Preprints of the 18th IFAC World Congress Milano (Italy) August 28 - September 2, 2011

Time-to-go Polynomial Guidance Laws with Terminal Impact Angle/Acceleration Constraints

Tae-Hun Kim Chang-Hun Lee Min-Jea Tahk Division of Aerospace Engineering, KAIST, Daejeon, Korea (Tel: +82-42-350-3758; e-mail: thkim@fdcl.kaist.ac.kr) Division of Aerospace Engineering, KAIST, Daejeon, Korea (e-mail: chlee@fdcl.kaist.ac.kr) Division of Aerospace Engineering, KAIST, Daejeon, Korea (e-mail: mjtahk@fdcl.kaist.ac.kr) Abstract: This paper deals with a generalized impact angle control guidance law with terminal acceleration constraint, which is called the Time-to-go Polynomial Guidance (TPG). The guidance command of TPG is initially assumed as a polynomial form of tgo , and then the coecients of this polynomial function are designed to satisfy the terminal impact angle and zero miss-distance constraints. TPG presents a simple form, and it can easily achieve the terminal acceleration and its time-derivative constraints which are predetermined by the specied values. In this study, linear trajectory solutions of TPG are derived and analytically investigated. Numerical and adjoint simulations are performed to investigate the performance of TPG. 1. INTRODUCTION Many advanced guidance laws have been researched to improve the performance of the conventional proportional navigation (PN) guidance law or to achieve some objectives, e.g. minimum energy, minimum ight time, impact angle control. The impact angle control in these objectives has been widely considered to insure a high kill probability, and to increase warhead eect of antitank missiles or antiship missiles. It is also used to satisfy ight-path constraints, which depend on missions of reconnaissance unmanned aerial vehicles (UAVs), such as path planning. A terminal attitude angle control guidance law was devised for the reentry vehicle in Kim et al. [1973]. This was probably the rst attempt to design a terminal impact angle control guidance law by solving the linear quadratic optimal control problem. In Bryson et al. [1975], a optimal solution of linear engagement problem, minimizing the control eort subject to lateral position and velocity constrains, has been investigated. Another optimal impact angle control guidance law for varying velocity missiles against non-stationary targets has been developed by Song et al. [1999]. The work in Ryoo et al. [2005] suggested a generalized form of optimal impact angle control laws with an arbitrary missile system order and practical time-to-go estimation methods for implementation of the proposed guidance law. The concept of the biased PN guidance law for controlling the impact angle is covered in Kim et al. [1998]. They introduced the conventional PN guidance law which combined with a time-varying bias term. A composite guidance law with the terminal impact angle constraint for a surfaceto-surface planar engagement against a stationary target has been investigated based on the properties of the PN guidance (Ratnoo et al. [2008]). The guidance law consists
Copyright by the International Federation of Automatic Control (IFAC)

of a orientation guidance (i.e. PN with N < 2) to cover the desired impact angle and a terminal homing PN guidance with N 2 for achieving the desired impact angle at the terminal time. Later, the composite law was further improved to cope with a non-stationary target in Ratnoo et al. [2010]. Although many guidance laws with only impact angle constraint have been introduced, there are few studies on impact angle and terminal acceleration constraints. In practical missile systems, terminal zero acceleration, which is directly proportional to angle of attack, is sometimes required to eectively destroy armored targets and to improve the robustness against external disturbances or uncertainties. This condition is also needed for warhead eect enhancement. Therefore, terminal zero acceleration constraint as well as impact angle constraint should be considered for guidance systems. Lee et al. [2003] proposed an optimal guidance law with constraints on impact angle and terminal acceleration, but the proposed guidance law may not be suitable for real systems because it is a very complicated form. An optimal impact angle control law for the energy cost weighted by a multiple order of the timeto-go was suggested in Ryoo et al. [2006]. If the guidance gain of the proposed law is greater than 0, the impact angle control command approaches zero. However, there is some limitation on the selection of the guidance gain. In this paper, we propose a generalized impact angle control guidance law with terminal acceleration constraint, which is called Time-to-go Polynomial Guidance (TPG) law. The guidance command of TPG is initially assumed as a polynomial function of time-to-go, and then the coecients of this function are designed to satisfy terminal conditions: zero impact angle error and zero miss-distance. From this procedure, TPG for both impact angle and

3915

Preprints of the 18th IFAC World Congress Milano (Italy) August 28 - September 2, 2011

terminal acceleration constraints can be derived. The guidance law is a simple form, which includes the some impact angle control guidance laws in previous researches. We also obtain closed-form solutions of TPG for a lag-free system by using a linear formulation, and the characteristics of the proposed guidance law are analytically investigated. This paper consists of three sections. In Section 2, the derivation of TPG and its closed-form solutions are discussed. Linear and adjoint simulations are performed to investigate the characteristics of TPG and to demonstrate its validity in Section 3. Finally, conclusions are oered in Section 4. 2. GUIDANCE LAW DESIGN 2.1 Derivation of Guidance Law In this section, the derivation of TPG is covered. The proposed guidance law providing the desired impact angle as well as zero terminal acceleration is derived based on the linearized engagement kinematics as shown in Fig. 1. In this gure, the coordination (XI , YI ) and (Xf , Yf ) represent the inertial reference frame and the impact angle frame, respectively. VM , M and f denote missile velocity, ight path angle, and desired impact angle at the terminal time, respectively. The missile and target are described by the notation of M and T , respectively. Other variables in Fig. 1 are self-explanatory. Under the assumption that the missile velocity is constant, a target is slowly moving relative to the missile maneuver, and heading error (impact angle error), = M f is small, we can obtain the linearized engagement kinematics for the impact angle control problem as follows, y VM = v v = aM where y and v are crossrange and velocity perpendicular to Xf -axis of the impact angle frame, respectively. aM represents the normal acceleration command. The gravity force and autopilot lag are ignored in Eq. (1). The feedback form of TPG command can be derived through two phases. Firstly, we assume that the command

is a polynomial form of tgo as shown in Eq. (2), and then design the guidance command by determining the coecients of this polynomial function to satisfy the predetermined terminal constraints such as the zero impact angle error and terminal zero acceleration constraint. aM (t) = c0 + c1 (tf t) + c2 (tf t) +
2

(2)

In this derivation, two constraints on the terminal acceleration are considered as (1) Constraint 1: terminal acceleration should be zero. (2) Constraint 2: time-derivative of terminal acceleration should be zero. The constraint 1 is required to eectively destroy an armored target and to increase the warhead eect. Because it is hard to accurately satisfy the rst constraint due to time-to-go estimation error, the second constraint is further desirable for a robust guidance law with respect to time-to-go error. In Eq. (2), the coecient c0 and c1 should be zero for the constraints 1 and 2, simultaneously. To control the desired impact angle and zero miss-distance, two boundary conditions y (tf ) = v (tf ) = 0 should be considered. Therefore, the required number of coecients in Eq (2) are two for a unique solution with these boundary conditions. Then, the guidance command can be assumed as aM (t) = cm (tf t) + cn (tf t)
m n

(3)

where n > m 0 and m, n are real numbers. Note that, in case of n > m > 0, the terminal acceleration can be converged to zero; it is corresponding to the constraint 1. Both the constraint 1 and 2 on the terminal acceleration can be achieved using the guidance gains with n > m > 1. For determining the coecients cm and cn , let us substitute the guidance command as shown in Eq. (3) into Eq. (1), and then integrate the resulting linear equation with initial conditions, y (t0 ) , v (t0 ). From this procedure, we can obtain the crossrange and velocity at nal time as go + cm y (tf ) = y (t0 ) + v (t0 ) t
+1 m t go m+1 +2 m t go m+2

(1)

+ cn

+2 n t go n+2

(4) v (tf ) = v (t0 ) + cm + cn

+1 n t go n+1

go = tf t0 and t0 is the initial engagement time. where t From Eq. (4), the coecients cm and cn satisfying the terminal boundary conditions can be obtained as cm = (m + 1) (m + 2) go (n + 2) y (t0 ) + v (t0 ) t +2 m (n m) t go

(5)

(n + 1) (n + 2) go (m + 2) y (t0 ) + v (t0 ) t cn = +2 n (m n) t go By substituting Eq. (5) into Eq. (3), then the guidance command of TPG at initial time t = t0 can be determined as (m + 2) (n + 2) (n + m + 3) y (t0 ) v (t0 ) (6) aM (t0 ) = 2 go t t go
3916

Fig. 1. The homing guidance geometry.

Preprints of the 18th IFAC World Congress Milano (Italy) August 28 - September 2, 2011

Table 1. Examples of TPGs

Name TPG-01 TPG-12 TPG-13 TPG-23 Type m = 0, n = 1 m = 1, n = 2 m = 1, n = 3 m = 2, n = 3 Guidance law 6 aM = t2 y t4 v
go go

Constraints None Only 1 Only 1 1 and 2

After some manipulation, the second fundamental solution is expressed as 1 y2 = tm+2 (13) n m go From Eqs. (10) and (13), the closed-form solution of the crossrange can be written as
+2 y = C1 tn go + C2 +2 tm go nm

aM = t12 2 y
go

6 v tgo 7 v tgo 8 v tgo

aM = t15 2 y
go

aM = t20 2 y
go

(14)

If the coecient cm and cn are initialized and recalculated at each time step, then feedback form of TPG is expressed as (m + 2) (n + 2) (n + m + 3) y (t) v (t) (7) aM (t) = 2 tgo tgo where tgo = tf t. From the above equation, we know that TPG has a simple form and various forms of guidance command according to several combinations of guidance gain m and n. Some examples of TPG are summarized in Table 1. Interesting characteristics of TPG can be observed in Table 1. In the case of m = 0, n = 1, TPG is identical to the optimal impact angle control guidance law for the lagfree system as proposed in Ryoo et al. [2005]. If n = m + 1 and m is a nonnegative integer, then TPG is the same as the time-to-go weighted optimal control guidance law in Ryoo et al. [2006]. TPG, however, does not require the conditions of guidance gain as n = m + 1 or integer values of m and n. In the other words, any nonnegative real values can be used for guidance gains of TPG, and we can form the various guidance types according to m and n. Therefore, it is more general form of impact angle control guidance law than previous studies. 2.2 Closed-form Solution of TPG The closed-form solutions of TPG in the linearized guidance model are derived to give an insight of TPG for designer in this section. When the guidance law Eq. (7) is applied to Eq. (1), the crossrange can be expressed as the second order dierential equation. (m + 2) (n + 2) (m + n + 3) y + y=0 (8) y + tgo t2 go This is one of homogeneous linear ordinary dierential equations and the form of general solution is given as y = C1 y1 + C2 y2 (9)

where C1 = 1 n2 n1 (m + 2) t y (t0 ) + t v (t0 ) go go mn

m2 m 1 C2 = (n + 2) t y (t0 ) + t v (t0 ) go go

(15)

Note that the crossrange solution consists of two distinct time-to-go polynomial terms which depend on the guidance gains. From the rst and second time-derivative of Eq. (14), we can obtain the closed-form solutions of v and aM . For n > m 0, terminal boundary conditions are always satised, however, n > m > 1 condition is required for the satisfaction of the terminal acceleration constraints. If 0 m 1, the terminal acceleration constraint 2 cannot be considered. 3. NUMERICAL ANALYSIS In order to verify the performance of the proposed guidance, two simulations are carried out in this section. The rst simulation is the comparison of the crossrange, heading error and acceleration command proles for TPGs with various guidance gains in linear system. In this case, missile autopilot lag is not considered. Second, terminal error analyses of TPG for a rst-order lag missile system are conducted by using the adjoint technique. m = 2 and n = 3 gains of the guidance law are used for this simulation. 3.1 Comparison of TPGs To investigate the basic properties for various guidance gains, linear simulations with four dierent kinds of gains in Table 1 are performed. we assume that the missile
140 120 100 80 60 40 20 0 20

And then, the second fundamental solution can be determined by using reduction of order technique. y2 = y1 where U= 1 2e y1
1 (m+n+3)t go dt

U dt

(11)

Crossrange y (m)

where the coecients C1 and C2 are constants, which are determined from initial conditions. The rst fundamental solution which satises Eq. (8) can be easily obtained as +2 (10) y1 = tn go

TPG01(m=0,n=1) TPG12(m=1,n=2) TPG13(m=1,n=3) TPG23(m=2,n=3)

10

t (sec)

(12)

Fig. 2. Crossrange prole.

3917

Preprints of the 18th IFAC World Congress Milano (Italy) August 28 - September 2, 2011

5 4

Heading Error (deg)

3 2 1 0 1 2 3 4 5 0 2 4 6

TPG01(m=0,n=1) TPG12(m=1,n=2) TPG13(m=1,n=3) TPG23(m=2,n=3)

Fig. 5. Adjoint loop for y (tf ) of the rst-order lag system with TPG
8 10

t (sec)

Fig. 3. Heading error prole.

20 10

Acceleration aM (m/s )

0 10 20 30 40 50

Fig. 6. Adjoint loop for v (tf ) of the rst-order lag system with TPG
TPG01(m=0,n=1) TPG12(m=1,n=2) TPG13(m=1,n=3) TPG23(m=2,n=3)
Normalized Miss Distance
0 2 4 6 8 10

TPG23 (m=2, n=3)

1

|y(t )|/y
0.9 0.8

f f

0 0

|y(t )|/(*v )

t (sec)

0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

Fig. 4. Acceleration command prole. velocity (VM ) is 300m/s, and the initial crossrange (y (t0 )) and heading error ( (t0 ) = v (t0 ) /VM ) are 100m and 5deg , respectively. The ight time tf is chosen as 10sec. Fig. 2 and 3 represent the crossrange and heading error proles, and the acceleration command prole of TPGs for the lag-free system is given in Fig. 4. Figures show that all of TPGs can achieve the zero impact angle error as well as zero miss-distance, and the crossrange and heading angle rapidly approach to the Xf -axis of the impact angle frame as the guidance gains increase. The guidance command of TPG-01 approaches to a nonzero value because of m = 0, but others (n > m > 0) always converges to 0. It is interesting that the time-derivative of terminal acceleration for TPG-23 (m = 2, n = 3) is also zero since m is greater than 1. 3.2 Adjoint Analysis of TPG-23 Although the TPG-23 has a good performance for the lagfree system, we need to investigate terminal errors of TPG23 due to rst-order lag system for practical applications. Since the impact angle control problem has two terminal boundary conditions (y (tf ) , v (tf )), the adjoint simulation for the terminal miss-distance (y (tf )) and miss-velocity (v (tf )) should be preformed independently in Figs. 5 and 6. In these gures, is time constant of the lag system. From Fig. 7, the miss-distance due to initial crossrange, in normalized form, goes to zero as the ratio of ight time

tf/

10

12

14

16

18

Fig. 7. Normalized terminal miss distance of TPG-23.

TPG23 (m=2, n=3)
40

|v(tf)|/(y0/)
35

|v(t )|/v
f

Normalized Miss Velocity

30

25

20

15

10

tf/

10

12

14

16

18

Fig. 8. Normalized terminal miss velocity of TPG-23. to the lag system time constant becomes large. The missdistance for initial velocity error (i.e. heading error) also converges to zero if the ight time is greater than 14 times

3918

Preprints of the 18th IFAC World Congress Milano (Italy) August 28 - September 2, 2011

. As shown in Fig. 8, The miss-velocity results due to both initial crossrange and velocity error are similar to the adjoint results for miss-distance. Therefore, the terminal errors can be made small when tf is large compared with . This implies that the missile should be launched far from the target for minimizing both terminal miss-distance and heading error, moreover, protection against the command saturation at the vicinity of the interception. 4. CONCLUSION In this paper, we propose the time-to-go polynomial guidance law for impact angle control and terminal zero acceleration. Since the guidance has various forms according to several combinations of guidance gains m and n, it can be regarded as a more general impact angle control guidance law than the previous proposed laws. The closedform solution of TPG for linear model with the lag-free system is derived in order to demonstrate the properties of the guidance law, depending on the choice of m, n. Based on the closed-form solution, the guidance law to satisfy the terminal acceleration constraints can be designed by choosing the arbitrary m with m > 1 condition. Linear and adjoint simulations show the performance and characteristics of the proposed law, and it is nd that the TPG can be applied to practical systems. Several factors such as seekers FOV limit, acceleration limit, and stability of the guidance loop including lag systems should be considered in further work. REFERENCES P. Zarchan. (1994). Tactical and Strategic Missile Guidance. 2nd ed. Washington, DC: AIAA Inc. M. Kim, and K.V. Girder. (1973). Terminal guidance for impact attitude angle constrained ight trajectories. IEEE Trans. on AES, volume AES-9, pages 852859. A.E. Bryson Jr, and Y.C. Ho. (1975). Apllied Optimal Control. New York: Wiley. T.L. Song, S.J. Shin, and H. Cho. (1999). Impact Angle Control for Planar Engagements. IEEE Trans. on AES, volume 35, pages 14391444. C.K. Ryoo, H. Cho, and M.J. Tahk. (2005). Optimal Guidance Laws with Terminal Impact Angle Constraint. Journal of Guidance, Control, and Dynamics, volume 28, pages 724732. C.K. Ryoo, H. Cho, and M.J. Tank. (2006). Timeto-go Weighted Optimal Guidance with Impact Angle Constraints. IEEE Trans. on CST, volume 14, pages 483492. B.S. Kim, J.G. Lee, and H.S. Han. (1998). Biased PNG Law for Impact Angle Constraint. IEEE Trans. on AES, volume 34, pages 277288. Y.I. Lee, C.K. Ryoo, and E. Kim. (2003). Optimal guidance with constraints on impact angle and terminal acceleration. AIAA Guidance, Navigation, and Control Conference, Austin, Texas, Aug. 2003. A. Ratnoo, and D. Ghose. (2008). Impact Angle Constrained Interception of Stationary Targets. Journal of Guidance, Control, and Dynamics, volume 31, pages 18161821. A. Ratnoo, and D. Ghose. (2010). Impact Angle Constrained Guidance Against Nonstaionary Nonmaneuvering Targets. Journal of Guidance, Control, and Dynamics, volume 33, pages 269275.
3919