Unmanned Aerial Vehicle Guidance for an All-Aspect

Approach to a Stationary Target

Group - 15
Abel Viji George AE21B001
Abhigyan Roy AE21B002
Shreeya Padte AE21B056

Indian Institute of Technology

Madras

November 8, 2024

Course Project (AS5570) IIT MADRAS November 8, 2024 1 / 29

Introduction
This study focuses on developing a multiphase guidance algorithm for a
UAV in the nonlinear engagement framework that enables an aerial vehicle
to approach a stationary target from any direction. The challenges faced
by the standard and two-phase Pure Proportional Navigation, mainly
approach angle control are addressed, providing solutions that take into
account the restrictions imposed by the seeker’s field of view (FOV).

Figure 1: Possible terminal heading of UAV to a stationary target

Course Project (AS5570) IIT MADRAS November 8, 2024 2 / 29

Background

1 Importance of controlling approach angles for effective UAV

operations.
Necessity of controlling terminal path angles includes requirements like
efficient way-point navigation, achieving rendezvous, and other desired
formations following a trajectory with sensing.
Controlling impact angle also becomes important in successful
interception against stationary or slowly moving targets, such as
aircraft carrier warships with efficient electronics countermeasures and
a close-in weapon system.

2 Previous studies and their limitations.

Course Project (AS5570) IIT MADRAS November 8, 2024 3 / 29

Previous Studies on Impact Angle-Constrained Guidance
Problem

Linearized Geometry-Based :

Work Main Focus

1 Optimal-angle-based control guidance.
2 Sliding-mode-control-based guidance.
3 Zero-effort collision triangle was incorporated in a linear optimal
guidance problem.

Limitation : The optimal-control-based approaches rely on linearization of the

engagement problems and are sensitive to the time-to-go estimation error, while
the sliding-mode-based approaches have limitations in dealing with scenarios with
a relatively large initial heading error causing a positive initial range rate.
1
Idan, M., Golan, O., and Guelman, M., “Optimal Planar Interception with Terminal Constraints,” *Journal of Guidance,
Control, and Dynamics*, Vol. 18, No. 6, Nov.–Dec. 1995, pp. 1273–1279. doi:10.2514/3.21541
2
Kumar, S. R., and Ghose, D., “Three-Dimensional Impact Angle Guidance with Coupled Engagement Dynamics,” *Journal
of Aerospace Engineering*, Vol. 231, No. 4, April 2017, pp. 621–641. doi:10.1177/0954410016641442
3
Cho, H., Ryoo, C. K., Tsourdos, A., and White, B., “Optimal Impact Angle Control Guidance Law Based on Linearization
About Collision Triangle,” *Journal of Guidance, Control, and Dynamics*, Vol. 37, No. 3, May–June 2014, pp. 958–964.
doi:10.2514/1.62910
Course Project (AS5570) IIT MADRAS November 8, 2024 4 / 29
Previous Studies contd.

PN-Based :

Work Main Focus

4 Time-varying biased pure PN.
5, 6 Two-stage planar pure PN (2pPPN) for stationary and lower-speed
targets.
7 Composite PN-based guidance for higher-speed targets.

Limitation : They considered the impact-angle control for two-dimensional

engagements and did not consider the FOV constraints.

4
Kim, B. S., Lee, J. G., and Han, H. S., “Biased PNG Law for Impact with Angular Constraint,” *IEEE Transactions on
Aerospace and Electronic Systems*, Vol. 34, No. 1, Jan. 1998, pp. 277–288. doi:10.1109/7.640285
5
Ratnoo, A., and Ghose, D., “Impact Angle Constrained Interception of Stationary Targets,” *Journal of Guidance, Control,
and Dynamics*, Vol. 31, No. 6, Nov.–Dec. 2008, pp. 1817–1822. doi:10.2514/1.37864
6
Ratnoo, A., and Ghose, D., “Impact Angle Constrained Guidance Against Nonstationary Nonmaneuvering Targets,”
*Journal of Guidance, Control, and Dynamics*, Vol. 33, No. 1, Jan.–Feb. 2010, pp. 269–275. doi:10.2514/1.45026
7
Ghosh, S., Ghose, D., and Raha, S., “Composite Guidance for Impact Angle Control Against Higher Speed Targets,”
*Journal of Guidance, Control, and Dynamics*, Vol. 39, No. 1, Jan. 2016, pp. 98–117. doi:10.2514/1.G001232
Course Project (AS5570) IIT MADRAS November 8, 2024 5 / 29
2-D Engagement Geometry

Figure 2: 2-D engagement geometry and UAV’s terminal heading.

Engagement Equations for Stationary Target

Ṙ = VR = −VP cos(αP − θ)
Vθ
θ̇ = = −VP sin(αP − θ)
R
aP
α˙P =
VP
The UAV’s look angle is defined as µ = αP − θ
Course Project (AS5570) IIT MADRAS November 8, 2024 6 / 29
PPN
The UAV’s lateral acceleration is based on the pure PN (PPN)
guidance law with the navigation gain N given by,
aP = VP α˙P = NVP θ̇ (1)

Course Project (AS5570) IIT MADRAS November 8, 2024 7 / 29

PPN
The UAV’s lateral acceleration is based on the pure PN (PPN)
guidance law with the navigation gain N given by,
aP = VP α˙P = NVP θ̇ (1)

Consider αP0 ∈ [θ0 , θ0 + π]. In this case, from above equations, it

follows that the achievable approach angle using the standard PPN
guidance is given by
αPf = αP0 + N(θf − θ0 )
A collision course with a stationary target assumes αPf = θf
(Nθ0 − αP0 )
αPf =
(N − 1)

Course Project (AS5570) IIT MADRAS November 8, 2024 7 / 29

PPN
The UAV’s lateral acceleration is based on the pure PN (PPN)
guidance law with the navigation gain N given by,
aP = VP α˙P = NVP θ̇ (1)

Consider αP0 ∈ [θ0 , θ0 + π]. In this case, from above equations, it

follows that the achievable approach angle using the standard PPN
guidance is given by
αPf = αP0 + N(θf − θ0 )
A collision course with a stationary target assumes αPf = θf
(Nθ0 − αP0 )
αPf =
(N − 1)

For N ≥ 2 (which is a condition for the bounded terminal lateral

acceleration of the UAV), the achievable approach angle using PPN
resides in the interval αPf ∈ [2θ0 − αP0 , θ0 ].
Course Project (AS5570) IIT MADRAS November 8, 2024 7 / 29
2pPPN

However, [2θ0 − αP0 , θ0 ) ⊂ [−π + θ0 , θ0 ) which implies that using the

standard PPN, a significant portion of the angular interval in the
half-space [−π + θ0 , θ0 ) cannot be achieved. To expand the
achievable approach angle set, a two-stage PPN guidance strategy
(2pPPN) was introduced.

Course Project (AS5570) IIT MADRAS November 8, 2024 8 / 29

2pPPN

However, [2θ0 − αP0 , θ0 ) ⊂ [−π + θ0 , θ0 ) which implies that using the

standard PPN, a significant portion of the angular interval in the
half-space [−π + θ0 , θ0 ) cannot be achieved. To expand the
achievable approach angle set, a two-stage PPN guidance strategy
(2pPPN) was introduced.
For impact angles outside [2θ0 − αP0 , θ0 ), an orientation guidance for
the initial phase of missile flight was proposed. The purpose of
orientation guidance is to take the missile from (αP0 , θ0 ) to
(αPORIf = θ0 , θORIf = θ0 − π/2). Thus, the impact band covers
αPf ∈ [−π + θ0 , θ0 ].
The gain for this phase is:
θ0 − αP0 2
NORI = = |αP0 − θ0 |
θ0 − (θ0 − π/2) π

Course Project (AS5570) IIT MADRAS November 8, 2024 8 / 29

2pPPN

Theorem 1: Consider αP0 > θ0 . A desired approach angle αPd f

αdP −αP
∈ [2θ0 − αP0 , θ0 ) could be attained by PPN with N = f
αdP −θ
≥ 2,
f
whereas αPd f ∈ [−π + θ0 , 2θ0 − αP0 ) could be achieved by 2pPPN with

αdP −αP

 π2 (αP0 − θ0 );
 if f
αdP −θ
<2
N= d
α −α
f
αdP −αP
 Pdf P ;
 if f
≥2
αP −θ αdP −θ
f f

Examples of trajectories using PPN for αPd f = -π/6 and 2pPPN for αPd f = -5π/6
are presented, with and without the look-angle constraint.

Course Project (AS5570) IIT MADRAS November 8, 2024 9 / 29

2pPPN
Observation 1: Consider αP0 < θ0 . A desired approach angle αPd f
αdP −αP
∈ (θ0 , 2θ0 − αP0 ] could be attained by PPN with N = f
αdP −θ
≥ 2,
f
whereas αPd f ∈ (2θ0 − αP0 , −π + θ0 ] could be achieved by 2pPPN with
αd −αP

 π2 |(αP0 − θ0 )| if αPdf −θ < 2

Pf
N = αd −α αdP −αP
Pf P

 d
α −θ
if αdf −θ ≥ 2
Pf Pf

Observation 2: If αP0 = θ0 or αP0 = −π + θ0 , then before applying

Theorem 1 and Observation 1, an engagement geometry (αP , θ) needs to
be attained outside the collision course or inverse collision course by using
a finite adjustment pulse bias Badj satisfying
(
−sgn(αPd f − θ0 ) if αP0 = θ0
sgn(Badj ) =
sgn(αPd f − θ0 ) if αP0 = −π + θ0

Course Project (AS5570) IIT MADRAS November 8, 2024 10 / 29

Three-Phase Composite Guidance

To overcome the limit of angular spectrum for achievable terminal

angles, the reversal of the LOS rate is found to be crucial. This can
be facilitated by adding a finite bias profile ”Bsgn(θ˙0 )” to the PPN
guidance command for a certain time interval (Biased PPN or BPPN).

Course Project (AS5570) IIT MADRAS November 8, 2024 11 / 29

Three-Phase Composite Guidance

To overcome the limit of angular spectrum for achievable terminal

angles, the reversal of the LOS rate is found to be crucial. This can
be facilitated by adding a finite bias profile ”Bsgn(θ˙0 )” to the PPN
guidance command for a certain time interval (Biased PPN or BPPN).
For PPN, µ̇ = (N − 1)θ̇ = −(N − 1)VP sin(µ)/R. Hence, depending
on the initial value of µ, it monotonically increases or decreases to 0,
which implies that θ̇ is sign-preserving.
Proof:
aP = NVP θ̇ + Bsgn(θ˙0 )
µ̇ = −(N − 1)VP sinµ/R + (B/VP )sgn(θ˙0 ).
From eqn of θ̇, θ˙0 decreases (increases) if µ0 > 0 (µ0 < 0). Hence,
d|µ|
dt < −(N − 1)VP |sinµ|/R. So, |µ| goes to zero faster than R and
hence, LOS rate changes sign.

Course Project (AS5570) IIT MADRAS November 8, 2024 11 / 29

αCIG Law and Algorithm:
Assume that the initial conditions are such that the desired approach angle
is not attainable by the PPN or the 2pPPN guidance laws alone. In that
case,
1 In phase-1, use BPPN guidance, aαCIG = NV θ̇ + Bsgn(θ˙ ) with
P P 0
N = 2 and B > 0 for an interval from t = 0 to t = t2 . t2 is the time
at which θ̇ changes sign. θ(t2 ) = θ2 , αP (t2 ) = αP2 .

Course Project (AS5570) IIT MADRAS November 8, 2024 12 / 29

αCIG Law and Algorithm:
Assume that the initial conditions are such that the desired approach angle
is not attainable by the PPN or the 2pPPN guidance laws alone. In that
case,
1 In phase-1, use BPPN guidance, aαCIG = NV θ̇ + Bsgn(θ˙ ) with
P P 0
N = 2 and B > 0 for an interval from t = 0 to t = t2 . t2 is the time
at which θ̇ changes sign. θ(t2 ) = θ2 , αP (t2 ) = αP2 .
2 In phase-2, use BPPN guidance again with N < 1 and B > 0 till a

time t = t3 . θ(t3 ) = θ3 such that θ3 ∈ [θ2 , θ0 ] or θ3 ∈ [θ0 , θ2 ] and

d ∈ [θ , θ + π] or αd ∈ [−π + θ , θ ] for α
αPf 0 3 Pf 3 0 P0 > θ0 or
αP0 < θ0 , respectively.

Course Project (AS5570) IIT MADRAS November 8, 2024 12 / 29

αCIG Law and Algorithm:
Assume that the initial conditions are such that the desired approach angle
is not attainable by the PPN or the 2pPPN guidance laws alone. In that
case,
1 In phase-1, use BPPN guidance, aαCIG = NV θ̇ + Bsgn(θ˙ ) with
P P 0
N = 2 and B > 0 for an interval from t = 0 to t = t2 . t2 is the time
at which θ̇ changes sign. θ(t2 ) = θ2 , αP (t2 ) = αP2 .
2 In phase-2, use BPPN guidance again with N < 1 and B > 0 till a

time t = t3 . θ(t3 ) = θ3 such that θ3 ∈ [θ2 , θ0 ] or θ3 ∈ [θ0 , θ2 ] and

d ∈ [θ , θ + π] or αd ∈ [−π + θ , θ ] for α
αPf 0 3 Pf 3 0 P0 > θ0 or
αP0 < θ0 , respectively.
3 In phase-3, use 2pPPN guidance law a
αCIG = NVP θ̇ with navigation
2
gain N defined as N = π |αP3 − θ3 | as long as
d − α )/(αd − θ) < 2; otherwise, set
(αPf P Pf
N = (αPfd − α )/(αd − θ). This applies if αd lies within
P Pf Pf
[2θ3 − αP3 , π + θ3 ] or [−π + θ3 , 2θ3 − αP3 ] for initial conditions
αP0 > θ0 or αP0 < θ0 , respectively.
Course Project (AS5570) IIT MADRAS November 8, 2024 12 / 29
µαCIG Law and Algorithm:

Similar to αCIG Law, but here we have the added constraint of the look
angle, which adds a few more additional steps. Given αP0 and αPd f ,
µαCIG
1 Use BPPN, a
P = NVP θ̇ + Bsgn(θ˙0 ) with N = 2 and B finite and
positive until time t1 , at which |µ| becomes just less than µs if
|µ0 | = |αP0 − θ0 | ∈ [µs , π]. Let αP (t1 ) = αP1 and θ(t1 ) = θ1 . Else, if
|µ0 | ∈ [0, µs ], then t1 = t0 , αP (t1 ) = αP1 = αP0 , and
θ(t1 ) = θ1 = θ0 .

Course Project (AS5570) IIT MADRAS November 8, 2024 13 / 29

µαCIG Law and Algorithm:

Similar to αCIG Law, but here we have the added constraint of the look
angle, which adds a few more additional steps. Given αP0 and αPd f ,
µαCIG
1 Use BPPN, a
P = NVP θ̇ + Bsgn(θ˙0 ) with N = 2 and B finite and
positive until time t1 , at which |µ| becomes just less than µs if
|µ0 | = |αP0 − θ0 | ∈ [µs , π]. Let αP (t1 ) = αP1 and θ(t1 ) = θ1 . Else, if
|µ0 | ∈ [0, µs ], then t1 = t0 , αP (t1 ) = αP1 = αP0 , and
θ(t1 ) = θ1 = θ0 .
µαCIG
2 In phase-1, use BPPN guidance, a
P = NVP θ̇ + Bsgn(θ˙0 ) with
N = 2 and B > 0 for an interval from t = 0 to t = t2 . t2 is the time
at which θ̇ changes sign. θ(t2 ) = θ2 , αP (t2 ) = αP2 .

Course Project (AS5570) IIT MADRAS November 8, 2024 13 / 29

µαCIG Law and Algorithm:
3 In the phase-2, use the look-angle-constrained bias-shaped BPPN guidance
aPµαCIG = NVP θ̇ + Bsgn(θ1 ) with N < 1 and B > 0 at time t ≥ t2 as long as,
for αP1 > θ1 , θ < αPd f and
Z t
(N − 1)VP (θ − θ2 ) < B dt < (N − 1)VP (θ − θ2 ) + VP µs ;
t2

or for αP1 < θ1 , θ > αPd f and

Z t
(N − 1)VP (θ2 − θ) < B dt < (N − 1)VP (θ2 − θ) + VP µs .
t2

upto a time t̂. θ(t̂) = θ̂ and αP (t̂) = αˆP .

Course Project (AS5570) IIT MADRAS November 8, 2024 14 / 29

µαCIG Law and Algorithm:
3 In the phase-2, use the look-angle-constrained bias-shaped BPPN guidance
aPµαCIG = NVP θ̇ + Bsgn(θ1 ) with N < 1 and B > 0 at time t ≥ t2 as long as,
for αP1 > θ1 , θ < αPd f and
Z t
(N − 1)VP (θ − θ2 ) < B dt < (N − 1)VP (θ − θ2 ) + VP µs ;
t2

or for αP1 < θ1 , θ > αPd f and

Z t
(N − 1)VP (θ2 − θ) < B dt < (N − 1)VP (θ2 − θ) + VP µs .
t2

upto a time t̂. θ(t̂) = θ̂ and αP (t̂) = αˆP .

If αPd f ∈ (θ̂, θ̂ + π] or αPd f ∈ [−π + θ̂, θ̂) for αP1 > θ1 or αP1 < θ1 ,
respectively, then t3 = t̂, αP3 = α̂P , and θ3 = θ̂. Begin the third phase as
specified in Step 4.
Continue with the look-angle-constrained BPPN guidance
aPµαCIG = NVP θ̇ + Bsgn(θ˙1 ) with N = 1 and B = 0 until θ = θ3 at some
time t3 such that αPd f ∈ (θ3 , θ3 + π] or αPd f ∈ [−π + θ3 , θ3 ), if θ̂ < −π + αPd f
or θ̂ > π + αPd f for αP1 > θ1 or αP1 < θ1 , respectively, let αP (t3 ) = αP3 . At
time t = t3 , start the third phase as in Step 4.
Course Project (AS5570) IIT MADRAS November 8, 2024 14 / 29
µαCIG Law and Algorithm:

4 In phase-3,
Use PPN guidance with aPµαCIG = NVP θ̇, where
N = (αPd f − αP3 )/(αPd f − θ3 ), if αPd f ∈ (θ3 , 2θ3 − αP3 ] or
αPd f ∈ [2θ3 − αP3 , θ3 ) for αP1 > θ1 or αP1 < θ1 , respectively.
Use 2pPPN guidance law aPµαCIG with navigation gain N defined as
N = π2 |αP3 − θ3 | as long as (αPf d
− αP )/(αPf d
− θ) < 2; otherwise, set
d d d
N = (αPf − αP )/(αPf − θ). This applies if αPf lies within
(2θ3 − αP3 , π + θ3 ] or [−π + θ3 , 2θ3 − αP3 ) for initial conditions
αP1 > θ1 or αP1 < θ1 , respectively.

Course Project (AS5570) IIT MADRAS November 8, 2024 15 / 29

Simulation Setup and Results

All the results presented were simulated in MATLAB.

The target is stationary and in line with the pursuer initially (θ0 = 0)
The initial angle of the pursuer (αP0 ) is 45◦ for 2pPPN and 135◦ for
the Three-phase Composite guidance, αCIG and µαCIG laws.

First phase Second phase

αPd f , rad Guidance
N B t2 , s N B t3 , s θ3 , rad
0 αCIG 2 VP |αP |/5 27.09 -2 VP |αP |/10 37.82 -0.071
0 µαCIG 2 VP |αP |/5 27.29 -2 VP |αP |/10 32.57 -0.213
π/3 αCIG 2 VP |αP |/5 27.09 -2 VP |αP |/10 37.82 -0.071
π/3 µαCIG 2 VP |αP |/5 27.29 -2 VP |αP |/10 32.57 -0.213
5π/6 αCIG 2 VP |αP |/5 27.09 -2 VP |αP |/10 37.82 -0.071
5π/6 µαCIG 2 VP |αP |/5 27.29 -2 VP |αP |/10 32.57 -0.213

Course Project (AS5570) IIT MADRAS November 8, 2024 16 / 29

Trajectory Analysis and Look Angle for Two Phase PPN
(paper)

Figure 3: UAV’s trajectory for PPN and

2pPPN for αdpf = −30◦ and −150◦ . Figure 4: Variation of Look-Angle with time.

Course Project (AS5570) IIT MADRAS November 8, 2024 17 / 29

Trajectory Analysis and Look Angle for Two Phase PPN
(code)

Figure 5: UAV’s trajectory for PPN and

2pPPN for αdpf = −30◦ and −150◦ . Figure 6: Variation of Look-Angle with time.

Course Project (AS5570) IIT MADRAS November 8, 2024 18 / 29

Example of three-phase composite guidance

Figure 7: UAV’s trajectory for αdpf = 150◦ . Figure 8: UAV’s trajectory for αdpf = 150◦ .
(Paper) (Code)

Course Project (AS5570) IIT MADRAS November 8, 2024 19 / 29

2-D engagement with αPd f = 0 (paper)

Figure 9: UAV’s Trajectory Figure 10: Look-Angle over Time

Figure 11: Lateral Acceleration Requirement Figure 12: Bias Variation over Time

Course Project (AS5570) IIT MADRAS November 8, 2024 20 / 29

2-D engagement with αPd f = 0 (code)

Figure 13: UAV’s Trajectory Figure 14: Look-Angle over Time

Figure 15: Lateral Acceleration Requirement Figure 16: Bias Variation over Time
Course Project (AS5570) IIT MADRAS November 8, 2024 21 / 29
2-D engagement with αPd f = π/3 (paper)

Figure 17: UAV’s Trajectory Figure 18: Look-Angle over Time

Figure 19: Lateral Acceleration Requirement Figure 20: Bias Variation over Time
Course Project (AS5570) IIT MADRAS November 8, 2024 22 / 29
2-D engagement with αPd f = π/3 (code)

Figure 21: UAV’s Trajectory Figure 22: Look-Angle over Time

Figure 23: Lateral Acceleration Requirement Figure 24: Bias Variation over Time
Course Project (AS5570) IIT MADRAS November 8, 2024 23 / 29
2-D engagement with αPd f = 5π/6 (paper)

Figure 25: UAV’s Trajectory Figure 26: Look-Angle over Time

Figure 27: Lateral Acceleration Requirement Figure 28: Bias Variation over Time
Course Project (AS5570) IIT MADRAS November 8, 2024 24 / 29
2-D engagement with αPd f = 5π/6 (code)

Figure 29: UAV’s Trajectory Figure 30: Look-Angle over Time

Figure 31: Lateral Acceleration Requirement Figure 32: Bias Variation over Time
Course Project (AS5570) IIT MADRAS November 8, 2024 25 / 29
 αCIG and µαCIG with varied constant Bias profiles
(paper)

Course Project (AS5570) IIT MADRAS November 8, 2024 26 / 29

αCIG and µαCIG with varied constant Bias profiles (code)

Figure 33: αCIG Trajectories Figure 34: αCIG Bias Profiles

Figure 35: µαCIG Trajectories Figure 36: µαCIG Bias Profiles

Course Project (AS5570) IIT MADRAS November 8, 2024 27 / 29
Summary

As seen, implementing the µαCIG Law , a full circle approach angle

set [−π, π) can be achieved while satisfying the FOV constraint which
gives a bias-shaped BPPN guidance .
Implications for UAV operations in real-world scenarios and Potential
applications of developed guidance strategies.
Precision strikes and surveillance missions
Infrastructure Inspection
Search and Rescue Missions

Course Project (AS5570) IIT MADRAS November 8, 2024 28 / 29

Future Work

Extensions
Furthering the work in this paper, we can implement a modified law
for non-stationary targets, starting with constant velocity targets and
extending to maneuvering targets.
Also, we can add environmental conditions like wind velocity or
turbulence like drag as variables in the engagement equations of the
UAV.
The paper also has implemented the same for 3D engagements and SITL simulations, which we
have not discussed due to time constraints.

Course Project (AS5570) IIT MADRAS November 8, 2024 29 / 29