simulation results show that the formation
the effectiveness of the single UAV path planning errors are big in the influence area of the obsta-
method. cles, and the formation errors are small in the no-
obstacle area (below 8m). Considering the remark 1
7.3 Path Following of Multi-UAV Formation in Section 3.3, these results are reasonable. There-
fore, the path planning of the multi-UAV forma-
The path planning of the multi-UAV formation uses tion is very successful in the complex simulation
the initial parameters which are listed in table 1. The environment.

Fig. 17 Multi-UAV
formation path planning
result in a complex
simulation
J Intell Robot Syst (2015) 77:229–246 243

Fig. 18 Formation error results of four wingmen in a complex space

7.4 Path Following of Multi-UAV Formation the blue lines mean the multi-UAV path planning
results.
The rationality of the multi-UAV formation path plan- Because the research focused in this paper isn’t the
ning results in the simple no-obstacle space are shown control system, we only show the simulation result of
in Fig. 13. But in the complex environment the multi- the lead plane which can be regarded as a representa-
UAVs face the effects of the multifold obstacles. So tive of the whole formation. The 3-axis path following
the rationality needs to be further verified by the path error of the lead plane is shown in Fig. 20.
following process in the complex environment. The Then the flight states are shown in Fig. 21. Accord-
path following result is shown in Fig. 19 where the ing to the path following error, there are six differ-
red lines mean the real path following results, and ent flying stages: In the first section (0- 300s), the

Fig. 19 Path following


error of multi-UAVs in a
complex simulation
244 J Intell Robot Syst (2015) 77:229–246

Fig. 20 Following error result of the lead plane

multi-UAVs set out from the different starting points. oscillatory and unstable motion which is one of four
Obviously, the flying stages of formation are fluctu- inherent defects of the APF method. The four inherent
ating and unstable. In the section (1000- 2700s), they defects are shown in [26]. The degree of the oscillation
enter the first radar obstacle area. Then, the UAVs is based on the time step t and the repulsion gradient.
need to give up the formation. Under the effect of the The bigger time step t and repulsion gradient will
repulsion, the path planning results are erratic. This cause the stronger oscillation. Then the multi-UAVs
phenomenon is called as the oscillation problem. In need to follow these erratic path planning results, so
the presence of obstacles, the multi-UAVs enter into the paths following errors are big. In the sections

Fig. 21 Flying state of the lead plane


J Intell Robot Syst (2015) 77:229–246 245

(3300- 4000s), they enter the second radar obstacle rithm. Sci. China Technol. Sci. 55(10), 2712–2719 (2012).
area. Uniformly, they face the same problems during doi:10.1007/s11431-012-4890-x
5. Roberge, V., Tarbouchi, M., Labonte, G.: Com-
this process. In the other sections (300- 1000s, 2800-
parison of Parallel Genetic Algorithm and Particle
3300s, 4000- 5600s), the whole formation gets into Swarm Optimization for Real-time UAV Path Plan-
the stable flying state. Their path planning processes ning. IEEE Trans. Ind. Inform. 9(1), 132-141 (2013).
are similar to ones in the simple simulation. The path doi:10.1109/TII.2012.2198665
6. Anderson, B.D.O., Yu, C., Baris, F.: Information Archi-
planning results are straight and stable. So the tracking
tecture and Control Design for Rigid Formations. In: 26th
error is below 1m when they get into the stable flying Chinese Control Conference, 26-31 July 2007, Piscataway,
state, comparing to the UAV formation scale which NJ, USA 2007. Proceedings of the 26th Chinese Control
is 1 %. All above, we can see the velocity and the Conference. IEEE (2007)
7. Xiaohua, W., Yadav, V., Balakrishnan, S.N.: Cooperative
acceleration are all below the given upper limits which
UAV Formation Flying with Obstacle/Collision Avoidance.
showed in the Section 6.1. Therefore, these results IEEE Trans. Control Syst. Technol. 15(4), 672–679 (2007).
verify the rationality of the path planning algorithm. doi:10.1109/TCST.2007.899191
The control system can help the multi-UAV forma- 8. Das, A.K., Fierro, R., Kumar, V., Ostrowski, J.P., Spletzer,
J., Taylor, C.J.: A Vision-based Formation Control Frame-
tion to achieve the multi-UAV formation path plannin
work. IEEE Trans. Robot. Autom. 18(5), 813–825 (2002).
g successfully. doi:10.1109/TRA.2002.803463
9. Chang Boon, L.: A Dynamic Virtual Structure Forma-
tion Control for Fixed-wing UAVs. In: 2011 9th IEEE
8 Conclusion International Conference on Control and Automation (ICCA
2011),19-21 Dec. 2011, Piscataway, NJ, USA. IEEE (2011)
10. McInnes, C.R.: Velocity Field Path-planning for Single
Based on the APF method updated by the additional and Multiple Unmanned Aerial Vehicles. Aeronaut. J.
control force, the virtual velocity rigid body and the 107(1073), 419–426 (2003)
virtual target point have been introduced to solve the 11. Bemporad, A., Rocchi, C.: Decentralized Linear Time-
varying Model Predictive Control of a Formation of
multi-UAV formation path planning problem. And Unmanned Aerial Vehicles. In: 2011 50th IEEE Conference
the optimal control method is applied to solve the on Decision and Control and European Control Conference
additional control force. This new method is favor- (CDC-ECC 2011), 12-15 Dec. 2011, Piscataway, NJ, USA,
able for the single UAV path planning in the given 7488-7493 (2011). IEEE
12. Chen Y.-b., Luo G.-c., Mei Y.-s., Yu J.-q., Su X.-l.:
planning space. At the same time, it can also solve UAV path planning using artificial potential field method
the multi-UAV formation path planning problem by updated by optimal control theory. Int. J. Syst. Sci. (2014).
applying the virtual structure. Then, the real UAV con- doi:10.1080/00207721.2014.929191
trol system designed by the simple PID controller is 13. Cetin, O., Zagli, I., Yilmaz, G.: Establishing Obstacle and
Collision Free Communication Relay for UAVs with Arti-
used to verify the results of path planning. Simula- ficial Potential Fields. J. Intell. Robot. Syst. 69(1-4), 361–
tion tests revealed: 1) the effectiveness of the path 372 (2013)
planning method; 2) the feasibility and availability of 14. Charifa, S., Bikdash, M.: Comparison of geometrical, kine-
the real path following according to the path planning matic, and dynamic performance of several potential field
methods. In: IEEE SoutheastCon 2009, 5-8 March 2009,
result. Piscataway, NJ, USA. IEEE (2009)
15. Luo, G.-c., Yu, J.-q., Zhang, S.-y., Zhang, W.: Artificial
Potential Field based Receding Horizon Control for path
planning. In: 2012 24th Chinese Control and Decision Con-
References
ference (CCDC), 23-25 May 2012, Piscataway, NJ, USA.
IEEE (2012)
1. Garcia, M., Viguria, A., Ollero, A.: Dynamic Graph-Search 16. Zhang, X., Duan, H., Yu, Y.: Receding Horizon Control for
Algorithm for Global Path Planning in Presence of Haz- Multi-UAVs Close Formation Control Based on Differen-
ardous Weather (2012) tial Evolution. Sci. China Ser. F (Inf. Sci.) 53(2), 223–235
2. Khatib, O.: Real-time Obstacle Avoidance for Manipulators (2010). doi:10.1007/s11432-010-0036-6
and Mobile Robots. Int. J. Robot. Res. 5(1), 90–98 (1986) 17. Yanyang, W., Tietao, W., Xiangju, Q.: Study of
3. Dunbar, W.B., Caveney, D.S.: Distributed Receding Hori- Multi-objective Fuzzy Optimization for Path Plan-
zon Control of Vehicle Platoons: Stability and String Sta- ning. Chin. J. Aeronaut. 25(1), 51–56 (2012).
bility. IEEE Trans. Autom. Control 57(3), 620–633 (2012). doi:10.1016/S1000-9361(11)60361-0
doi:10.1109/TAC.2011.2159651 18. Hua, S., You, Y., Zhang, H., Song, H.: Receding Horizon
4. Pei, L., HaiBin, D.: Path Planning of Unmanned Aerial Control of UAV Formations. Electron. Opt. Control. 19(3),
Vehicle Based on Improved Gravitational Search Algo- 1–5 (2012)
246 J Intell Robot Syst (2015) 77:229–246

19. Wu, S.: Optimal Control Theory and Application. China 34(5), 519–524 (2012). doi:10.3724/SP.J.1218.2012.
Machine Press, Beijing (2008) 00519
20. Xie, L.-j., Xie, G.-r., Chen, H.-w., Li, X.-l.: Solution to 24. Salih, A.L., Moghavvemi, M., Mohamed, H.A., Gaeid,
Reinforcement Learning Problems with Artificial Potential K.S.: Modelling and PID Controller Design for a Quadrotor
Field. J. Cent. South Univ. Technol. 15(4), 552–557 (2008). Unmanned air Vehicle. In: Automation Quality and Testing
doi:10.1007/s11771-008-0104- Robotics (AQTR), 2010 IEEE International Conference on.
21. Mohamed, H.A., Yang, S., Moghavvemi, M.: Sliding Mode IEEE (2010)
Controller Design for a Flying Quadrotor with Simplified 25. Efe, M.Ö.: Neural Network Assisted Computationally Sim-
Action Planner. In: ICCAS-SICE, 2009. IEEE (2009) ple PID Control of a Quadrotor UAV. IEEE Trans. Ind.
22. Bouabdallah, S., Siegwart, R.: Full Control of a Quadrotor. Informat. 7(2), 354–361 (2011)
In: 2007 IEEE/RSJ International Conference on Intelligent 26. Koren, Y., Borenstein, J.: Potential Field Methods and
Robots and Systems, 29 Oct.-2 Nov. 2007, Piscataway, NJ, their Inherent Limitations for Mobile robot naviga-
USA 2007. IEEE (2007) tion. In: Robotics and Automation, 1991. Proceed-
23. Bai, Y., Liu, H., Shi, Z., Zhong, Y.: Robust Flight Con- ings., 1991 IEEE International Conference on. IEEE
trol of Quadrotor Unmanned Air Vehicles. Jiqiren/Robot (1991)

You might also like