Robotics and Autonomous Systems 98 (2017) 158–173

Contents lists available at ScienceDirect

Robotics and Autonomous Systems

journal homepage: www.elsevier.com/locate/robot

Two-layer obstacle collision avoidance with machine learning for

more energy-efficient unmanned aircraft trajectories
Youngjun Choi *,1 , Hernando Jimenez 1 , Dimitri N. Mavris 2
Georgia Institute of Technology, Atlanta, GA, 30332-0150, United States

highlights

• Novel optimal obstacle avoidance algorithm with two-layer structure.

• Global-path optimization resolves multiple obstacles via a clustering algorithm and specifies a flight path.
• Local-path optimization is defined as a multi-phase optimal trajectory problem.
• The proposed two-phase collision avoidance algorithm is demonstrated and shown to yield more energy-efficient collision avoidance trajectories in a
simulated multi-obstacle environment when compared to a traditional one-layer obstacle algorithm.

article info a b s t r a c t
Article history: This paper proposes a new two-layer obstacle avoidance algorithm that allows an unmanned aircraft
Received 6 October 2015 system to avoid multiple obstacles with minimal effort. The algorithm includes a global-path optimization
Received in revised form 6 July 2017 that identifies the number of obstacles resulting from a clustering technique based on obstacle infor-
Accepted 3 September 2017
mation from an airborne sensor, and specifies a potential threat. A local-path trajectory optimization
Available online 18 September 2017
employs a model predictive control structure based on a multi-phase optimal trajectory resulting from
approximated dynamics, vehicle constraints, and the result of the global-path optimization. Numerical
Keywords:
Obstacle avoidance flight simulations are conducted with a conventional one-layer obstacle avoidance algorithm and the
Optimal trajectory two-layer obstacle avoidance algorithm. The results of the numerical simulation show that the proposed
Clustering algorithm two-layer optimal obstacle avoidance algorithm generates more energy-efficient avoidance trajectories
Model predictive control when an unmanned aircraft meets multiple obstacles.
UAV © 2017 Elsevier B.V. All rights reserved.
Path-planning

1. Introduction The current separation assurance paradigm for unmanned air-

craft builds upon that for manned traffic as a multi-layered frame-
Unmanned aircraft and their integration into the airspace sys- work with overlapping capabilities that include operating pro-
tem have gained significant attention and popularity given the eco- cedures, strategic separation, tactical separation, self separation,
nomic potential stemming from diverse applications, such as par- and collision avoidance, following progressively shorter time hori-
cel delivery, support of rescue missions, and surveillance missions zons [2,3]. Strategic separation, for instance, is exercised by air traf-
using various platforms. However, the prospect of fully integrated
fic management services and addresses conflict resolution for time
airspace operations is limited by technological and regulatory chal-
horizons roughly between 3 min and 10 min. Collision avoidance
lenges, including separation assurance, human system integration
and communication. Among core challenges, detect and avoid has is realized by on-board systems for conflict resolution within ap-
been recognized as a key enabler for separation assurance by key proximately 15 s. The detect and avoid capability of an unmanned
stakeholders such as the Federal Aviation Administration (FAA) [1]. aircraft is a key enabler for self-separation and collision avoidance,
and it contributes to top-level operational capabilities in fully
integrated operations given its coupling with other critical ele-
* Corresponding author.
E-mail addresses: ychoi95@gatech.edu (Y. Choi), ments, such as sensing/surveillance, communications, functional
hernando.jimenez@asdl.gatech.edu (H. Jimenez), allocation, and human factors.
dimitri.mavris@aerospace.gatech.edu (D.N. Mavris).
1 Research Engineer, School of Aerospace Engineering, Georgia Institute of There are numerous studies investigating collision avoidance
Technology.
concepts and techniques, which can be generally divided into
2 S.P. Langley Distinguished Regents Professor, School of Aerospace Engineering, five groups: stochastic methods, road map methods, potential
Director Aerospace Systems Design Laboratory, Georgia Institute of Technology. field approaches, geometric methods, and optimization-based

http://dx.doi.org/10.1016/j.robot.2017.09.004
0921-8890/© 2017 Elsevier B.V. All rights reserved.
 Y. Choi et al. / Robotics and Autonomous Systems 98 (2017) 158–173 159

methods. Stochastic methods effectively search nonconvex high- detected obstacle information from an airborne sensor, an on-
dimensional spaces for global or local obstacle avoidance paths board system identifies the number of obstacles through a cluster-
based on the environment as perceived by airborne sensors. The ing technique. This clustering routine solves a distance-based con-
rapid random tree (RRT) [4,5] is a popular method of this kind. Road straint optimization problem that includes minimum-separation
map methods use visual graph characteristics and path-planning distance between pairs of adjacent obstacles. Then, based on the
algorithms to partition collision-free paths and create a piecewise cluster results and the velocity information, the on-board system
linear path or curved path using smoothing techniques [6]. The judges a potential threat among the identified multiple obstacles.
Theta∗ algorithm has recently been introduced to overcome the The clustering algorithm has been successfully applied to resolve
restricted heading angle issue of the classical A∗ algorithm to solve multiple obstacles as separate objects in the field of ground mobile
the 3D path-planning problem in an urban environment [7]. robots [24]. However, the UAV collision avoidance problem is a
The fundamental concept of artificial potential field methods higher dimensional clustering problem that requires more com-
is the creation of a force map where a waypoint generates an putational resources to handle three-dimensional airborne LiDAR
attractive force and obstacles generate repulsive forces. Based on information. Therefore, this paper also explores diverse techniques
the result of the force map, an algorithm generates an optimal and compares them to achieve a more computationally favorable
collision-free path. The artificial potential field has been intro- approach. In the local-path optimization, we employ a multi-phase
duced by Khatib et al. to solve an obstacle avoidance problem for optimal obstacle avoidance problem.
robotics applications [8]. This approach has the benefits of a simple Many clustering techniques have been introduced because of
structure and ease of implementation but also has some impor- their simplicity and scalability for a broad spectrum of applications.
tant shortcomings, such as local minima where the total force is The k-means algorithm, one of the most popular and classical
close to zero and non-reachable waypoints due to proximity to clustering methods, partitions a data set into k clusters by sequen-
obstacles [9–11]. Improvements to address these issues have been tially assigning points to clusters based on their proximity and
proposed, for instance in work by Li et al. [12] and Mac et al. [11], updating the cluster definitions to reflect said point assignment.
by introducing additional force terms and condition exceptions. However, the k-means algorithm requires the specification of the
Among geometric methods, one approach uses the point of number of clusters k a priori. To circumvent this issue, several
closest approach for a collision avoidance problem against a mov- approaches have been proposed, for instance, by evaluating some
ing airplane. This technique specifies a conflict condition based top-level partition quality metric for which said metric is optimal
on the vehicles’ geometric relationship, and the two aircraft ma- (see for example Ref. [25]). These approaches often fail to provide
sufficiently tractable results and are computationally expensive.
neuver away from the conflict region through a vector sharing
Jia et al. posit that if the point cloud of each group has a non-
resolution [13]. Another geometric approach utilizes a collision
convex shape, these clustering algorithms yield results that only
cone to yield a closed-form solution for an aiming point [14]. The
guarantee a local minimum [26]. Another clustering algorithm is
collision cone approach has some inherent shortcomings for which
the NJW algorithm proposed by Ng et al. as a classical spectral
solutions have been developed in more recent work [15].
clustering method, which is computationally tractable [27]. In this
Optimization-based and geometric methods are closely re-
approach, the number of clusters k can be determined from the
lated, and have presented a range of promising solutions. For
top k eigenvalues of the Laplacian matrix. However, in that case,
instance, Chakravarthy and Ghose proposed a collision cone tech-
determining the number k of clusters does not have a tractable
nique to avoid a moving obstacle with an irregular shape in
standard, and the top k eigenvectors do not always produce correct
two-dimensional space [16]. Watanabe et al. extended this ap-
clustering results [26].
proach with a minimal-effort optimization framework in three-
For the obstacle avoidance trajectory problem, the number of
dimensional space [17]. Another example is that of Schouwe-
clusters k cannot be user-specified. Moreover, resolution of clus-
naars et al. [18] who suggested mixed-integer linear programming
ters as separate objects must be defined according to an explicitly-
(MILP) that incorporates binary constraints based on the area infor- defined minimum separation distance. To solve the issue of the
mation outside of an obstacle. Yoshiaki et al. [19] propose a more NJW algorithm, the number of clusters k can be determined from k-
precise MILP algorithm for a three-dimensional collision avoid- block diagonality of the Laplacian matrix [28]. Another clustering
ance problem that includes the construction of a coarse cost map technique that does not need any a priori information of k is DB-
and a detailed trajectory optimization. These MILP frameworks SCAN, proposed Ester et al. [29], a density-based spatial clustering
have been employed for a real-time collision avoidance problem algorithm. In DBSCAN, the number of clusters is not defined a priori
in diverse platforms [20]. However, solving MILP is commonly but rather results from the clustering process itself, driven by the
implemented via the branch and bound method, which is compu- grouping of points that meet a user-defined minimum density.
tationally expensive [20,21]. Another optimization-based method, For the clustering algorithm in the global-path optimization, we
proposed by Moon et al. [22], solves for an optimal avoidance select two clustering algorithms as candidate solutions and com-
trajectory based on a safe position around an obstacle determined pare their performance with respect to clustering accuracy and
from on-board sensor information and a flight envelope protection computational efficiency.
function. Concurrent detection of multiple proximate obstacles For the online trajectory optimization, we employ a model
results in them being mishandled as a single obstacle, even when predictive control (MPC) scheme that produces computationally
there is sufficient clearance to fly safely between them, resulting feasible solutions. That is, solving actual nonlinear constrained
in highly inefficient trajectories. Kang et al. [23] expanded this dynamics trajectory optimization is an NP-hard problem, but MPC
framework to by incorporating external surveillance sources with solves approximated linear dynamics and simplified constraints,
obstacle information. Although some improvements are observed, and regularly updates the results of the optimal trajectory in a
this concept is vulnerable to surveillance data loss and essentially real-time manner. This MPC approach provides a computationally
reduces to the same local-optimization approach it seeks to im- tractable solution for an online optimal collision avoidance prob-
prove upon. In this paper we address the question of how an lem and has been successfully implemented in the past [22,23,30].
optimal collision avoidance algorithm can produce highly energy- We hypothesize that our two-layer collision avoidance algo-
efficient trajectories in a multi-obstacle environment while relying rithm yields more energy efficient trajectories without incurring
solely on on-board sensors and capabilities. a prohibitive computational burden relative to the single-layer
This work proposes a two-layer obstacle collision avoidance approach. In the remainder of this paper, we first present the
algorithm that incorporates a global-path optimization and a local- formulation and implementation of our algorithm, and then test
path optimization. In the global-path optimization, based on the the hypothesis via direct comparison of numerical simulations.
160 Y. Choi et al. / Robotics and Autonomous Systems 98 (2017) 158–173

2. New path planning architecture using a learning algorithm 2.3. Global trajectory optimization

2.1. Two-level algorithm concept 2.3.1. Obstacle resolution

The first step in the identification of an optimal global trajectory
The multi-obstacle avoidance problem is inherently challeng- is the resolution of multiple obstacles ahead of the aircraft’s current
ing. In general, the mathematical complexity and computational flight path. This function entails a suitable characterization of
expense of an optimal trajectory solution grows quickly with the obstacles based on their location relative to the aircraft and each
explicit treatment of multiple obstacle constraints. It follows that other, as detected by the on-board sensor. We adopt a general sen-
most optimal trajectory frameworks in recent work do not explic- sor model that captures the fundamental mechanism underlying
itly consider multi-obstacle scenarios [22,23,31]. Instead, existing many such instruments. The model features an array of distance-
algorithms employ the collision avoidance framework for one ob- measuring rays originating at the sensor, uniformly structured
stacle regardless of the number of downstream trajectory threats, over the horizontal and vertical field of view. We assume the field
producing inefficient trajectories whenever safe flight between of view is symmetric about the x-axis in the vehicle coordinate
obstacles is feasible. To address this major shortcoming, the new system GV and forward looking. The distance to an external body
algorithm proposed here features a two-layer structure accommo- along any ray is known whenever the ray intersects a surface of that
dating online global-path and local-path optimization processes, body, unless it exceeds a detection range specified in the model.
illustrated in Fig. 1. The relative angular placement of all sensor rays within the field of
The global-path optimization uses on-board sensor data to ef- view is prescribed so that the angle of each the ray in the vehicle
ficiently resolve multiple downstream obstacles based on their coordinate system is known. Accordingly, with measured distances
relative location and separation. It identifies the potential threat along known angles in GV the sensor data array provides the
among the resolved obstacles. The local path optimizer produces location of all detection points. Sensor capabilities may be defined
an optimal collision avoidance trajectory using the information of by adjusting the vertical and horizontal field of view, density of
the potential threat. To do so, it solves a multi-phase optimal tra- distance-measuring rays, and the detection range. We assume that
jectory problem based on vehicle dynamics, constraints, obstacle instrument error and signal return time are negligible, so that
information, and mission waypoints. Two fundamental phases are the model is deterministic and instantaneous. We also assume
defined for this problem: obstacle avoidance and recovery. In the the detection range is isometric so that no variation exists along
obstacle avoidance phase, the terminal trajectory point is an in- vertical or horizontal field of view directions.
termediate waypoint that satisfies three linear constraints around Obstacle resolution is hence realized as an operation on the
the obstacle, one above and one on either side, guaranteeing a sensor data array where separate objects must be characterized
minimum safe separation distance between the aircraft and the and distinguished. For the obstacle resolution, we consider two
obstacle. In the recovery phase, the trajectory takes the aircraft clustering techniques: DBSCAN and spectral clustering technique.
from the intermediate waypoint to a prescribed target, or final In DBSCAN, the number of clusters is determined by the clustering
waypoint. The avoidance path is continuously optimized based on process itself, driven by the grouping of points that meet a user-
updated airborne sensor information and vehicle states. defined minimum density. The latter is established by two param-
eters that define the ϵ -neighborhood of a point p: the minimum
2.2. Two-level algorithm in the guidance, navigation, and control number of points minPts that must be within a distance ϵ from
architecture point p. These required user-defined algorithm parameters are
lower-level data attributes that are more intuitive, meaningful,
The guidance navigation and control architecture employed in and practical than emergent clustering characteristics, such as
this work, illustrated in Fig. 2, is consistent with typical constructs number of clusters. Point density also provides an ideal mechanism
in current practice. Optimal trajectory generation realized with to resolve adjacent clusters while guaranteeing that a minimum
guidance and navigation elements, as well as application of air- safe distance between them exists. Specifically for the current
craft controller to the flight dynamics, ensures adherence to said application of obstacle resolution, ϵ can be used to explicitly set
trajectory. In our two-level concept, guidance and navigation have the minimum distance between obstacles beyond which they are
an off-line block where the trajectory is defined a priori in accor- resolved as separate objects.
dance with a prescribed mission profile, pre-defined waypoints, In the spectral clustering algorithm, the number of clusters k
and available obstacle information (if applicable or available). The is initially identified through k-block diagonality of the Laplacian
on-line counterpart regularly updates the trajectory based on the matrix L. Once k is known, clusters are resolved with any algorithm
current vehicle state and sensor data to compensate for perturba- requiring k as an input, such as k-means. An adjacency matrix is
tions or the avoidance of unknown obstacles. The global and local binary and denotes pairwise connectivity of nodes with a ‘‘1’’, and
components of the proposed trajectory optimization concept are ‘‘0’’ otherwise. We introduce a characteristic distance d as the basis
realized in this region of the architecture. As shown in Fig. 2, sensor of distance-based adjacency and define elements in A as follows:
data corresponding to detected obstacles in the operating environ- Aij = 1,
 
if xi − xj  ≤ d
ment are passed to the global block where distinct obstacles are (1)
Aij = 0, if xi − xj  > d
 
resolved, and potential threats are specified.
Once the potential threats are identified, the local component where the vectors xi and xj denote data points, or vertices in the
solves a two-phase trajectory optimization problem. Note how- graph-theoretic sense.
ever that if a potential threat does not exist, the local component The overall approach is summarized in Algorithm 1. Caution
solves a one-phase trajectory optimization problem. The aircraft should be exercised whenever noise is present in the data because
controller generates input commands to follow the optimal trajec- it degrades k-block diagonality [28]. For applications where data
tory, and applies them to the vehicle flight dynamics. The aircraft features high signal-to-noise ratio, or in the extreme case of com-
controller loop monitors vehicle states and updates control inputs pletely deterministic data such as that produced by our determin-
with feedback rate fa . The guidance and navigation loop updates istic sensor model, concerns associated with k-block diagonality
the optimal trajectory with feedback rate fg given aircraft states may be ignored.
and the current global path approximation.
 Y. Choi et al. / Robotics and Autonomous Systems 98 (2017) 158–173 161

Fig. 1. Concept formulation and flow of the two-layer collision avoidance algorithm.

Fig. 2. New path planning architecture using a machine learning algorithm.

ways. First, it expedites the runtime of the clustering function.

We denote the distance Do as the user-defined minimum ob- Second, the need to resolve obstacles based on depth (along the
longitudinal x axis of the vehicle coordinate system), and the
served distance between obstacles for resolution via sensor data
need to incorporate this information into a more complex global
clustering. Accordingly, in our implementation of the two clus-
trajectory optimization, is circumvented altogether.
tering approaches discussed here, the characteristic distance pa-
On the other hand, exclusion of depth information can poten-
rameter, ϵ for DBSCAN and d for spectral clustering with k-block tially degrade the ability to solve for optimal avoidance trajectories
diagonality, are set to guarantee that sensed obstacles that appear under certain conditions. Specifically, without depth information,
separated by a distance Do or more are resolved as separate objects. obstacles that are partially or totally blocked by nearer obstacles
Each of the clustering algorithms is implemented in a two- are not resolved as separate objects. As a result, optimal trajec-
dimensional domain where the sensor data has been collapsed tories that may exist such that aircraft can navigate between up-
onto the yz plane of the vehicle coordinate system. Eliminating stream obstacles and partially blocked obstacles downstream are
obstacle depth affords significant computational efficiency in two not realized. This condition is examined in depth with a numerical
162 Y. Choi et al. / Robotics and Autonomous Systems 98 (2017) 158–173

Algorithm 1 Spectral clustering optimization. When a potential threat is not detected, the local tra-
Inputs: point cloud information P ∈ Rn×d , distance constraint jectory optimization solves the one-phase trajectory optimization
between obstacles Do problem to reach the target position xt . When a potential treat is
(1) Distance-based adjacency n×n detected, the local trajectory optimization solves the two-phase
∑ matrix Aij ∈ R , d = Do
(2) Degree matrix Dii = A ij if i = j, otherwise Dij = 0 trajectory optimization problem to avoid the obstacle resolved
j
(3) Unnormalized Laplacian matrix L = D − A with cluster C̃ and then reach the target position xt . We recognize
(4) Solve for k with k-block diagonality: rank(L) = n - k that obstacle avoidance and recovery towards a prescribed target
(5) Treat each row of L as a point in Rk , and cluster into k clusters encompass the two fundamental phases of collision avoidance, and
via k-means that a multi-phase approach to trajectory optimization is therefore
Outputs: Clustering result C1 , C2 , ... , Ck well-suited for this problem. In this subsection, we only illustrate
the multi-phase framework since the one-phase framework is
relatively simple and can be easily defined from the multi-phase
framework.
In general, the multi-phase problem divides the trajectory into
n phases or segments (p ∈ 0, 1, 2, . . . , n) and sequentially solves
for the optimal trajectory in each phase p. The formulation is pred-
icated on the definition of a performance index or cost function,
dynamic constraints, path constraints, event constraints, and link
constraints. The cost function can be written as
n
∑ (p) (p)
min J(x̄(p) , u(p) , t) = 8(p) (x̄f , tf )
p=1
n ∫ (p)
∑ tf
+ L(p) (x̄(p) , u(p) , t)dt
(p)
p=1 t0

p
Fig. 3. Identification of a potential threat. where the superscript denotes the pth phase, x̄ is a state vector, u
is a control input vector, t is time, t0 is initial time, and tf is terminal
time. 8 and L are terminal and transient costs, also called Mayer
and Lagrange costs in the optimal control theory context. Inclusion
simulation of a representative scenario in Section 3.3, where we
of transient and terminal costs represents the most general for-
note that the performance of the proposed two-layer approach is
mulation of the cost function. However, for subclasses of optimal
reduced to that of the single-layer alternative.
trajectory problems, only transient or terminal costs will suffice.
Vehicle state equations are captured as dynamic constraints,
2.3.2. Identification of a potential threat
typically expressed in the form:
The second step in global trajectory optimization is to construct
global path alternatives commensurate with the obstacle informa- dx̄(p)
tion produced in the previous step. We note however that there = f (p) (x̄(p) , u(p) , t). (3)
dt
exists an infinite number of feasible trajectories that maintain a
Path constraints are algebraic inequalities that capture vehicle
minimum safe distance from all obstacles. We proposed to iden-
flight performance limitations such as velocity, acceleration, and
tify potential threats in the global optimization loop, and solve
thrust limits. These constraints can be represented as follows:
the two-phase trajectory optimization problem using the threat
information in the local trajectory optimization. Note that if the (p)
cmin ≤ c(p) (x̄(p) , u(p) , t) ≤ c(p)
max . (4)
potential threat does not exist, the local trajectory optimization
solves the one-layer trajectory problem. Event constraints establish the conditions that must be satisfied
The potential threat is selected by individual cluster informa- at the final time tf of each phase p, and are generally written in the
tion and the vehicle velocity vector v = [vx vy vz ]. More specif- form
ically, we project each point cluster onto a plane for which the (p)
normal vector is parallel to x-direction (vx ) of the velocity vector, Emin ≤ E(p) (x̄(p) , u(p) , t) ≤ E(p)
max . (5)
and located at a distance corresponding to the nearest point of the Phase link constraints ensure continuous state transition be-
cluster along the x-direction of the navigation coordinate system. tween phases. For the link s between phases p − 1 and p, the
It then projects the velocity vector onto individual planes. For constraints are expressed as:
instance, in the case of Fig. 3, we project the velocity vector onto
three planes (plane 1, plane 2, and plane 3). The projected velocity χ(s) (x̄f
(p−1)
, x̄(p) (p−1)
(tf ) − x̄(p) (t0 ) = 0, (p = s + 1).
0 ) = x̄ (6)
vector judges a potential threat. In other words, if the projected
vector on a plane is inside of the plane boundary, that cluster is
identified as a potential threat. In the example, the cluster on plane 2.4.2. Two-phase trajectory optimization framework
2 is a potential threat. In some cases, there may not be any potential To formulate the local-trajectory optimization, we suggest a
threats. This case solves as a one-phase optimal trajectory problem two-phase approach based on the multi-phase optimal trajectory
in the local trajectory optimization. framework. The first phase (p = 1) is an avoidance phase, and
the second phase (p = 2) is a recovery phase. The cost function
2.4. Local trajectory optimization and constraints are defined accordingly. Trajectory solutions can
be time-optimal, effort- or energy-optimal, or hybrid by combin-
2.4.1. General multi-phase optimal trajectory problem ing the two. Other performance index formulations linked to the
In the online process, the local trajectory optimization solves vehicle dynamics are certainly possible, although time and energy
two types of problems, namely, one- and two-phase trajectory are very commonly used as the basis. The time-based approach
 Y. Choi et al. / Robotics and Autonomous Systems 98 (2017) 158–173 163

minimizes trajectory duration, say in a hostile operating environ- are defined as

ment, and has been employed in a number of studies [22,23,31]. 1
⎡ ⎤
− 0 0
However, minimal-time trajectories impose high energy require- ⎢ τx ⎥
⎢ 1 ⎥
ments associated with aggressive maneuvers and accelerations, Ã = ⎢ 0 −
⎢ 0 ⎥,
⎥
which in turn strain energy management requirements and can ⎢ τy ⎥
⎣ 1⎦
degrade mission feasibility. On the other hand, a minimal-energy 0 0 −
τz
solution emphasizes mission energy management and efficiency,
1
⎡ ⎤
and is not driven by time or duration limitations. For this study, 0 0
⎢ τx ⎥ (9)
we adopt an energy-based performance index to assess effort- ⎢0 1 0⎥
⎢ ⎥
B̃ = ⎢ ⎥,
efficiency improvements attainable with the proposed collision ⎢ τy ⎥
avoidance concept. The performance index is defined as:
⎣ 1⎦
0 0
τz
n=2 ∫ (p)
tf
1∑
⎡ ⎤
0 ··· 0
J = u(p)T W(p) u(p) dt (7)
2 ⎢. .. ⎥ .
0̃n×n = ⎣ ..
(p)
t0
p=1 .⎦
0 ··· 0
where phase p (here limited to p = 1, 2) is denoted by the super-
script p , t0 and tf are initial and final times, respectively, u(p) ∈ R3 is τx , τy , and τz are time constants with respect to the three accelera-
(p) (p) (p)
(p) (p) (p)
the acceleration command input vector [ucx ucy ucz ]T , and W(p) tion state variables [ax ay az ]T .
Event constraints in the first phase prescribe the terminal vehi-
is a 3 × 3 weighting matrix presumed to be the same in both phases.
cle states that guarantee obstacle avoidance, and they are therefore
The 1/2 factor is inherent in the transient cost function L and is
fundamental for the entire collision avoidance trajectory problem.
simply placed outside the summation in Eq. (7) above. The cost In our approach, prescription of this terminal state is predicated on
function adopted here is based on the notion that work expended two conditions. First, the aircraft is in unaccelerated, level flight.
follows the commanded acceleration, squared to ensure positive This flight condition has been commonly used in the past, for
quantities, and has been used frequently in prior optimal trajectory instance by Moon and Kang [22]. It offers a simple and tractable
state that is justifiable in most cases when the target point after
problems (see for instance Ref. [32]). In the present formulation,
successful collision avoidance prescribes an altitude and velocity
the effort associated with the trajectory is entirely captured in the
comparable to that of the initial condition. We adopt this condition
transient cost expression, and a terminal cost component is not but note that it is simply suggested here, and that other conditions
necessary. may also be considered. Second, to enforce a safe distance from
In general, constraint complexity has a significant effect on any avoided obstacle, we introduce the variable (rs ). Three linear
the runtime performance of the online obstacle avoidance algo- inequality constraints are defined at a distance rs around the ob-
stacle cluster C̃ , one to the left, one to the right, and one above, as
rithm, so the choice of constraints and their inherent sophisti-
shown in Fig. 4.
cation must be considered carefully. This is particularly relevant
Accordingly, we define the terminal state as
for dynamic constraints where simplified flight dynamics can be
(1)
adopted whenever possible instead of complex non-linear alter- tf = free
natives. Moon and Kang [22,23] propose first-order acceleration x(1) (tf ) = [x(1) (tf ) free free]T (x(1) (tf ) = min C̃x − rs )
dynamic equations for simplified kinematics of an optimal tra- (10)
v(1) (tf ) = [u(1) (tf ) 0 0 ]T
jectory problem. We employ this first-order approximation for
dynamic constraints to achieve a computationally and dynamically a(1) (tf ) = [0 0 0]T
feasible solution. This choice for low order dynamics constraints
where C̃x is the x position information of C̃ . Note that the safe
is a practical one, driven by the present focus to demonstrate the position along the x axis, x(1) (tf ), is prescribed according to rs too.
proposed two-layer approach in relation to the classical one-layer The y and z positions are free variables that are only restricted by
alternative. An implementation of our method with higher order the following inequality constraints, pursuant to the safe distance
dynamics is immediately feasible, but increased runtime should be from the obstacle for the avoidance trajectory:
⎧
expected as a result. ⎪ y ≤ C̃ymin + Mt1
⎨
The simplified kinematic constraints are as follows: y ≥ C̃ymax − Mt2
⎪
⎩
z ≥ C̃zmax − Mt3 (11)
⎡ ⎤ ⎡ ⎤⎡ ⎤
ẋ(p) 0̃3×3 Ĩ3×3 0̃3×3 x(p) [ ]
0̃6×3
⎣v̇(p) ⎦ = ⎣0̃6×6 Ĩ3×3 ⎦ ⎣ v(p) ⎦
+ u(p) (8) t = [t1 t2 t3 ], ti ∈ [1, 0]
B̃
ȧ(p) Ã a (p)
M is a large number
(p) (p) (p) (p) T (p)
where x is the position vector [x y z ] , v is the where C̃ymin and C̃ymax are the minimum and maximum y positions,
velocity vector [u(p) v (p) w (p) ]T , a is the acceleration vec- respectively, and C̃zmax is the maximum z position, in the point
(p) (p) (p)
tor [ax ay az ]T , and u(p) is the acceleration command cloud C̃ . The binary vector t identifies which of the three obstacle
(p) (p) (p) T
[ucx ucy ucz ] . These states are expressed in a navigation coor- avoidance constraints is active with vector element ti = 0, and
dinate system. Ĩ3×3 is a 3 × 3 identity matrix, and Ã, B̃, and 0̃n×m tj̸=i = 1 otherwise. Identification of the active constraint, and
164 Y. Choi et al. / Robotics and Autonomous Systems 98 (2017) 158–173

stated as simple load factor constraints:

|ax | < gxmax
⏐ay ⏐ < gymax
⏐ ⏐
(15)

|az + g | < gzmax

where gxmax , gymax , and gzmax are maximum allowable accelera-
tions. We also adopt a minimum altitude constraint as a path
constraint to guarantee flight above some operational minimum
to prevent possible collisions with other ground assets.

z ≥ zmin . (16)

The last constraint is the linkage constraint at the phase transition

point to enforce state continuity across phases:

x̄(p−1) (tf ) = x̄(p) (t0 ). (17)

Fig. 4. Notional depiction of the three obstacle avoidance trajectory event con-
Since the proposed framework of the optimal collision avoidance
straints at a safe distance rs around the obstacle cluster C̃ .
problem has two phases, in the phase linkage constraint (Eq. (17)),
p is two. The formulated framework of the multi-phase optimal
trajectory problem is solved by the Gauss Pseudospectral method
(GPM) provided by the open-source software GPOPS. This Gauss
therefore definition of the binary vector t, is solved for as follows: Pseudospectral technique was developed by Benson [33,34], and
⏐ ⏐ ⏐ ⏐ it was advanced and validated by empirical cases studies from
D = [⏐(min C̃y − rs ) − yp ⏐ , ⏐(max C̃y + rs ) − yp ⏐ ,
⏐ ⏐ ⏐ ⏐
Huntinton et al. [35–37]. This GPM technique employs an orthogo-
⏐ ⏐ nal collocation method based on the Legendre–Gauss points. The
⏐(max C̃z + rs ) − zp ⏐] GPOPS software provides a MATLAB interface with a non-linear
⏐ ⏐
⏐ ⏐ programming problem solver, SNOPT [38,39].
t = [0 1 1], if min D = ⏐(min C̃y − rs ) − yp ⏐
⏐ ⏐
(12) We further examine the proposed two-layer algorithm in terms
⏐ ⏐ of the rotation angle of the cuboid obstacle relative to the initial
t = [1 0 1], if min D = ⏐(max C̃y + rs ) − yp ⏐
⏐ ⏐ navigation coordinate system. The impact of this geometric feature
is assessed by a comparative assessment of a representative sce-
⏐ ⏐
nario where a cuboid obstacle faces the aircraft navigation coordi-
t = [1 1 0], if min D = ⏐(max C̃z + rs ) − zp ⏐ .
⏐ ⏐
nate system at the start of the simulation, and a second case where
the same object is rotated. The time sequence of the avoidance
Here, yp and zp are the vehicle position in y and z projected with
(1) trajectories that result from the simulation, as well as the vehicle
the current velocity vector v0 (t) onto a plane perpendicular to the response, are captured in Figs. 5 and 6. In both figures, the yellow
velocity vector at a downrange distance x(1) (tf ). lines are the sensor rays, the black line is the vehicle response
Event constraints for the second phase dictate the terminal state trajectory, and the blue line is the result of the collision avoidance
at the target position, for which we recommend unaccelerated trajectory at a specific time t. The results show that in the first
level flight, as follows: case where the obstacle faces the aircraft, the resulting avoidance
trajectory is generated to the right of the obstacle. For the case
(2) where the obstacle has been rotated, the energy optimal trajectory
tf = free
]T is to the left of the obstacle, which is indicative that the projected
x(2) (tf ) = x(2) (tf ) y(2) (tf ) z (2) (tf )
[
velocity vector is closer to the left edge of the sensor data point
(13) projection of the rotated building. In both cases, the solutions are
v(2) (tf ) = [vx(2) (tf ) 0 0] found to be energy optimal, and more generally demonstrate that
a(2) (tf ) = [0 0 0]. alignment of the cuboid obstacle faces to the navigation coordinate
system is not required to attain energy-optimal trajectories.
Path and link constraints are defined equally for the avoidance and
recovery phases. Path constraints curtail vehicle dynamic perfor- 3. Numerical simulation
mance and include limits on velocity and acceleration, keeping the
3.1. Simulation of unmanned aircraft dynamics, controller, and sensor
vehicle within the flight envelope to prevent unfeasible maneu-
vers. Path constraints are defined as follows:
For the numerical simulation, we assume a small electric fixed-
umin ≤ u ≤ umax wing unmanned aircraft, and we adopt the Aerosonde specification
taken directly from the literature [40,41]. Table 1 summarizes
vmin ≤ v ≤ vmax airframe parameters, along with those assumed for the airborne
(14)
wmin ≤ w ≤ wmax sensor and collision avoidance scheme. For simplicity, the vehicle
√ equations of motion assume a point mass model, as reported in
Vmin ≤ u2 + v 2 + w 2 ≤ Vmax previous work [42–44]. For the aircraft tracking controller, we em-
ploy the standard feedback control structure. The on-board sensor
where umin , vmin , and wmin are the low speed limits, and umax , vmax , is modeled as a generic light detection and ranging instrument, or
and wmax are the high speed limits. The acceleration constraints are LiDAR, with no instrument uncertainties.
 Y. Choi et al. / Robotics and Autonomous Systems 98 (2017) 158–173 165

Fig. 5. Time sequence of the optimal trajectory when an obstacle is aligned with the initial navigation frame.

Fig. 6. Time sequence of the optimal trajectory when an obstacle is not aligned with the initial navigation frame.

3.2. Comparative analysis of clustering algorithms for obstacle reso- the down-range distance between the unmanned aircraft and each
lution of the two obstacles is varied with the x axis location of the obstacle
centroid as follows: L1, L2 = [1100 1200 1300 1400 1500].
We compare the DBSCAN and spectral clustering algorithms We combine the three independent factor variations into a full
for the present application in terms of prediction success rate and factorial design of experiments with 125 samples. The UAV sensor
computational time. These figures of merit are evaluated over a parameters and the minimum separation distance between obsta-
design of experiments based on the setup illustrated in Fig. 7, cles, Do , are described in Table 1. The parameter Do is mapped to
where two identical, generic, cuboid obstacles lay ahead of the the characteristic distance in the clustering algorithm of choice, so
aircraft flight path. The x, y, and z axes shown are the vehicle that obstacles are resolved as separate objects only if the distance
coordinate system GV . The sensor data collection is assumed to between them is greater than Do . Note that to allow safe flight
be instantaneous, and the clustering algorithms executed at that between resolved obstacles, the distance between them must be at
instant only, so that there is no trajectory optimization or flight least twice the minimum separation distance between the aircraft
simulation, only clustering evaluation at the instant depicted. The and any obstacle, rs .
distance 2D between obstacles, referenced to the origin of the y
axis, is sampled as follows: D = [20 40 60 80 100]. Similarly, Do ≥ 2rs . (18)
166 Y. Choi et al. / Robotics and Autonomous Systems 98 (2017) 158–173

Table 1
Description of UAV parameters.
UAV parameters Variable Value Unit
Weight w 29.76 [lb]
Planform area S 6.1 [ft2 ]
Area swept out by the propeller Sp 0.1348 [ft2 ]
Propeller aerodynamic coefficient Cp 1
Efficiency constant of the motor Km 8
Vehicle parameters
Lift coefficient at zero angle of attack CL0 0.28
Lift curve slope CLα 3.45
Aspect ratio AR 10.7
Span efficiency e 0.9
Zero-lift drag coefficient CD0 0.03

Distance range Sr 1500 [ft]

Azimuth range AZ 60 [deg]
Sensor parameters
Elevation range EL 45 [deg]
Sensor resolution θsen 1.5 [deg]

Minimum distance constraint between obstacles Do 150 [ft]

Minimum separation distance between an obstacle and a vehicle rs 70 [ft]
Collision avoidance parameters
Minimum altitude zmin 200 [ft]
Maximum acceleration gmax 64.34 [ft/s2 ]

Update rate of guidance and navigation loop fg 1 [Hz]

Other parameters
Update rate of aircraft control loop fa 10 [Hz]

Figs. 8–11 illustrate four representative cases in the evaluation of

clustering algorithms. Based on the results obtained, we opt for
DBSCAN as the preferred clustering algorithm in the global-path
optimization.
The clustering algorithm has a computational burden on the
order of 0.1 [s]. Next, the global optimization takes, on average,
0.019 [s]. If an obstacle is found to present a conflict for the current
trajectory, the two-phase local trajectory optimization takes, on
average, 3.5 [s], and if no obstacle threat exists then the single
phase local optimization takes approximately 0.49 [s]. This means
that for cases where obstacle conflicts are found, the relative cost
of the clustering algorithm is marginal, on the order of 3.4%. On
the other hand, for cases where there are no obstacle threats, the
computational burden of the clustering algorithm is 24.3%.
(a) Top view.
3.3. Assessment of two-layer collision avoidance with numerical sim-
ulation

To test the hypothesis that our two-layer collision avoid-

ance algorithm yields more energy efficient trajectories without
incurring a prohibitive computational burden relative to the
single-layer approach, we conduct an experiment with numerical
simulations. The experiment is comprised of three use cases de-
signed to be relevant and representative for UAS applications of
(b) Side view. interest, and to capture critical arrangements of relative obstacle
placement in terms of on-board sensing and optimal avoidance
Fig. 7. Simulated experiment setup for comparative analysis of clustering algo- trajectory execution. Simulations for the four use cases are run
rithms. (Aircraft not shown to scale). with both methods, single- and two-layer, using the same initial
conditions, set to steady level flight at 500 feet altitude with a
velocity of 70 feet/s.
This experiment was conducted on a commercially available The first use case is illustrated in Fig. 12 with the trajectory
desktop computer with a 3.40 GHz Intel(R) Core(TM) i7-2600 pro- simulation results of the two collision avoidance methods. The
cessor and 8 GB of RAM. Results of the clustering experiments are obstacle-free trajectory between the initial and final points is
summarized in Table 2 and show that both algorithms accurately shown in red, and the executed collision avoidance trajectory
resolve clusters with a 100% success rate. That is, when the gap with each method is shown in dotted blue. The first scenario
between obstacles is larger than Do (i.e., 2D ≥ Do ), both algorithms features two obstacles of comparable altitude (∼1000 feet) and
exactly resolve two clusters regardless of variation of the distances footprint, located at the same downrange distance (∼3800 feet),
L1 and L2 . and separated by a lateral distance (∼175 feet) greater than twice
In terms of computational efficiency, DBSCAN is faster and fea- the safe distance defined for aircraft-to-obstacle separation. This
tures lower variance across experiments. We attribute the greater initial scenario is intended to characterize the benefits of safe flight
runtime mean and variance of spectral clustering in part to the use between adjacent obstacles. Results show that with the one-layer
of k-means with random initialization for the initial center points. obstacle avoidance algorithm the aircraft maneuvers left around
 Y. Choi et al. / Robotics and Autonomous Systems 98 (2017) 158–173 167

Table 2
Comparison assessment results of the clustering algorithms.
DBSCAN Spectral method
Prediction success rate (%) Computational time (s) Prediction success rate (%) Computational time (s)
Average 100 0.101 100 0.133
Standard deviation 0 0.027 0 0.214

(a) Raw data.

(a) Raw data.

(b) Clustering results.

(b) Clustering results.
Fig. 8. Clustering results (D = 20 ft, L1 = 1100 ft, L2 = 1100 ft).
Fig. 9. Clustering results (D = 40 ft, L1 = 1100 ft, L2 = 1500 ft).

the left obstacle, since that route is more energy efficient than
flying over the two obstacles or around the right obstacle. As maneuvering to the left of the left obstacle, and then keeping to
expected, in this simulation, the aircraft treats both obstacles as the left of the third obstacle in the back. The two-layer algorithm
a single object and executes the most energy-efficient trajectory executes an avoidance trajectory between the first two obstacles,
to avoid it and meet the final point. In contrast, the result of and then around the right of the third obstacle behind them. Fig. 15
two-layer obstacle avoidance algorithm shows that the aircraft presents the time history of velocity, acceleration, and the separa-
flies a trajectory through the gap between the two obstacles, with tion distance. As with the previous use case, the time responses
only a minor course correction too small to be appreciated in the indicate that the two-layer approach features smaller velocity
figure. Again, the result is reasonably expected as the gap between variations, less energy required, and comparable separation to
the two obstacles exceeds the minimum separation, allowing the obstacles.
global optimizer to resolve them as two separate objects, and The third use case, depicted in Fig. 16, reverses the obstacle
for the local optimizer to produce an energy-optimal avoidance arrangement, presenting first a single obstacle whose footprint
trajectory between the two. Fig. 13 shows the time history of and orientation are intentionally chosen to significantly block the
velocity, acceleration, and separation distance between the aircraft two obstacles behind it. This scenario is intended to test the case
where information about downrange obstacles cannot be acquired
and the obstacles. The velocity responses show that the single-
until a closer obstacle is cleared. Moreover, it tests the feasibility
layer solution features greater variation than that for the proposed
and efficacy of the proposed method with regard to our choice
two-layer structure, in this particular case due to the maneuver
to cluster sensor data in a 2D domain, that is, collapsed to the
around the left obstacle. Hence, unlike the one-layer structure, the xy plane, so that obstacles are not resolved based on depth data.
acceleration of the two-layer structure remains approximately at Results shows that the two obstacle avoidance algorithms have
zero, which is indicative of a more energy efficient trajectory. The the same initial obstacle avoidance trajectory to avoid the first
response of the separation distance shows that both techniques obstacle. After avoiding the first obstacle the aircraft detects two
generate a safe avoidance route. downstream obstacles and from that point on the two algorithms
The second use case, depicted in Fig. 14, builds upon the first produce different avoidance trajectory solutions. In the one-layer
one and is designed to test cases when there is a third obstacle obstacle avoidance trajectory, the aircraft flies above the two ob-
immediately behind the gap between the first two obstacles. As stacles because it computes the energy-efficient avoidance trajec-
before, the one-layer algorithm avoids the first two obstacles by tory resolving the two obstacles as a single object. The two-layer
168 Y. Choi et al. / Robotics and Autonomous Systems 98 (2017) 158–173

(a) Raw data.

(a) One-layer obstacle avoidance algorithm.

(b) Clustering results.

Fig. 10. Clustering results (D = 100 ft, L1 = 1500 ft, L2 = 1100 ft).

(b) Two-layer obstacle avoidance algorithm.

Fig. 12. Numerical simulation results of obstacle avoidance algorithms in the first
scenario.

(a) Raw data.

algorithm resolves two separate obstacles and executes an energy
optimal trajectory through the gap between them. Fig. 17 is the
time response. The time responses show that the velocity profile is
exactly the same before the aircraft detects the two downstream
obstacles, with notable differences in velocity variations after that.
The one-layer acceleration response is consistent with greater
energy demands.
The fourth case, depicted in Fig. 18, includes three obstacles
intentionally arranged so that view of the two downstream obsta-
cles is partially and totally blocked, respectively, by the upstream
obstacle. Like the third scenario discussed above, this scenario is
representative of conditions where the choice to exclude obstacle
depth information can be characterized in terms of impacts on
energy requirements for avoidance trajectories. The result shows
that both algorithms present the same avoidance trajectories be-
cause the clustering algorithm in the global optimization does
(b) Clustering results. not resolve the two obstacles and instead treats them as a single
obstacle. Fig. 19 is the time response. Both algorithms generate
Fig. 11. Clustering results (D = 100 ft, L1 = 1500 ft, L2 = 1500 ft). exactly the same response for velocity, acceleration, and separation
distance.
 Y. Choi et al. / Robotics and Autonomous Systems 98 (2017) 158–173 169

(a) One-layer obstacle avoidance algorithm.

Fig. 13. Time response of velocity, acceleration and minimum distance between the
aircraft and obstacles in the first scenario.

We quantitatively assess the performance of both algorithms

for the four scenarios by estimating the work done by the aircraft,
or conversely, the energy required to execute each trajectory.
We adopt the classical energy formulation for flight performance,
where the work performed by the aircraft, i.e. the energy ex-
pended, is estimated as thrust (force) applied over distance, where
it is assumed that the thrust and velocity vectors are aligned so
that thrust is always in the direction of motion. In this formulation,
there is no work associated with lift as it is always perpendic- (b) Two-layer obstacle avoidance algorithm.
ular to the trajectory. The work associated with drag is always
an energy loss. Weight work corresponds to the change in po- Fig. 14. Numerical simulation results of obstacle avoidance algorithms in the
tential energy. The work associated with the imbalance between second scenario.
thrust, drag, and the component of weight aligned with them,
resulting in longitudinal accelerations, corresponds to the change
in kinetic energy. The latter is often adjusted via thrust input.
Accordingly, executing a trajectory is given by
T = D + W sin γ + F ∫
WT = Tds, (20)
T ds = Dds + W ds sin γ + F ds c
(19)
WT = WD + WW + WF where T is thrust, c is an avoidance trajectory, and ds is the length
of a discrete segment along the avoidance trajectory.
WT = WD + 1PE + 1KE . Table 3 summarizes the measurement results of each method
Here, γ is the flight path angle, F is the thrust-controlled force im- under the four different scenarios. Results clearly indicate that the
balance between thrust, drag, and the weight component aligned proposed method requires less energy than the one-layer collision
to the flight path, ds is an infinitesimal segment along the flight avoidance method. It is worth noting too that both algorithms
trajectory, and WT denotes the work associated with force T , as establish energy-optimal trajectories around resolved obstacles, so
is also the case for forces D, W , and F . Note that for cases where
that all energy-efficiency benefits observed can only be attributed
γ = 0, the work associated with weight is zero, and there is no
to the resolution of separate objects from the proposed two-layer
change in potential energy, so 1PE = 0. Similarly, for the case of
null imbalance force, i.e. F = 0, the work associated with F is zero, approach.
and there is no change in kinetic energy, so 1KE = 0. Results clearly show that the proposed two-layer approach
Recognizing that thrust, and the work it performs over ds, generates more energy-efficient obstacle avoidance trajectories.
equate to changes in kinetic and potential energy while over- The inclusion of the global trajectory layer results in additional
coming drag, then the work performed (or energy expended) in computational expense, where the most demanding step is the
170 Y. Choi et al. / Robotics and Autonomous Systems 98 (2017) 158–173

Table 3
Summary of numerical simulation.
Scenario One-layer structure (lb ft) Two-layer structure (lb ft) Energy difference (lb ft)
W1 W2 W1 − W2
1 19 816 18 690 1125
2 19 816 18 896 920
3 23 353 21 645 1708
4 18 773 18 773 0

(a) One-layer obstacle avoidance algorithm.

Fig. 15. Time response of velocity, acceleration and minimum distance between the
aircraft and obstacles in the second scenario.

solution of the clustering problem. Based on our runtime assess-

ments of DBSCAN presented in Table 2, we find that this cost
is acceptably low in comparison to the typical runtime of the
local trajectory optimization. Additional runtime benefits may be
attained via programming improvements or parallel computing
techniques.

3.4. Simulation in a realistic urban environment (b) Two-layer obstacle avoidance algorithm.

In this section, we demonstrate the proposed two-layer obsta-

Fig. 16. Numerical simulation results of obstacle avoidance algorithms in the third
cle avoidance algorithm in a representative urban environment.
scenario.
In order to generate a fully realistic environment, we generate a
representation of a section of downtown San Diego, as shown in
Fig. 20(a). To generate this model of a real urban setting, we use
the actual geometric information from publicly available LiDAR altitude is assumed to be 350 feet, and the initial flight condition is
surveys. The result of the urban model is presented in Fig. 20(b). a level flight at a speed of 70 feet/s.
In this model, buildings are represented as cuboid bodies with a Table 4 summarizes the energy quantification results of the
minimum altitude of 200 feet. ten avoidance trajectories. The results clearly illustrate that the
We identify the ten trajectories traversing this environment proposed two-layer algorithm yields more energy efficient trajec-
with the most conflicts with obstacles, as illustrated in Fig. 21 with tories, with the exception of the sixth case, where the optimal
the initial and terminal waypoints. To identify these challenging trajectory is exactly the same as that of the single-layer method.
routes and to avoid trivial cases with no obstacle conflicts, we The average improvement is approximately 7.1%. Two representa-
define five initial and terminal conditions along the x and y axis tive simulation results, shown in Figs. 22 and 23, indicate typical
and compute the initial optimal trajectories. Then, we select the ten differences observed, specifically where the traditional one-layer
routes with the greatest number of conflicts. Note that the initial approach yields trajectories that fly over the all obstacles in a dense
 Y. Choi et al. / Robotics and Autonomous Systems 98 (2017) 158–173 171

Table 4
Energy quantification of the urban scenarios.
Scenario no. 1 2 3 4 5 6 7 8 9 10
One layer [lb ft] 19746.39 18074.16 17997.87 16962.03 17131.05 15879.09 19681.15 16095.72 19713.92 18725.62
Two layer [lb ft] 16777.26 16436.40 17178.83 16137.37 16381.19 15879.09 17161.34 15644.07 17371.21 17621.44
Improvement [%] 15.04 9.06 4.55 4.86 4.38 0.00 12.80 2.81 11.88 5.90

Fig. 19. Time response of velocity, acceleration and minimum distance between the
Fig. 17. Time response of velocity, acceleration and minimum distance between the aircraft and obstacles in the fourth scenario.
aircraft and obstacles in the third scenario.

4. Conclusion

Many existing collision avoidance concepts using on-board

sensor data fail to generate efficient trajectories because, even
though multiple objects may be detected, the collision avoidance
algorithm treats them as a single conglomerate regardless of their
number and relative location. Resulting avoidance trajectories are
therefore optimal only in the context of a large avoidance region
and fail to exploit free space between obstacles that may allow
for more efficient solutions. The approach proposed in this pa-
per features a two-layer multi-phase optimal collision avoidance
algorithm that entails global- and local-path optimization. The
purpose of the global-path optimization is to resolve multiple
obstacles as separate objects, given that they satisfy a separation
minimum, with sensor data clustering. The global-path optimiza-
tion also identifies a potential threat using the cluster data of
resolved objects. We identify two candidate clustering techniques
that allow for the parametric specification of a cluster separation
minimum for obstacle resolution: DBSCAN and spectral clustering.
A design of simulation experiments is executed to evaluate both
Fig. 18. Numerical simulation results of obstacle avoidance algorithms (one-layer,
and second-layer) in the fourth scenario. techniques in terms of prediction success rate and computational
runtime. Although both methods are found to have 100% rates
for the simulations performed, DBSCAN featured lower runtime
mean and variance, thus emerging as the recommended clustering
region, whereas the proposed two-layer approach yields trajec- technique for the proposed concept.
Numerical simulations with status-quo one-layer collision
tories where the aircraft flies through the gap between obstacles
avoidance and the proposed two-layer concept are conducted
whenever sufficient distance exists between them. to compare their performance in terms of the energy-efficiency
172 Y. Choi et al. / Robotics and Autonomous Systems 98 (2017) 158–173

(a) San Diego downtown (Google Earth image).

Fig. 22. Numerical simulation result in the case with high energy difference (second
scenario).

Fig. 23. Numerical simulation result in the case with low energy difference (third
scenario).

(b) Simulation environment.

those improvements may be. The actual flight tests and HIL tests to
address these factors remain as future work.
Fig. 20. Urban simulation environment (San Diego).
Collision avoidance with on-board sensors is a relatively new
and quickly evolving field where many areas of additional work are
warranted. Regarding the work here reported, we argue in favor of
a broader benchmarking study where a large set of multi-obstacle
scenarios is utilized, potentially with hardware-in-the-loop real-
time simulations. Moreover, this two-layer collision avoidance al-
gorithm has a flexible structure. In other words, the local trajectory
optimization can be realized with any of a myriad of obstacle
avoidance formulations.
We also believe that extension to a probabilistic study is a natu-
ral next step so that uncertainties associated with instrumentation,
on-board computer runtime, and flight dynamics and controls
governing trajectory execution, are treated explicitly. Ultimately,
the work presented here outlines a new concept that offers the
Fig. 21. Ten initial trajectories in the urban scenario.
potential of significant benefits for the rapidly growing field of UAS
applications and the increasing need to develop safety-assuring
technologies.
of executed trajectories. Four relevant use cases are utilized to
examine multi-obstacle placements of interest, assess the feasi- Acknowledgment
bility of the proposed concept, and test the main hypothesis of
this work. Results indicate that trajectory solutions with the two-
The authors thank Dr. Jongki Moon for his insights and expert
layer avoidance method provide tangible energy efficiency benefits input on this work.
while incurring a very small additional computational cost.
In order to further test, demonstrate, and validate the effec-
References
tiveness of the proposed algorithm, HIL (Hardware-In-the-Loop) or
unmanned flight tests are requisite, particularly since actual sensor [1] U. S. Department of Transportation, Air traffic organization policy (2014).
measurement data can contain significant noise and uncertainty [2] A.R. Lacher, D.R. Maroney, A.D. Zeitlin, Unmanned aircraft collision avoidance–
for which modifications to the method may be required. At this technology assessment and evaluation methods, in: The 7th Air Traffic Man-
point, however, we do not attempt to propose or speculate on what agement Research & Development Seminar. Barcelona, Spain (2007).
 Y. Choi et al. / Robotics and Autonomous Systems 98 (2017) 158–173 173

[3] K. Dalamagkidis, K.P. Valavanis, L.A. Piegl, On unmanned aircraft systems [30] S. Maniatopoulos, D. Panagou, K.J. Kyriakopoulos, Model predictive control for
issues, challenges and operational restrictions preventing integration into the the navigation of a nonholonomic vehicle with field-of-view constraints, in:
national airspace system, Progress in Aerospace Sciences 44 (7) (2008) 503– IEEE (Ed.), American Control Conference (ACC) (2013), pp. 3967–3972.
519. [31] S. Sundar, Z. Shiller, Time-optimal obstacle avoidance, in: Proceedings of the
[4] S.M. La Valle, Planning Algorithms, Cambridge University Press, 2006. IEEE International Conference on Robotics and Automation. vol. 3. Nagoya,
[5] D. Levine, B. Luders, J.P. How, Information-rich path planning with general Japan (2013).
constraints using rapidly-exploring random trees, in: AIAA Infotech Aerospace [32] M. Anderson, J. Sverdrup, J. Lopez, J. Evers, A comparison of trajectory deter-
Conference (2010). mination approaches for small uav’s, in: AIAA Atmospheric Flight Mechanics
[6] Y.-J. Ho, J.-S. Liu, Collision-free curvature-bounded smooth path planning Conference and Exhibit. Keystone, Colorado (2006).
using composite bezier curve based on voronoi diagram, in: Computational [33] D. Benson, A gauss pseudospectral transcription for optimal control, Ph.D. The-
Intelligence in Robotics and Automation (CIRA), 2009 IEEE International Sym- sis, Department of Aeronautics and Astronautics. Ph.D. Thesis, Massachusetts
posium on. IEEE (2009), pp. 463–468. Institute of Technology (2004).
[7] L. De Filippis, G. Guglieri, F. Quagliotti, Path planning strategies for UAVs in 3d [34] D.A. Benson, G.T. Huntington, T.P. Thorvaldsen, A.V. Rao, Direct trajectory
environments, J. Intell. Robot. Syst. 65 (1) (2012) 247–264. optimization and costate estimation via an orthogonal collocation method, J.
[8] O. Khatib, Real-time obstacle avoidance for manipulators and mobile robots, Guid. Control Dyn. 29 (6) (2006) 1435–1440.
Int. J. Robot. Res. 5 (1) (1986) 90–98. [35] G.T. Huntington, D.A. Benson, A.V. Rao, Design of optimal tetrahedral space-
[9] G. Li, A. Yamashita, H. Asama, Y. Tamura, An efficient improved artificial craft formations, Journal of the Astronautical Sciences 55 (2) (2007) 141–169.
potential field based regression search method for robot path planning, in: [36] G.T. Huntington, D.A. Benson, J.P. How, C.L.D.N. Kanizay, A.V. Rao, Computation
IEEE International Conference on Mechatronics and Automation (2012), pp. of boundary controls using a gauss pseudospectral method, in: 2007 Astrody-
1227–1232. namics Specialist Conference. Mackinac Island, Michigan (2007).
[10] D. Panagou, Motion planning and collision avoidance using navigation vector [37] G.T. Huntington, A.V. Rao, Optimal reconfiguration of spacecraft formations
fields, in: IEEE International Conference on Robotics and Automation (ICRA) using the gauss pseudospectral method, J. Guid. Control Dyn. 31 (3) (2008)
(2014), pp. 2513–2518. 689–698.
[11] T.T. Mac, C. Copot, A. Hernandez, R.D. Keyser, Improved potential field method [38] P.E. Gill, W. Murray, M.A. Saunders, SNOPT: An SQP algorithm for large-scale
for unknown obstacle avoidance using UAV in indoor environment, in: IEEE constrained optimization. Tech. rep., SIAM Review (2005).
14th International Symposium on Applied Machine Intelligence and Informat- [39] P.E. Gill, W. Murray, M.A. Saunders, User’S Guide for SNOPT Version 7: Software
ics (SAMI) (2016). for Large Scale Nonlinear Programming, Stanford University, 2006.
[12] X.-Y. Li, L. Jie Guo, Constructing affinity matrix in spectral clustering based on [40] F. Wieland, D. DeLaurentis, G. Kubat, Modeling and simulation for UAS in the
neighbor propagation, Neurocomputing 97 (2012) 125–130. Elsevier. NAS. Tech. rep., NASA (2012).
[13] J.-W. Park, H.-D. Oh, M.-J. Tahk, UAV collision avoidance based on geometric [41] R.W. Beard, T.W. McLain, Small Unmanned Aircraft: Theory and Practice,
approach, in: SICE Annual Conference, 20-22 August (2008). Princeton University Press, 2012.
[14] A. Mujumdar, Nonlinear geometric and differential geometric guidance of [42] W. Lin, Distributed UAV formation control using differential game approach,
UAVs for reactive collision avoidance. Tech. rep., DTIC Document (2009). Aerosp. Sci. Technol. 35 (2014) 54–62.
[15] A. Mujumdar, R. Padhi, Reactive collision avoidance of using nonlinear geo- [43] P.K. Menon, G.D. Sweriduk, B. Sridhar, Optimal strategies for free-flight air
metric and differential geometric guidance, J. Guid. Control Dyn. 34 (1) (2011) traffic conflict resolution, J. Guid. Control Dyn. 22 (2) (1999) 202–211.
303–311. [44] Y. Choi, H. Jimenez, D.N. Mavris, Statistical gain-scheduling method for aircraft
[16] A. Chakravarthy, D. Ghose, Obstacle avoidance in a dynamic environment: A flight simulation, Aerosp. Sci. Technol. 46 (2015) 493–505.
collision cone approach, Systems, Man and Cybernetics, Part A: Systems and
Humans 28 (5) (1998) 562–574.
[17] Y. Watanabe, A.J. Calise, E.N. Johnson, Vision-based obstacle avoidance for
UAVs, in: AIAA Guidance, Navigation and Control Conference. Hilton Head, Youngjun Choi is research faculty at the School of
South Carolina (2007). Guggenheim Aerospace Engineering, the Georgia Institute
[18] S. Tom, B.D. Moor, E. Feron, J. How, Mixed integer programming for multi- of Technology. Dr. Choi received Ph.D. from that institution
vehicle path planning, in: European control conference. vol. 1. (2001), pp. in 2016. He was a visiting researcher at the Centre for
2603–2608. Power Transmission and Motion Control at the University
[19] Y. Kuwata, J. How, Three dimensional receding horizon control for UAVs, in: of Bath and worked at the Agency for Defense Develop-
AIAA Guidance, Navigation, and Control Conference and Exhibit. Providence, ment. His research areas are guidance, navigation, and
Rhode Island (2004). control for autonomous vehicles, machine learning algo-
[20] D. Mellinger, A. Kushleyev, V. Kumar, Mixed-integer quadratic program tra- rithms, and aerospace system design.
jectory generation for heterogeneous quadrotor teams, in: Robotics, (ICRA), A.
(Eds.), IEEE International Conference. (2012), pp. 477–483.
[21] G.N. Vanderplaats, Multidiscipline Design Optimization, Vanderplaats Re-
search & Development, Inc., 2007.
Dr. Jimenez is research faculty at the Daniel Guggen-
[22] J. Moon, J. Prasad, Minimum-time approach to obstacle avoidance constrained
heim School of Aerospace Engineering, Georgia Institute
by envelope protection for autonomous UAVs, in: Proceedings of AHS 65th
of Technology. Dr. Jimenez received his Ph.D. from that
Annual Forum. Grapevine, TX (2009).
institution in 2009. He works at the Aerospace Systems
[23] K. Kang, J.V.R. Prasad, Development and flight test evaluations of an au-
Design Laboratory where he serves as the chief of UAS-
tonomous obstacle avoidance system for a rotary-wing UAV, Unmanned Sys-
Airspace Integration branch. In addition, he leads research
tems 1 (1) (2013) 3–19. efforts in areas of systems analysis, general aviation safety,
[24] Y. Peng, D. Qu, Y. Zhong, S. Xie, J. Luo, J. Gu, The obstacle detection and obstacle and aviation environmental impacts.
avoidance algorithm based on 2-d lidar, in: 2015 IEEE International Conference
on Information and Automation (2015).
[25] D.T. Pham, S.S. Dimov, C.D. Nguyen, Selection of k in k-means clustering.
Proceedings of the Institution of Mechanical Engineers, Part C: Journal of
Mechanical Engineering Science 219 (1) (2005) 103–119.
[26] H. Jia, S. Ding, X. Xu, R. Nie, The latest research progress on spectral clustering, Dimitri Mavris is S.P. Langley Regents Professor at the
Neural Comput. Appl. 24 (2014) 1477–1486. Daniel Guggenheim School of Aerospace Engineering,
[27] A.Y. Ng, M.I. Jordan, Y. Weiss, On spectral clustering: Analysis and an algorithm, Georgia Institute of Technology. He is director of the
Advances in Neural Information Processing System 2 (2002) 849–856. Aerospace Systems Design Laboratory where over 40 re-
[28] J. Feng, Z. Lin, H. Xu, S. Yan, Robust subspace segmentation with block-diagonal search faculty and 200 graduate students conduct research
prior. (2014), pp. 3818–3825. in civil aviation, propulsion and power, advanced con-
[29] M. Ester, H.-P. Kriegel, J. Sander, X. Xu, A density-based algorithm for dis- cepts, systems engineering, and defense and space.
covering clusters in large spatial databases with noise. vol. 96, International
Conference on Knowledge Discovery and Data Mining (KDD-96) (1996).