Pith. sign in

REVIEW 3 major objections 7 minor 32 references

RCOA lets quadrotor NMPC avoid obstacles and thread narrow slits even when the horizon never reaches them, at over 30 Hz.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.5

2026-07-12 06:58 UTC pith:MXQW2B2R

load-bearing objection Solid 3D/UAV engineering extension of the authors' prior RCOA; the short-horizon narrow-passage demo is useful, but the horizon-independence story still rides on an unextended 2D KKT result. the 3 major comments →

arxiv 2607.02797 v1 pith:MXQW2B2R submitted 2026-07-02 eess.SY cs.SY

RCOA Extension and Applications

Ricardo Tapia , Iman Soltani This is my paper
classification eess.SY cs.SY
keywords obstacle avoidancemodel predictive controlquadrotorUAV navigationconvex optimizationvolumetric collisionprediction horizon
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

This paper extends Relaxed Convex Obstacle Avoidance (RCOA) from 2D into 3D UAV flight and adds multi-point vehicle geometry so a rotating quadrotor body, not a point mass, can be kept clear of prismatic obstacles. The authors argue that RCOA's continuous penalty, derived from optimality conditions, still supplies an anticipatory guidance force that turns the vehicle before an obstacle ever enters the finite prediction window. In simulations they show the nonconvex multi-point version (RNCOA) solves as fast or faster than standard ellipsoidal and dual-set collision constraints, and that a short-horizon NMPC using RNCOA can roll a quadrotor through a slit only centimeters wider than the vehicle while successive open-loop predictions stop short of the slit. The practical claim is that local planners no longer have to stretch the horizon (and the compute budget) just to see obstacles in time.

Core claim

Extending RCOA to three dimensions with a multi-point vehicle footprint preserves its defining property: the KKT-derived continuous penalty regulates the vehicle away from an obstacle even when that obstacle lies outside every open-loop prediction horizon, enabling a closed-loop NMPC to execute aggressive high-roll passage maneuvers at mean solve times of 11.3 ms while matching or beating the latency of standard volumetric OA formulations.

What carries the argument

RNCOA: axis-split big-M constraints applied to the min/max extents of a set of vehicle vertices, plus a linear penalty on the continuous indicators; the resulting optimality relation (generalizing the 2D KKT result) supplies horizon-independent anticipatory clearance.

Load-bearing premise

The 2D optimality relation that forces early clearance still governs the nonconvex multi-point 3D formulation under short horizons, relaxed solver tolerances, and hand-tuned big-M and penalty weights.

What would settle it

Run the identical narrow-passage NMPC with the RCOA penalty weight set to zero (or with pure hard spatial constraints of matching geometry) and check whether the early roll still appears before any prediction node reaches the slit; absence of anticipatory roll would falsify the claimed horizon-independent mechanism.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 7 minor

Summary. The paper extends the authors’ prior Relaxed Convex Obstacle Avoidance (RCOA) method from 2D point-mass settings to 3D UAV navigation. It introduces a multi-point vehicle-geometry variant (RNCOA) that replaces per-vertex big-M constraints with max/min extent conditions, models obstacles as rectangular prisms, and embeds the formulation in a quadrotor NMPC. Two simulation studies are reported: (i) open-loop OCP latency comparisons against point-to-ellipsoid (EOA), ellipsoid–ellipsoid (EEOA), and dual minimum-distance (DMDOA) baselines across several volumetric footprints and topological corridors; and (ii) a closed-loop narrow-slit maneuver with a severely truncated horizon (Tp = 0.25 s, N = 9) in which RNCOA produces anticipatory high-roll trajectories at mean solve times of ~11.3 ms while EOA becomes infeasible. The authors attribute the early roll and horizon-independent behavior to the KKT-derived regulation previously proved for 2D convex RCOA (Eq. 10 / [7, Theorem 1]).

Significance. If the computational and horizon-independence claims hold for rotating 3D vehicle footprints under real-time NMPC, the work addresses a genuine operational tension in agile UAV flight: long horizons for safety versus short horizons for high control rates. The manuscript supplies a concrete multi-point volumetric extension, a reasonably broad baseline comparison (EOA/EEOA/DMDOA), and a closed-loop demo that visually shows successive open-loop horizons terminating before the slit while the vehicle still rolls. These are useful engineering contributions for the NMPC/OA community. The main limitation is that the load-bearing analytical property is inherited from the 2D convex case and is not re-established for the nonconvex multi-point controller actually deployed, so the strongest abstract claim currently rests more on simulation phenomenology than on transferred theory.

major comments (3)
  1. [§III-A Eq. (10); §III-B Eqs. (14); §IV-B Eqs. (18), Figs. 7–8] The central horizon-independence claim (Abstract; §I; §IV-B; Figs. 7–8) is justified by citing the 2D point-mass KKT relation Y* ≥ y_o_max − (M3/M1)(x_o_min − X*) from [7, Theorem 1], restated as Eq. (10) in §III-A. The closed-loop controller that produces the early high-roll is the nonconvex multi-point RNCOA of Eqs. (14) and (18): max/min over rotating body-frame vertices, dual corridor big-M pairs, L1 path costs, hand-tuned M_i and w3 = 190, and solver tolerance relaxed to 10^−1. No stationarity or KKT analysis is given for this setting, so it is not established that the anticipatory roll is caused by the claimed analytical gradient rather than by continuous penalty cost shaping, corridor seeding, or the fixed reference (ry, rz) = (1.1, 2.0). Either extend the optimality analysis to RNCOA under the deployed NMPC, or qualify the claim to simulation evidence only and remove the direct a
  2. [Abstract; §I; §III-B] The abstract and introduction present RCOA as enabling a fully convex OCP for obstacle avoidance, yet the primary 3D volumetric and narrow-passage results use RNCOA, which the authors themselves describe as nonconvex because of the max/min extent operators. The computational and real-time claims therefore rest on a nonconvex formulation whose relationship to the original convex RCOA theory is not fully characterized. Clarify in the abstract and §I which results are strictly convex RCOA versus nonconvex RNCOA, and state explicitly that the >30 Hz narrow-passage demo is nonconvex NMPC.
  3. [§IV-A, Tables VI–VII, Fig. 4] Baseline fairness is only partially controlled. Table VI shows different prediction horizon times (RNCOA/EOA 4.8 s vs EEOA 3.5 s vs DMDOA 3.75 s) “mandated by solver convergence,” DMDOA is forced onto IPOPT+MA57 while others use FATROP, and topological corridors for baselines require hard quadrant restrictions plus seeded initial guesses (§IV-A1). These differences can inflate relative latency and variance for EEOA/DMDOA. Report a common-horizon, common-solver subset (or document failed common settings) so that the median/IQR advantage in Fig. 4 can be attributed primarily to the OA formulation rather than to transcription or solver choice.
minor comments (7)
  1. [Abstract] Abstract grammar: “and apply it to Unmanned Aerial Vehicle (UAV) navigation” should be “and applies it…”.
  2. [Throughout] Inconsistent spacing in “UA V” / “UAV” and “NMPC” throughout; standardize.
  3. [Fig. 4] Figure 4 y-axis label “Time (seconds)” is ambiguous (wall-clock solve time vs horizon length); caption should state “OCP solve time”.
  4. [§IV-B1] Table VIII is referenced as “Table IV-B1” in the text; renumber consistently.
  5. [§III-B Eq. (12)] Eq. (12) writes the SE(3) transform with “ROW” in the rotation block; notation is nonstandard and should be R_OW or similar for readability.
  6. [§IV-B Eq. (18l)] The dual-corridor constraint sum γ_j ≤ 2 in (18l) is stated without discussion of how the bound interacts with the two independent corridor pairs; a short remark would help.
  7. [§II] Related-work coverage of recent real-time quadrotor NMPC OA (e.g., successive convexification / SCP pipelines) is thin; a brief placement relative to those methods would strengthen §II.

Circularity Check

1 steps flagged

Horizon-independence claim imports 2D KKT property via self-citation of authors' prior Theorem 1; empirical NMPC demos and external baselines remain independent.

specific steps
  1. self citation load bearing [§III-A, Eq. (10) and surrounding text; also Abstract and §IV-B attribution]
    "As derived from KKT analysis in [7, Theorem 1], the optimal spatial states satisfy the following relationship: Y∗ ⩾ yo_max − (M3/M1)(xo_min − X∗) (10) This analytical property dictates that as the vehicle approaches the lateral boundary (xo_min), its vertical position is proportionally regulated toward the safe boundary (yo_max). Crucially, this KKT derived regulation acts independently of the OCP prediction horizon."

    The paper's unique claimed characteristic (horizon-independent OA efficacy) is justified solely by restating the authors' own prior Theorem 1 for 2D point-mass RCOA. The key 3D multi-point nonconvex RNCOA controller of (14)/(18) used in the narrow-passage demo receives no independent stationarity analysis, so the attribution of early roll (Fig. 8) to that analytical gradient rests on the self-citation rather than a derivation internal to this manuscript.

full rationale

The paper is primarily an engineering extension with numerical comparisons against independent external OA formulations (EOA from (2), EEOA from (1), DMDOA from (3)). Computational latency results (Fig. 4, Table X) and closed-loop narrow-passage trajectories (Figs. 7–9) are self-contained simulation evidence, not forced by construction from fitted inputs. The sole load-bearing theoretical inheritance is the anticipatory regulation of Eq. (10), restated from the authors' own prior work [7, Theorem 1] for the original 2D point-mass convex RCOA and then attributed to the new nonconvex multi-point RNCOA of (14)/(18) without a fresh KKT derivation. This is ordinary method inheritance for an extension paper, not a definitional loop or a prediction that reduces to its own fit; weights, big-M constants, and corridor choices are free design parameters. Hence only minor self-citation circularity (score 2). No self-definitional equations, no fitted-then-predicted quantities, and no uniqueness theorem used to forbid alternatives appear.

Axiom & Free-Parameter Ledger

4 free parameters · 5 axioms · 1 invented entities

The central claims rest on the prior RCOA construction and its KKT property, standard quadrotor rigid-body dynamics, big-M / penalty design choices, and the multi-point geometric approximation. No new physical entities are postulated; free parameters are controller weights and big-M constants chosen for feasibility and real-time solve quality.

free parameters (4)
  • RCOA/RNCOA penalty weights w, w1, w2, w3
    Hand-tuned cost weights that dominate path error and enforce OA; held constant across geometries but not derived from first principles (§IV-A1, Table VIII).
  • Big-M constants M1–M4
    Sufficiently large constants in the relaxed indicator constraints; set to values such as 50 and 3 in the narrow-passage case without a systematic selection rule (Eqs. 6–8, 14, 17–18; Table VIII).
  • Prediction horizon length Tp and node count N
    Chosen differently per formulation for convergence (Table VI) and aggressively truncated (N=9, Tp=0.25 s) for the real-time claim; directly affects latency and the horizon-independence demonstration.
  • Solver tolerance (down to 1e-1 for real-time NMPC)
    Relaxed from 1e-6 to 1e-1 to prioritize frequency in the narrow-passage experiment; affects reported feasibility and timing (§IV-B2).
axioms (5)
  • ad hoc to paper 2D RCOA formulation and KKT anticipatory property (Y* regulated toward safe boundary as X approaches obstacle edge) from prior work [7, Theorem 1]
    Load-bearing analytical claim for horizon independence is imported from the authors' previous paper and applied without re-derivation to 3D multi-point RNCOA (§III-A, Eq. 10).
  • domain assumption Standard rigid-body quadrotor dynamics with 3-2-1 Euler angles and linear rotor thrust/torque map
    Plant model taken from Beard-style quadrotor equations; assumed accurate enough for the simulated NMPC claims (§III-C, Eqs. 15–16).
  • domain assumption Obstacles as axis-aligned rectangular prisms (or equivalent ellipsoids for baselines) in a known local frame
    OA constraints are defined only for primitive prism/ellipsoid geometry with known pose (§III-B, Eq. 11).
  • ad hoc to paper Multi-point vertex approximation of vehicle volume is sufficient for collision avoidance when max/min extents trigger clearance for all vertices
    RNCOA replaces continuous volumetric non-intersection with discrete points and max/min operators, explicitly sacrificing continuous guarantees (§III-B, Eq. 14).
  • domain assumption Distinct spatial corridors can be selected a priori (or by a global planner) so one parallel OCP topology is solved at a time
    RCOA requires separate OCPs for over/under/left/right; hierarchical selection is assumed available (§III-B, §V).
invented entities (1)
  • RNCOA (nonconvex multi-point RCOA) no independent evidence
    purpose: Reduce constraint/slack growth when applying RCOA to rotating multi-vertex vehicle geometries in 3D NMPC
    Introduced as a modification that uses max/min over vertices and shared gamma variables; independent evidence is only the paper's own simulations, not external validation.

pith-pipeline@v1.1.0-grok45 · 19873 in / 3614 out tokens · 38608 ms · 2026-07-12T06:58:58.766273+00:00 · methodology

0 comments
read the original abstract

The Relaxed Convex Obstacle Avoidance (RCOA) formulation is the first approach to enable a fully convex optimal control problem (OCP) for obstacle avoidance. Convergence analysis of RCOA yields an analytical framework that defines a unique characteristic: the ability to maintain obstacle avoidance (OA) efficacy even when obstacles reside beyond the controller's prediction horizon. In this paper, RCOA is extended to three-dimensional environments and apply it to Unmanned Aerial Vehicle (UAV) navigation. Furthermore, the formulation is enhanced to incorporate vehicle geometries, moving beyond point-mass representations to enable collision avoidance between 3D objects. Numerical simulations demonstrate that RCOA provides computational performance on par or exceeding state-of-the-art methods. Notably, RCOA is demonstrated to enable a Nonlinear Model Predictive Controller (NMPC) to execute aggressive maneuvers through narrow passages with reduced prediction horizons, ensuring real-time feasibility at frequencies exceeding 30~Hz.

Figures

Figures reproduced from arXiv: 2607.02797 by Iman Soltani, Ricardo Tapia.

Figure 1
Figure 1. Figure 1: Typical obstacle with coordinate frame centered and axis normal to [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Volumetric representation of the vehicle: A. Vehicle frame fixed area [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Evaluation scenario of a quadrotor navigating around two joined [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Computational evaluation summary across all designated paths and volumetric vehicle boundaries (including the unconstrained default trajectory). [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 6
Figure 6. Figure 6: Locally optimal 3D trajectories generated by EEOA and EOA [PITH_FULL_IMAGE:figures/full_fig_p007_6.png] view at source ↗
Figure 9
Figure 9. Figure 9: Failed NMPC trajectory utilizing the standard EOA formulation. [PITH_FULL_IMAGE:figures/full_fig_p009_9.png] view at source ↗
Figure 7
Figure 7. Figure 7: Closed-loop RNCOA trajectory navigating a narrow slit. A 2D [PITH_FULL_IMAGE:figures/full_fig_p009_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Sequential NMPC prediction horizons demonstrating receding horizon [PITH_FULL_IMAGE:figures/full_fig_p009_8.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

32 extracted references · 10 canonical work pages · 1 internal anchor

  1. [1]

    Model predictive contouring control for time-optimal quadrotor flight,

    A. Romero, S. Sun, P. Foehn, and D. Scaramuzza, “Model predictive contouring control for time-optimal quadrotor flight,”IEEE Transac- tions on Robotics, vol. 38, 6 2022,ISSN: 19410468.DOI: 10.1109/ TRO.2022.3173711

    arXiv 2022

  2. [2]

    Iapf: An improved artificial potential field framework for asymmetric dual-arm manipulation with real-time inter-arm collision avoidance,

    S. K. S. Prakash, D. Prajapati, B. Narula, and A. Shukla, “Iapf: An improved artificial potential field framework for asymmetric dual-arm manipulation with real-time inter-arm collision avoidance,”Frontiers in Robotics and AI, vol. V olume 12 - 2025, 2025,ISSN: 2296-9144. DOI: 10.3389/frobt.2025.1604506 [Online]. Available: https://www. frontiersin.org/jo...

    work page doi:10.3389/frobt.2025.1604506 2025

  3. [3]

    Dynamic adaptive dynamic window approach,

    M. Dobrevski and D. Sko ˇcaj, “Dynamic adaptive dynamic window approach,”IEEE Transactions on Robotics, vol. 40, pp. 3068–3081, 2024.DOI: 10.1109/TRO.2024.3400932

    doi:10.1109/tro.2024.3400932 2024

  4. [4]

    A hybrid rrt-dwa path planning framework for uavs in dynamic environments,

    Q. Han, X. Ma, J. Liu, H. Liu, Y . Yan, and Q. Yang, “A hybrid rrt-dwa path planning framework for uavs in dynamic environments,”Scientific Reports, vol. 16, p. 3089, 1 2025,ISSN: 2045-2322.DOI: 10.1038/ s41598- 025- 32993- w [Online]. Available: https://doi.org/10.1038/ s41598-025-32993-w

    2025

  5. [5]

    A real-time framework for kinody- namic planning in dynamic environments with application to quadrotor obstacle avoidance,

    R. E. Allen and M. Pavone, “A real-time framework for kinody- namic planning in dynamic environments with application to quadrotor obstacle avoidance,”Robotics and Autonomous Systems, vol. 115, pp. 174–193, 2019,ISSN: 0921-8890.DOI: https://doi.org/10.1016/j. robot.2018.11.017 [Online]. Available: https://www.sciencedirect.com/ science/article/pii/S092188...

    doi:10.1016/j 2019

  6. [6]

    Lqr-rrt*: Optimal sampling-based motion planning with automatically derived extension heuristics,

    A. Perez, R. Platt, G. Konidaris, L. Kaelbling, and T. Lozano-Perez, “Lqr-rrt*: Optimal sampling-based motion planning with automatically derived extension heuristics,” in2012 IEEE International Conference on Robotics and Automation, 2012, pp. 2537–2542.DOI: 10 . 1109 / ICRA.2012.6225177

    arXiv 2012

  7. [7]

    Tapia and I

    R. Tapia and I. Soltani,A convex obstacle avoidance formulation, 2025. arXiv: 2512.13836[eess.SY]. [Online]. Available: https://arxiv.org/ abs/2512.13836

    Pith/arXiv arXiv 2025

  8. [8]

    Constrained iterative lqr for on-road autonomous driving motion planning,

    J. Chen, W. Zhan, and M. Tomizuka, “Constrained iterative lqr for on-road autonomous driving motion planning,” in2017 IEEE 20th International Conference on Intelligent Transportation Systems (ITSC), 2017, pp. 1–7.DOI: 10.1109/ITSC.2017.8317745

    work page doi:10.1109/itsc.2017.8317745 2017

  9. [9]

    Mixed integer programming for multi-vehicle path planning,

    T. Schouwenaars, B. D. Moor, E. Feron, and J. How, “Mixed integer programming for multi-vehicle path planning,” in2001 European Control Conference, ECC 2001, Institute of Electrical and Electronics Engineers Inc., 2001, pp. 2603–2608,ISBN: 9783952417362.DOI: 10. 23919/ecc.2001.7076321

    arXiv 2001

  10. [10]

    Khatib,Real-Time Obstacle Avoidance for Manipulators and Mobile Robots

    O. Khatib,Real-Time Obstacle Avoidance for Manipulators and Mobile Robots. Apr. 1985, vol. 1, pp. 500–505.DOI: 10.1109/ROBOT.1985. 1087247

    work page doi:10.1109/robot.1985 1985

  11. [11]

    Borenstein and Y

    J. Borenstein and Y . Koren,Real-time obstacle avoidance for fast mobile robots in cluttered environments. Jun. 1990, 572–577 vol.1, ISBN: 0-8186-9061-5.DOI: 10.1109/ROBOT.1990.126042

    work page doi:10.1109/robot.1990.126042 1990

  12. [12]

    A fast procedure for computing the distance between complex objects in three-dimensional space,

    E. G. Gilbert, D. W. Johnson, and S. S. Keerthi, “A fast procedure for computing the distance between complex objects in three-dimensional space,”IEEE Journal on Robotics and Automation, vol. 4, pp. 193–203, 2 1988.DOI: 10.1109/56.2083

    doi:10.1109/56.2083 1988

  13. [13]

    Optimized model predictive control-based path planning for multiple wheeled mobile robots in uncertain environments,

    Y . She, C. Song, Z. Sun, and B. Li, “Optimized model predictive control-based path planning for multiple wheeled mobile robots in uncertain environments,”Drones, vol. 9, 1 2025,ISSN: 2504-446X. DOI: 10.3390/drones9010039 [Online]. Available: https://www.mdpi. com/2504-446X/9/1/39

    work page doi:10.3390/drones9010039 2025

  14. [14]

    Mpc control and path planning of omni-directional mobile robot with potential field method,

    Hailin, W. Chengcheng, H. Fang, Y . X. L. Xiaofeng, and Chen, “Mpc control and path planning of omni-directional mobile robot with potential field method,” inIntelligent Robotics and Applications, Alexandre, Y . Yamin, C. S. C. Zhiyong, and Mendes, Eds., Springer International Publishing, 2018, pp. 170–181,ISBN: 978-3-319-97589- 4

    2018

  15. [15]

    3DSSD: Point-based 3D Single Stage Object Detector

    L. Ros, A. Sabater, and F. Thomas, “An ellipsoidal calculus based on propagation and fusion,”IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics), vol. 32, pp. 430–442, 4 2002.DOI: 10.1109/TSMCB.2002.1018763

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1109/tsmcb.2002.1018763 2002

  16. [16]

    Efficient avoidance of ellipsoidal obsta- cles with model predictive control for mobile robots and vehicles,

    M. Rosenfelder, H. Carius, M. Herrmann-Wicklmayr, P. Eberhard, K. Flaßkamp, and H. Ebel, “Efficient avoidance of ellipsoidal obsta- cles with model predictive control for mobile robots and vehicles,” Mechatronics, vol. 110, p. 103 386, Oct. 2025,ISSN: 0957-4158.DOI: 10.1016/J.MECHATRONICS.2025.103386 [Online]. Available: https: //www.sciencedirect.com/sci...

    doi:10.1016/j.mechatronics.2025.103386 2025

  17. [17]

    Leprich, M

    D. Leprich, M. Rosenfelder, M. Herrmann-Wicklmayr, K. Flaßkamp, P. Eberhard, and H. Ebel,Efficient collision-avoidance constraints for ellipsoidal obstacles in optimal control: Application to path-following mpc and uavs, 2025. [Online]. Available: https://arxiv.org/abs/2510. 26531

    2025

  18. [18]

    Differentiable collision detection for a set of convex primitives,

    K. Tracy, T. A. Howell, and Z. Manchester, “Differentiable collision detection for a set of convex primitives,” in2023 IEEE International Conference on Robotics and Automation (ICRA), 2023, pp. 3663–3670. DOI: 10.1109/ICRA48891.2023.10160716

    doi:10.1109/icra48891.2023.10160716 2023

  19. [19]

    Nonlinear model pre- dictive control on se(3) for quadrotor aggressive maneuvers,

    J. C. Pereira, V . J. S. Leite, and G. V . Raffo, “Nonlinear model pre- dictive control on se(3) for quadrotor aggressive maneuvers,”Journal of Intelligent & Robotic Systems, vol. 101, p. 62, 3 2021,ISSN: 1573- 0409.DOI: 10.1007/s10846-021-01310-8 [Online]. Available: https: //doi.org/10.1007/s10846-021-01310-8

    doi:10.1007/s10846-021-01310-8 2021

  20. [20]

    Optimization-based colli- sion avoidance,

    X. Zhang, A. Liniger, and F. Borrelli, “Optimization-based colli- sion avoidance,”IEEE Transactions on Control Systems Technology, vol. 29, pp. 972–983, 3 2021.DOI: 10.1109/TCST.2019.2949540

    work page doi:10.1109/tcst.2019.2949540 2021

  21. [21]

    Morozov et al.,Multi-query shortest-path problem in graphs of convex sets, 2024

    S. Morozov et al.,Multi-query shortest-path problem in graphs of convex sets, 2024. [Online]. Available: https://arxiv.org/abs/2409.19543

    Pith/arXiv arXiv 2024

  22. [22]

    S. Garg, T. Cohn, and R. Tedrake,Planning shorter paths in graphs of convex sets by undistorting parametrized configuration spaces, 2025. [Online]. Available: https://arxiv.org/abs/2411.18913 11

    Pith/arXiv arXiv 2025

  23. [23]

    New distances for the separation and penetration of objects,

    E. G. Gilbert and C. J. Ong, “New distances for the separation and penetration of objects,” inProceedings of the 1994 IEEE International Conference on Robotics and Automation, 1994, 579–586 vol.1.DOI: 10.1109/ROBOT.1994.351237

    work page doi:10.1109/robot.1994.351237 1994

  24. [24]

    A direct method for checking overlap of two hyperellipsoids,

    I. Gilitschenski and U. D. Hanebeck, “A direct method for checking overlap of two hyperellipsoids,” in2014 Sensor Data Fusion: Trends, Solutions, Applications (SDF), 2014, pp. 1–6.DOI: 10.1109/SDF.2014. 6954724

    work page doi:10.1109/sdf.2014 2014

  25. [25]

    S. P. Boyd and L. Vandenberghe,Convex Optimization. Cambridge University Press, 2014,ISBN: 978-0-521-83378-3.DOI: 10 . 1017 / CBO9780511804441 [Online]. Available: https: // web. stanford. edu/ %7Eboyd/cvxbook/

    2014

  26. [26]

    Control of systems integrating logic, dynamics, and constraints,

    A. Bemporad and M. Morari, “Control of systems integrating logic, dynamics, and constraints,”Automatica, vol. 35, pp. 407–427, 1999

    1999

  27. [27]

    Quadrotor dynamics and control,

    R. W. Beard, “Quadrotor dynamics and control,” 2008. [Online]. Available: https://api.semanticscholar.org/CorpusID:195351003

    2008

  28. [28]

    Casadi – a software framework for nonlinear optimization and optimal control,

    J. A. E. Andersson, J. Gillis, G. Horn, J. B. Rawlings, and M. Diehl, “Casadi – a software framework for nonlinear optimization and optimal control,”Mathematical Programming Computation, 2018

    2018

  29. [29]

    Vanroye, A

    L. Vanroye, A. Sathya, J. D. Schutter, and W. Decr ´e,Fatrop : A fast constrained optimal control problem solver for robot trajectory optimization and control, 2023. [Online]. Available: https://arxiv.org/ abs/2303.16746

    Pith/arXiv arXiv 2023

  30. [30]

    On the implementation of an interior- point filter line-search algorithm for large-scale nonlinear program- ming,

    A. W ¨achter and L. T. Biegler, “On the implementation of an interior- point filter line-search algorithm for large-scale nonlinear program- ming,”Mathematical Programming, vol. 106, pp. 25–57, 1 2006,ISSN: 1436-4646.DOI: 10.1007/s10107- 004- 0559- y [Online]. Available: https://doi.org/10.1007/s10107-004-0559-y

    doi:10.1007/s10107- 2006

  31. [31]

    Ma57 - a code for the solution of sparse symmetric definite and indefinite systems,

    I. S. Duff, “Ma57 - a code for the solution of sparse symmetric definite and indefinite systems,”ACM Transactions on Mathematical Software, vol. 30, 2 2004,ISSN: 00983500.DOI: 10.1145/992200.992202

    doi:10.1145/992200.992202 2004

  32. [32]

    Trajectory planning for bertha — a local, continuous method,

    J. Ziegler, P. Bender, T. Dang, and C. Stiller, “Trajectory planning for bertha — a local, continuous method,” in2014 IEEE Intelligent Vehicles Symposium Proceedings, 2014, pp. 450–457,ISBN: 1931- 0587.DOI: 10.1109/IVS.2014.6856581

    work page doi:10.1109/ivs.2014.6856581 2014