UAV Surveillance Under Visibility and Dwell-Time Constraints: A Sampling-Based Approach

Amit Surana; Francesco Bullo; Grant S. Taylor; Jeffrey R. Peters; Terry S. Turpin

REVIEW 3 major objections 4 minor 38 references

UAV Surveillance Under Visibility and Dwell-Time Constraints: A Sampling-Based Approach

T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Sampling the UAV flight space turns surveillance routing with viewing and dwell-time requirements into a generalized traveling salesperson problem, and provably approaches the continuous optimum as sampling is refined.

desk verdict A genuinely useful sampling-based framework for UAV surveillance with visibility and dwell-time constraints, but Theorem 3's resolution-completeness guarantee has a gap that goes beyond the exact-GTSP caveat. read the letter →

arxiv 1908.05347 v1 pith:5JA7SYEZ submitted 2019-08-08 eess.SY cs.MAcs.ROcs.SYmath.OC

Jeffrey R. Peters , Amit Surana , Grant S. Taylor , Terry S. Turpin , Francesco Bullo This is my paper · ORCID

classification eess.SYcs.MAcs.ROcs.SYmath.OC

keywords UAVsurveillancedwell-timeconstraintsvisibilityregionsDubinsvehiclegeneralizedtravelingsalespersonproblemsampling-basedpathplanningresolutioncompletenessepsilon-constraintscalarization

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper is trying to establish that a practical, sampling-based path planner can produce near-optimal tours for a fixed-wing UAV that must image multiple ground targets, each with its own required camera azimuth, tilt angle, and dwell time. The authors reformulate the mission as an epsilon-constraint optimization — minimize the closed tour length subject to a bound on the initial maneuver time — and then show that careful discretization of the UAV's configuration space turns this infinite-dimensional problem into a finite graph search. In many instances the discrete problem is exactly a generalized traveling salesperson problem (GTSP), which can be attacked with existing solvers. The paper's central theoretical result is a resolution-completeness theorem: under a non-degeneracy condition and a structural condition on the feasible initial-maneuver set, tour lengths produced by the heuristic converge to the length of an optimal continuous tour as the sampling density tends to infinity. This matters because it provides a principled justification for a heuristic pipeline that is otherwise easy to distrust.

What carries the argument

The central object is a weighted directed graph $G=(V\cup\{v_0\},E,W)$ built from sampled UAV configurations, where each node is a configuration (position and heading) paired with a feasible dwell-time loop at a particular target, and each edge weight is the duration of the dwell loop at the source node plus the optimal Dubins path time from the source to the destination configuration. The named identity doing the work is the augmented Dubins distance; because the Dubins distance satisfies a triangle inequality, the graph weights inherit it, which permits the reduction to a GTSP. The equivalence theorem (Theorem 2) and the resolution completeness theorem (Theorem 3) rest on the structure of the set $\text{INL}_\epsilon$ of configurations reachable from the start within the allowed initial time, together with the continuity of the closed-tour length function away from a finite set of degenerate surfaces.

What would settle it

Run the two-target example of Section 6.3 with the initial-time bound set to 16.26 s, the value at which the reachable set collapses to a single configuration: any sampling grid that omits that exact configuration makes the discrete problem infeasible, so no sequence of tours exists and the claimed resolution completeness fails.

Watch

Extended reading notes

Core claim

The central claim is that the continuous, multi-objective UAV tour problem with visibility and dwell-time constraints can be approximated to arbitrary accuracy by a discrete graph problem. By sampling configurations that are each the start and end point of a feasible dwell-time maneuver, and by weighting edges with the sum of the dwell time at the source and the optimal Dubins travel time to the destination, the authors obtain a finite graph whose optimal tours approximate those of the continuous problem. The approximation is exact in a well-defined sense: whenever the set of configurations reachable within the allowed initial time is nested either inside or around the dwell-time set of a single target (Theorem 2), optimal solutions of the corresponding GTSP map to optimal solutions of the discrete problem. The resolution completeness theorem (Theorem 3) then shows that, for non-degenerate instances, the length of tours produced by the full Algorithm 5 converges to the continuous optimum as the number of samples per target grows, provided the GTSP is solved exactly at each step.

Load-bearing premise

The convergence theorem assumes that the generalized traveling salesperson problem on the sampled graph is solved to global optimality at every sampling density, but exact solvers are not generally available and the numerical implementation uses a heuristic solver, so the proven guarantee does not apply to the implemented algorithm.

Editorial extensions

If this is right

When target spacing is large, the condition for Theorem 2 is typically satisfied, so optimal GTSP solutions translate directly into optimal solutions of the discrete approximation, and hence into near-optimal continuous tours.
Varying the epsilon bound produces a family of routes that trace an approximate Pareto front trading initial response time against total tour duration, which can be computed with existing GTSP and asymmetric TSP solvers.
The framework reduces to the polygon-visiting Dubins traveling salesperson problem when all dwell-times are zero and epsilon is unconstrained, showing it is a genuine generalization of prior work.
The resolution completeness result guarantees that the discrete approximation faithfully reflects the continuous problem in the limit; any residual suboptimality in practice comes from the finiteness of the sampling grid and from the heuristic rather than exact GTSP solving.
As dwell-time requirements increase, the advantage of the GTSP-based heuristic over greedy nearest-target planning grows, since greedy choices can strand the vehicle far from targets that require multiple loops.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Because the proven convergence requires an exact GTSP solver, a testable extension is to compute exact solutions via branch-and-bound for small target counts and compare the observed convergence rate with the resolution-completeness prediction.
The nesting condition on the reachable set is likely to hold when the initial position is much closer to one target than to any other; in cluttered or symmetric layouts where the condition fails, the heuristic has no such equivalence guarantee, and degraded performance is plausible.
The same sampling-plus-GTSP machinery extends naturally to multiple vehicles or to targets with time windows, but the equivalence theorems would need to be re-derived, since the initial-maneuver constraint would couple vehicles.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Worth a read if you work on Dubins routing or sampling-based planning. The paper does something concrete: it takes the polygon-visiting Dubins TSP from Obermeyer et al., adds per-target visibility regions, dwell-time loops, and an epsilon-constraint on the initial maneuver, and shows how to approximate the continuous problem by a GTSP. The graph construction is careful, Theorems 1 and 2 are correct and clean, and the numerical study gives sensible evidence that the method works. The connection to prior work is honest—Remark 3 states clearly when the problem reduces to [7]. That part of the paper is solid.

The soft spots are in the resolution-completeness claim. The reader's report flags the exact-GTSP issue: Theorem 3 assumes Algorithm 4 solves the GTSP to global optimality, while the implementation uses LKH. The appendix concedes this, but the abstract still promises resolution completeness without qualifiers. That is a real mismatch.

There is a sharper problem that the report doesn't capture. The proof of Theorem 3 argues that because sampling is dense, some sample will fall inside a neighborhood U of the optimal first configuration. That step only works if U contains an open ball inside INL_epsilon. If the optimal first configuration lies on the boundary of INL_epsilon—for instance, if epsilon equals the minimum Dubins distance and the minimizer set is a curve—then INL_epsilon has empty relative interior there. A dense sampling sequence can be chosen to miss that lower-dimensional set entirely, so the discrete INL_epsilon is empty for every N and Algorithm 5 returns 'infeasible.' Non-degeneracy (Definition 4) only rules out isolated points, not boundary points on a continuum, so the theorem's hypotheses don't prevent this. The proof is not just incomplete in practice; the stated theorem is false as written. The authors do say at the end of the appendix that Theorem 3 should be viewed as intuition, but the main text and abstract don't carry that caveat.

Bottom line: the framework is worth keeping, and most of the mathematics is honest. But the paper needs a corrected Theorem 3—either add a condition that the optimal first component lies in the interior of INL_epsilon, or downgrade the claim to a conjecture—and the abstract should match what the implemented heuristic actually guarantees. I'd send it to a serious referee, but I'd ask them to focus on the appendix. Also, the authors don't provide code or data; for a sampling-based method that would help a lot.

For a reading group: maybe, if you want a concrete example of how sampling-based proofs can slip on boundary cases. I'd recommend a cautious revise.

Referee Report

3 major / 4 minor

Summary. The paper proposes a sampling-based framework for planning a fixed-wing UAV surveillance tour under per-target visibility and dwell-time constraints. A bi-objective formulation (closed-tour time and initial-maneuver time) is scalarized via an epsilon-constraint method, then discretized by sampling the UAV configuration space at configurations that support feasible dwell maneuvers. The resulting discrete problem (Problem 2) is solved by a heuristic reduction to a Generalized Traveling Salesperson Problem (GTSP) via Algorithm 4. Theorems 1 and 2 state, respectively, that Algorithm 4 always returns a feasible solution and that, under structural conditions on the initially reachable node set, optimal GTSP solutions map to optimal solutions of Problem 2. Theorem 3 claims resolution completeness of the full construction under non-degeneracy, dense sampling, and exact GTSP solving. The paper closes with numerical examples illustrating Pareto-front behavior, comparisons against a greedy baseline, and a hand-constructed convergence example.

Significance. If the advertised convergence property held for the implemented algorithm, the paper would be a useful extension of the prior Polygon-Visiting Dubins TSP sampling framework to dwell-time behaviors and a second objective. The problem formulation is careful and the modular reduction to GTSP is practically appealing; Theorem 2's equivalence result is clean and the numerical study is transparent about the solvers used. However, the central resolution-completeness claim is substantially weaker than the abstract suggests: Theorem 3 requires an exact GTSP oracle that the implementation does not provide, and its denseness argument has a gap for feasible sets of empty relative interior. The paper therefore needs either a repaired theorem or a reframed set of claims before it can be accepted.

major comments (3)

[Appendix, Theorem 3 proof] The step "Since the sampling procedure is dense, for some N, there will be a discrete node placed inside of the set U" is not justified. Dense sampling is defined on each DWL_j, but the first component of the required discrete node must lie in DWL_j∩INL_epsilon. If INL_epsilon∩DWL_j has empty relative interior—for example if epsilon equals the minimum of DIST(v0,·) over a positive-dimensional continuum in DWL_j—then a dense sequence in DWL_j can be chosen that avoids this set entirely. In that case Algorithm 4 returns "Problem 2 Infeasible" for every N, contradicting the claimed feasibility of each INL_MNVRN. Definition 4 only excludes a first component that is isolated in INL_epsilon; it does not exclude a non-isolated point lying on a positive-codimension reachable set. The theorem needs an additional relative-interior assumption around the optimal first component, or the conclusion must be weakened to approximate feasibility with an initial-maneuver slack that tends to zero.
[Theorem 3 and Section 6 (experimental setup)] The statement of Theorem 3 explicitly assumes "(iii) an optimal GTSP solution is found (Algorithm 4, line 5)", but the numerical implementation in Section 6 solves GTSPs through a Noon-Bean transformation followed by the Lin-Kernighan heuristic (LKH). The Appendix itself concedes that "it is not generally possible to find optimal solutions to GTSPs" and that the theorem's utility is mainly to provide intuition. Consequently, the proven result applies to an idealized algorithm with an exact GTSP oracle, not to the Algorithm 5 as implemented. The abstract's claim that the heuristic "is shown to have resolution completeness properties" should be qualified accordingly, or a separate convergence statement for the heuristic should be supplied.
[Section 6.3 and Definition 4] The authors' own example at epsilon=16.26 s shows that when the feasible initial set is a single configuration, Algorithm 5 is not resolution complete, and they argue this is a degenerate case. This is an isolated-point degeneracy, which Definition 4 does exclude. However, it does not address the non-isolated lower-dimensional case in the first major comment, where no point is isolated in INL_epsilon but the intersection DWL_j∩INL_epsilon still has empty relative interior. The numerical evidence in Figure 9 is limited to one hand-constructed nondegenerate instance, so it provides only weak support for the claim that the theorem's conditions are typically satisfied in practice.

minor comments (4)

[Abstract and Section 5.1] The phrase "in many cases, this approximation is equivalent to a generalized traveling salesperson problem" is imprecise: the discrete problem is solved by a heuristic reduction to a GTSP, and equivalence to Problem 2 holds only under the conditions of Theorem 2. Please adjust the wording to avoid overstating the scope of the equivalence.
[Remark 8] The notation "ÐM j=1[ϵ_j, ϵ_j]" appears to be a rendering artifact; it should presumably be a union of intervals, e.g., ∪_{j=1}^{M}[ϵ_j, ϵ_j].
[Figure 7] The y-axis label "Closed Trajectory Time Error (s)" in Figure 7 is misleading: the plotted quantity is the difference between the greedy method and Algorithm 5, not an error with respect to an optimal solution. Consider renaming it "Difference in closed trajectory time" or "Improvement over greedy."
[Algorithm 4, line 3] It would clarify the procedure to state explicitly that the subset INL*_epsilon is chosen nonempty; otherwise the GTSP instance constructed in line 5 may have an empty node set for the selected target.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity found; the derivation is self-contained and the noted proof gaps are correctness concerns, not input-output equivalences.

full rationale

The derivation chain is self-contained: Problem 1 is a continuous Dubins routing optimization, Problem 2 is an independent discretization with sampled nodes and Dubins-distance edge weights, and Theorem 2 establishes an equivalence between Problem 2 and a GTSP under structural conditions on the reachable set rather than defining the two problems to be identical. Theorem 3's resolution-completeness proof uses a standard density-and-continuity argument and does not fit or rename any parameter. The appendix explicitly limits Theorem 3 to the idealized case where an optimal GTSP solution is available and concedes that 'it is not generally possible to find optimal solutions to GTSPs'; this is an honest limitation, not a circular prediction. The skeptical concern that dense sampling in DWL_j may miss a lower-dimensional INL_epsilon is a potential gap in the proof of Theorem 3, but a proof gap is not circularity: no equation is identified with its own input, and no fitted parameter is renamed as a prediction. Self-citations (e.g., [15], [24]) appear only in the literature review and are not load-bearing. The dependence on the external prior work [7] for continuity of LGTH is a citation of independent, non-author prior work used as a lemma, not as the theorem's conclusion. Score 0.

Assumptions & free parameters 2 free parameters · 7 assumptions · 0 invented entities

The central result relies on standard vehicle and imaging modeling assumptions, a specific circular-loop model for dwell behavior, and external continuity and triangle-inequality results. The most significant added assumption is exact GTSP optimality for the resolution-completeness theorem, which is not met by the LKH-based implementation and is acknowledged by the authors.

free parameters (2)

epsilon (initial maneuver time bound) = varied from about 16 to 205 seconds in the examples
User-chosen bound in Problem 1 that controls the tradeoff between initial maneuver time and closed-tour time; varied to construct approximate Pareto fronts. It is not fitted to data but is a hand-chosen parameter of the method.
sampling density parameters (N, or grid spacings delta_r, delta_theta, delta_alpha) = N not fixed in examples; spacing conditions 1 through 7 use delta_r from 1000 m to 125 m and angles from pi to pi/8
Algorithm 2 samples N configurations per target; the numerical examples replace N with a grid whose spacings are chosen by hand. Finer spacing improves tour quality, and the resolution-completeness theorem requires the number of samples to tend to infinity.

assumptions (7)

domain assumption The UAV is modeled as a Dubins vehicle moving in a plane at fixed altitude and speed with a minimum turning radius r.
Section 3.1 defines the vehicle model; all path costs use optimal Dubins curves, citing [37].
domain assumption Imaging parameters are chosen so that at least one feasible dwell-time maneuver exists at each target inside its visibility region.
Stated in Section 3.2 before Problem 1 as a feasibility assumption for the whole framework.
domain assumption Nontrivial dwell-time imaging is performed by flying complete circles of radius r around a pivot point inside the visibility region, or around the target for FULL behavior, with the entire loop inside the visibility region.
Section 3.2 defines DWL_j and restricts the solution space to circular loiter patterns; this is the paper's specific modeling of dwell maneuvers.
standard math The tour-length function LGTH is continuous except on a finite set of (3M-1)-dimensional smooth surfaces, with one-sided continuity up to each surface.
Invoked in Definition 4 and the proof of Theorem 3 in the Appendix, and attributed to [7].
standard math The optimal Dubins path length satisfies a triangle inequality.
Used in Remark 5 to equate the exact-one-node and at-least-one-node GTSP formulations, cited to [29].
ad hoc to paper Algorithm 4, line 5, returns a globally optimal GTSP solution at every sampling density N.
The proof of Theorem 3 explicitly says 'Since the GTSP is solved exactly'; the numerical implementation uses the LKH heuristic, and the appendix concedes that exact GTSP solutions are generally unavailable.
ad hoc to paper The Problem 1 instance is non-degenerate in the sense of Definition 4.
Theorem 3 assumes non-degeneracy; degenerate instances, such as a unique admissible first configuration, break the convergence argument, as shown for epsilon = 16.26 s in Section 6.3.

how reviews work

0 comments

Cite this review

Pith. "Pith review of UAV Surveillance Under Visibility and Dwell-Time Constraints: A Sampling-Based Approach." pith.science (2026). https://pith.science/paper/5JA7SYEZ

@misc{pith2026190805347,
  author       = {Pith},
  title        = {Pith review of: UAV Surveillance Under Visibility and Dwell-Time Constraints: A Sampling-Based Approach},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5JA7SYEZ}},
  note         = {Machine review of arXiv:1908.05347}
}

read the original abstract

A framework is introduced for planning unmanned aerial vehicle flight paths for visual surveillance of ground targets, each having particular viewing requirements. Specifically, each target is associated with a set of imaging parameters, including a desired (i) tilt angle, (ii) azimuth, with the option of a 360-degree view, and (iii) dwell-time. Tours are sought to image the targets, while minimizing both the total mission time and the time required to reach the initial target. An epsilon-constraint scalarization is used to pose the multi-objective problem as a constrained optimization, which, through careful discretization, can be approximated as a discrete graph-search. It is shown that, in many cases, this approximation is equivalent to a generalized traveling salesperson problem. A heuristic procedure for solving the discrete approximation and recovering solutions to the full routing problem is presented, and is shown to have resolution completeness properties. Algorithms are illustrated through numerical studies.

Figures

Figures reproduced from arXiv: 1908.05347 by the authors.

**Figure 2.** An example visibility region VISj when BEHj , ANGLE (left), and when BEHj = ANGLE (right). two angles (lower, upper angular limits). Algorithm 1: Visibility Region Construction Input : a; φ T j , φA j , ∆ T j , ∆ A j for each j ∈ {1, . . ., M } Output : {VISj }j∈{1, . . ., M} for Each Tj do if BEHj , ANGLE then 1 Define VISj as the annulus in R 2 centered at tj with radial limits a/tan(φ T j ± ∆ T j ) else 2 Define … view at source ↗

**Figure 3.** Example imaging behaviors at target Tj for various choices of BEHj and τj . BEHj = FULL, τj != 0 BEHj = ANGLE, τj != 0 BEHj = ANY, τj != 0 BEHj = ANGLE, τj = 0 BEHj = ANY, τj = 0 [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

Figures from the paper (5 more)

**Figure 4.** Figure 4: Examples of valid configuration samples associated with [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]

**Figure 5.** Figure 5: Example of when INL (blue nodes) satisfies condition 1 (left) and 2 (right) of Theorem 2. Proof. The GTSP solution (line 5) will contain some v ∈ INL∗ ⊆ INL . Thus, the permutation operation in line 6 will produce v1, v2, . . ., vM with v1 ∈ INL . It follows readi…

**Figure 6.** Figure 6: Approximate Pareto-Optimal Front and Example Routes for the Spacing Conditions in Table 2 [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]

**Figure 7.** Figure 7: Relative performance of the greedy algorithm. [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]

**Figure 9.** Figure 9: Relative cost error when = 130 s for the spacing conditions in [PITH_FULL_IMAGE:figures/full_fig_p010_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

38 extracted references · 38 canonical work pages

[7]

Sampling-based path planning for a visual reconnais- sance UAV

Obermeyer, K. J., Oberlin, P., and Darbha, S., 2012. “Sampling-based path planning for a visual reconnais- sance UAV”.AIAA Journal of Guidance, Control, and Dynamics, 35(2), pp. 619–631

work page 2012
[1]

Special issue on uninhabited aerial vehicles

Roberts, J., 2006. “Special issue on uninhabited aerial vehicles”.Journal of Field Robotics,23(3–4)

work page 2006
[2]

P., 2008.Advances in Unmanned Aerial Vehicles: State Of The Art And the Road To Autonomy

Valavanis, K. P., 2008.Advances in Unmanned Aerial Vehicles: State Of The Art And the Road To Autonomy. Springer

work page 2008
[3]

Report on technology hori- zons, a vision for Air Force Science And Technol- ogy during 2010–2030

US Air Force, 2010. Report on technology hori- zons, a vision for Air Force Science And Technol- ogy during 2010–2030. Tech. rep., AF/ST-TR-10- 01-PR, United States Air Force. Retrieved from http://www.defenseinnovationmarketplace.mil/ resources/AF_TechnologyHorizons2010-2030.pdf on Feb. 8, 2016

work page 2010
[4]

USOﬃceoftheSecretaryofDefense,2005.Unmanned aircraft systems (UAS) roadmap, 2005-2030

work page 2005
[5]

Attack reconnaissance heli- copter operations

Army, U., 2007. Attack reconnaissance heli- copter operations. Tech. Rep. FM 3-04.126, De- partment of the Army (US), Feb. Retrieved from http://usacac.army.mil/sites/default/ﬁles/misc/doctrine/ CDG/fms.html on Dec. 5, 2016

work page 2007
[6]

Nonlinear Multiobjective Opti- mization

Miettinen, K., 1998. Nonlinear Multiobjective Opti- mization. Springer

work page 1998
[8]

A resourceallocationalgorithmformulti-vehiclesystems with non holonomic constraints

Rathinam, S., Sengupta, R., and Darbha, S., 2007. “A resourceallocationalgorithmformulti-vehiclesystems with non holonomic constraints”.IEEE Transactions onAutomationSciencesandEngineering, 4(1),pp.98– 104

work page 2007

Show all 38 references

[9]

Decentralized perimeter surveillance using a team of UAVs

Kingston, D. B., Beard, R. W., and Holt, R. S., 2008. “Decentralized perimeter surveillance using a team of UAVs”. IEEE Transactions on Robotics, 24(6), pp. 1394–1404

2008
[10]

Cooperative surveillance and pursuit using unmanned aerial vehicles and unattended ground sensors

Las Fargeas, J., Kabamba, P., and Girard, A., 2015. “Cooperative surveillance and pursuit using unmanned aerial vehicles and unattended ground sensors”.Sen- sors,15(1), pp. 1365–1388

2015
[11]

Trackingmultiple groundtargetsinurbanenvironmentsusingcooperating unmanned aerial vehicles

Shaferman,V.,andShima,T.,2015. “Trackingmultiple groundtargetsinurbanenvironmentsusingcooperating unmanned aerial vehicles”.Journal of Dynamic Sys- tems, Measurement, and Control,137(5), pp. 051010– 051010–11

2015
[12]

Guest editorial introduction to the special issue on multivehicle systems cooperative control with applica- tion

Rabbath, C. A., Su, C. Y., and Tsourdos, A., 2007. “Guest editorial introduction to the special issue on multivehicle systems cooperative control with applica- tion”.IEEE Transactions on Control Systems Technol- ogy,15(4), pp. 599–600

2007
[13]

Tools and techniques for mobile sensor network con- trol

Hedrick, J., Basso, B., Love, J., and Lavis, B., 2011. “Tools and techniques for mobile sensor network con- trol”.Journal of Dynamic Systems, Measurement, and Control,133(2), pp. 024001–024001–7

2011
[14]

P., 2007

Gutin, G., and Punnen, A. P., 2007. The Traveling Salesman Problem and Its Variations. Springer

2007
[15]

Traveling Salesperson Problems for the Dubins vehicle

Savla, K., Frazzoli, E., and Bullo, F., 2008. “Traveling Salesperson Problems for the Dubins vehicle”.IEEE Transactions on Automatic Control,53(6), pp. 1378– 1391

2008
[16]

J., Savla, K., Frazzoli, E., and Bullo, F.,

Enright, J. J., Savla, K., Frazzoli, E., and Bullo, F.,
[17]

An approximation algorithm for a symmetric generalized multiple depot, multiple travelling salesman problem

Malik, W., Rathinam, S., and Darbha, S., 2007. “An approximation algorithm for a symmetric generalized multiple depot, multiple travelling salesman problem”. Operations Research Letters,35(6), pp. 747–753

2007
[18]

Stability of solutions to classes of traveling salesman problems

Niendorf, M., Kabamba, P. T., and Girard, A. R., 2016. “Stability of solutions to classes of traveling salesman problems”.IEEE Transactions on Cybernetics,46(4)

2016
[19]

A robust ap- proach to the UAV task assignment problem

Alighanbari, M., and How, J. P., 2008. “A robust ap- proach to the UAV task assignment problem”.Interna- tionalJournalonRobustandNonlinearControl, 18(2), pp. 118–134

2008
[20]

An eﬀective implementation of theLin–Kernighantravelingsalesmanheuristic

Helsgaun, K., 2000. “An eﬀective implementation of theLin–Kernighantravelingsalesmanheuristic”.Euro- peanJournalofOperationalResearch, 126(1),pp.106– 130

2000
[21]

Stochastic surveillance strategies for spatial quickest detection

Srivastava, V., Pasqualetti, F., and Bullo, F., 2013. “Stochastic surveillance strategies for spatial quickest detection”.InternationalJournalofRoboticsResearch, 32(12), pp. 1438–1458

2013
[22]

Robotic surveillanceandMarkovchainswithminimalweighted Kemeny constant

Patel, R., Agharkar, P., and Bullo, F., 2015. “Robotic surveillanceandMarkovchainswithminimalweighted Kemeny constant”.IEEE Transactions on Automatic Control,60(12), pp. 3156–3167

2015
[23]

Strategiesandtech- niques for node placement in wireless sensor networks: A survey

Younis,M.,andAkkaya,K.,2008.“Strategiesandtech- niques for node placement in wireless sensor networks: A survey”.Ad Hoc Networks,6(4), pp. 621–655

2008
[24]

Dynamic vehicle routing for robotic sys- tems

Bullo,F.,Frazzoli,E.,Pavone,M.,Savla,K.,andSmith, S. L., 2011. “Dynamic vehicle routing for robotic sys- tems”.ProceedingsoftheIEEE, 99(9),pp.1482–1504

2011
[25]

Multirobot rendezvous planning for recharging in per- sistent tasks

Mathew, N., Smith, S. L., and Waslander, S. L., 2015. “Multirobot rendezvous planning for recharging in per- sistent tasks”.IEEE Transactions on Robotics,31(1), pp. 128–142

2015
[26]

Gossip cov- erage control for robotic networks: Dynamical systems on the space of partitions

Bullo, F., Carli, R., and Frasca, P., 2012. “Gossip cov- erage control for robotic networks: Dynamical systems on the space of partitions”.SIAM Journal on Control and Optimization,50(1), pp. 419–447

2012
[27]

Dividing a territory among several vehicles

Carlsson, J. G., 2012. “Dividing a territory among several vehicles”.INFORMS Journal on Computing, 24(4), pp. 565–577

2012
[28]

Pathplanningforauavperform- ingreconnaissanceofstaticgroundtargetsinterrain

Obermeyer,K.,2009. “Pathplanningforauavperform- ingreconnaissanceofstaticgroundtargetsinterrain”.In AIAA Guidance, Navigation, and Control Conference, pp. 10–13

2009
[29]

Dubins traveling salesman problem with neighborhoods: A graph-based approach

Isaacs, J., and Hespanha, J. P., 2013. “Dubins traveling salesman problem with neighborhoods: A graph-based approach”.Algorithms,6(1), pp. 84–99

2013
[30]

PlanningAlgorithms

LaValle,S.M.,2006. PlanningAlgorithms. Cambridge University Press

2006
[31]

A transformation for a heterogeneous, multiple depot, multiple traveling salesmen problem

Oberlin, P., Rathinam, S., and Darbha, S., 2009. “A transformation for a heterogeneous, multiple depot, multiple traveling salesmen problem”. In American Control Conference, pp. 1292–1297

2009
[32]

A Lagrangian based approach for the asymmetric generalized travel- ing salesman problem

Noon, C. E., and Bean, J. C., 1991. “A Lagrangian based approach for the asymmetric generalized travel- ing salesman problem”.Operations Research,39(4), pp. 623–632

1991
[33]

Arandom-key geneticalgorithmforthegeneralizedtravelingsalesman problem

Snyder,L.V.,andDaskin,M.S.,2006. “Arandom-key geneticalgorithmforthegeneralizedtravelingsalesman problem”.European Journal of Operational Research, 174(1), pp. 38–53

2006
[34]

Survey of multi-objectiveoptimizationmethodsforengineering

Marler, R. T., and Arora, J. S., 2004. “Survey of multi-objectiveoptimizationmethodsforengineering”. Structural and multidisciplinary optimization, 26(6), pp. 369–395

2004
[35]

Many-objective optimization: An engineering design perspective

Fleming,P.J.,Purshouse,R.C.,andLygoe,R.J.,2005. “Many-objective optimization: An engineering design perspective”. In Evolutionary Multi-Criterion Opti- mization,C.A.C.Coello,A.H.Aguirre,andE.Zitzler, eds. Springer, pp. 14–32

2005
[36]

Formulation and solution of the target visitation problem

Grundel, D., and Jeﬀcoat, D., 2004. “Formulation and solution of the target visitation problem”. In Proceed- ingsoftheAIAA1stIntelligentSystemsTechnicalCon- ference. AIAA 2004-6212

2004
[37]

Oncurvesofminimallengthwith a constraint on average curvature and with prescribed initial and terminal positions and tangents

Dubins,L.E.,1957. “Oncurvesofminimallengthwith a constraint on average curvature and with prescribed initial and terminal positions and tangents”.American Journal of Mathematics,79, pp. 497–516

1957
[2009]

Stochastic and dynamic routing problems for multiple UAVs

“Stochastic and dynamic routing problems for multiple UAVs”.AIAA Journal of Guidance, Control, and Dynamics,34(4), pp. 1152–1166

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [7]

Sampling-based path planning for a visual reconnais- sance UAV

Obermeyer, K. J., Oberlin, P., and Darbha, S., 2012. “Sampling-based path planning for a visual reconnais- sance UAV”.AIAA Journal of Guidance, Control, and Dynamics, 35(2), pp. 619–631

work page 2012

[2] [1]

Special issue on uninhabited aerial vehicles

Roberts, J., 2006. “Special issue on uninhabited aerial vehicles”.Journal of Field Robotics,23(3–4)

work page 2006

[3] [2]

P., 2008.Advances in Unmanned Aerial Vehicles: State Of The Art And the Road To Autonomy

Valavanis, K. P., 2008.Advances in Unmanned Aerial Vehicles: State Of The Art And the Road To Autonomy. Springer

work page 2008

[4] [3]

Report on technology hori- zons, a vision for Air Force Science And Technol- ogy during 2010–2030

US Air Force, 2010. Report on technology hori- zons, a vision for Air Force Science And Technol- ogy during 2010–2030. Tech. rep., AF/ST-TR-10- 01-PR, United States Air Force. Retrieved from http://www.defenseinnovationmarketplace.mil/ resources/AF_TechnologyHorizons2010-2030.pdf on Feb. 8, 2016

work page 2010

[5] [4]

USOﬃceoftheSecretaryofDefense,2005.Unmanned aircraft systems (UAS) roadmap, 2005-2030

work page 2005

[6] [5]

Attack reconnaissance heli- copter operations

Army, U., 2007. Attack reconnaissance heli- copter operations. Tech. Rep. FM 3-04.126, De- partment of the Army (US), Feb. Retrieved from http://usacac.army.mil/sites/default/ﬁles/misc/doctrine/ CDG/fms.html on Dec. 5, 2016

work page 2007

[7] [6]

Nonlinear Multiobjective Opti- mization

Miettinen, K., 1998. Nonlinear Multiobjective Opti- mization. Springer

work page 1998

[8] [8]

A resourceallocationalgorithmformulti-vehiclesystems with non holonomic constraints

Rathinam, S., Sengupta, R., and Darbha, S., 2007. “A resourceallocationalgorithmformulti-vehiclesystems with non holonomic constraints”.IEEE Transactions onAutomationSciencesandEngineering, 4(1),pp.98– 104

work page 2007