{"id":"fdce760f-2d8d-4ce4-81f1-94b858587a20","arxiv_id":"2509.10349","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"Acetrans presents a LiDAR-IMU perception, corridor-based planning, and NMPC control system for cable-suspended UAV transport, validated in simulation and indoor/outdoor experiments.","lead":"A drone with a cable-suspended payload is taught to see the cable, plan a safe path, and fly the whole body through cluttered spaces, day or night. It combines a LiDAR-cable estimator, a corridor-based planner, and a controller so the UAV, payload, and even the bent cable stay inside safe regions.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Bent-cable safety rests on an unverified trapezoid containment assumption; the planner's corridor guarantee covers only the taut-cable case.","rationale":"The reader's weakest-assumption analysis and my stress-test converge on the same point: the taut-cable differential-flatness model underpins the corridor-based whole-body safety guarantee, while the bent-cable trapezoid is a separate, less formal patch. The paper itself says in Sec. III-B that bent-cable trajectory optimization is avoided precisely because the payload becomes uncontrollable, so the NMPC trapezoid is load-bearing for the advertised robust safety under bending. My concern is more specific than the reader's: even if the trapezoid is correctly constructed from the estimated catenary, the estimator's planar-catenary assumption (Sec. IV-B) is not justified for real nonuniform wind, and no experiment measures whether the true cable stays inside T. This is not a manufactured or 'against-consensus' objection; it is an internal gap between assertion and proof. I nonetheless agree with the reader's CONDITIONAL verdict rather than moving to REJECT, because the system has real independent evidence: MuJoCo success rates, indoor/outdoor flights including nighttime, and corridor/perception evaluations. The concrete test would settle whether the safety claim is merely overstated or actually false. If the test shows cable excursions outside T, the central guarantee would be unsupported and a REJECT or major rewriting would be warranted; if it shows zero violations, the conditional concern is resolved.","tokens_in":29573,"tokens_out":3769,"duration_ms":51749,"concrete_test":"In MuJoCo, model the cable as a multi-link chain (or flexible spline) with distributed aerodynamic drag. Run the full Acetrans stack under wind fields that are nonuniform in space and time and not aligned with any single plane (e.g., the ten random 3D wind fields from Sec. VII-B.2, plus a shear-wind case). At each control step, record (a) the maximum distance from any cable point to the estimated trapezoid T and (b) the maximum distance from any cable point to the corridor P. If either distance is positive in any trial before a collision, the B⊆T assumption underlying Eq. (67) and the NMPC safety constraint Eq. (69d) is falsified. Report the fraction of time steps with a violation; the guarantee claim requires exactly 0%.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim—guaranteed whole-body safety under cable bending and external disturbances—depends on an unproven containment step. Planning (Sec. V-D, Eq. 56) constrains only x_Q and x_L to lie in the corridor; convexity then covers the cable only if the cable is straight, i.e., taut. For execution, Sec. VI asserts B⊆T (Eq. 67), where T is the trapezoid built from the catenary fit of Sec. IV-B. That fit assumes the cable lies in a plane and follows a catenary in a uniform resultant acceleration field. Real outdoor wind is nonuniform and time-varying, and cable drag is distributed along the cable, so the actual bent shape can leave the fitted plane and, potentially, leave T. No theorem, simulation, or hardware measurement in the paper quantifies this excursion. Thus the planner's 'guarantee' is contingent on tautness, while the controller's 'guarantee' is contingent on the unverified catenary/trapezoid enclosure. If the enclosure fails, the NMPC constraints (Eq. 69d) are applied to the wrong set, and no safety guarantee remains—only empirical robustness. This is the exact gap between the abstract's strong safety claim and the formal statements, and Sec. V-D itself softens to 'highly likely' for unsampled trajectory states.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents Acetrans, a full autonomy stack for a quadrotor carrying a cable-suspended payload. The system has three main components: (i) a LiDAR-IMU perception module that estimates the payload pose and fits the cable shape in both taut and bent conditions, and filters cable points from the map; (ii) a corridor-based planning module, including the MACIRI algorithm for generating safe flight corridors with multi-size convex-hull seeds, and a MINCO trajectory optimizer that uses differential flatness of the taut-cable dynamics; and (iii) an NMPC controller that incorporates trapezoidal cable-bending envelopes to maintain whole-body safety during execution. The paper claims that this is the first fully autonomous suspended-load framework unifying perception, planning, and control, and that it achieves 1–3 orders of magnitude faster optimization than the Autotrans and Impactor baselines, with 100% success in narrow-gap and thin-pole simulations and successful indoor and nighttime outdoor flights.","tokens_in":29949,"tokens_out":4069,"duration_ms":55113,"significance":"If the claims hold, Acetrans would be a significant systems contribution: it is one of the few suspended-load pipelines that closes the perception-planning-control loop with onboard LiDAR, it extends corridor-based planning to multi-size convex-hull seeds, and it avoids cable sampling by exploiting convexity of the flight corridor. The paper contains useful and largely correct building blocks: the convex-inclusion argument of Section III-C is elementary but load-bearing; the differential-flatness derivation in Section III-B is standard; and the catenary fitting scheme in Section IV-B is a well-posed nested optimization with an explicit reparameterization. The simulation and hardware results, if reproducible, demonstrate a working system. The main weakness is that the paper's central safety claim—guaranteed whole-body safety under cable bending and external disturbances—is not backed by the formal statements in Sections V-D and VI, which rely on an unproven trapezoid-containment assumption and on soft penalty-based constraint enforcement. These gaps are substantial but addressable with additional analysis, experiments, or careful claim softening.","major_comments":[{"comment":"The NMPC safety argument depends on the assertion B ⊆ T, where B is the set of all UAV, payload, and bent-cable positions and T is the trapezoid constructed from the catenary fit of Section IV-B. This containment is asserted, not proved. The catenary model of Section IV-B assumes the cable lies in a plane and follows a catenary in a uniform resultant acceleration field (Eq. (20)). Real wind is nonuniform and time-varying, and aerodynamic drag is distributed along the cable, so the actual bent shape can leave the fitted plane and, potentially, leave T. If T fails to contain B, the constraints (69d) are applied to the wrong set and no safety guarantee remains. The planner's corridor guarantee (Eq. (56)) covers only the straight taut cable by convexity. Please either provide a theorem with explicit assumptions under which B ⊆ T provably holds, quantify the possible excursion with hardware m","section":"Section VI, Eq. (67)"},{"comment":"The paper claims in the Introduction and contributions that the planner guarantees whole-body safety, but the constraints are enforced as time-integral penalties with finite quadrature, not as hard constraints (Eqs. (53)-(55) and (62)). The text itself admits at the end of Section V-D that unsampled trajectory states are only 'highly likely' to be contained in the chain of convex hulls. Thus the formal statement does not match the 'guarantee' language. If a hard safety guarantee is intended, the trajectory should satisfy A_i x_Q(t) ≤ b_i and A_i x_L(t) ≤ b_i for all t, not only at quadrature points; otherwise the safety claim should be explicitly probabilistic or empirical.","section":"Section V-D, Eqs. (52)-(56)"},{"comment":"The catenary model assumes that the wind field exerts a force that is equal in magnitude and parallel on every infinitesimal cable segment, so the cable remains a catenary in a uniform resultant field. This is a strong modeling assumption: aerodynamic drag on a cable depends on the local relative velocity, the cable orientation, and the local wind field, so a nonuniform or unsteady wind will not produce a uniform resultant acceleration. The MuJoCo simulation in Section VII-B.2 provides some evidence for the fitting accuracy under the tested conditions, but it does not validate the containment assumption for outdoor wind. Please state the domain of validity of the catenary assumption and provide experimental evidence that the trapezoid envelope contains the real cable under the disturbances used in the indoor and outdoor flights.","section":"Section IV-B.3, Eq. (20)"},{"comment":"The differential flatness derivation includes F_Q and F_L as quantities that must be known for the flatness map, but the text then states that they are 'assumed to be constant during planning, such that their derivatives are 0.' This is a significant simplification: if the actual external forces vary along the trajectory, the reconstructed quadrotor position and thrust from the flat outputs will be inaccurate, which can affect both dynamic feasibility and the corridor constraints on x_Q. The paper should quantify the resulting error or justify why constant external forces are sufficient for the claimed safety and tracking performance.","section":"Section III-B, Eqs. (3a)-(3c)"}],"minor_comments":[{"comment":"The text refers to 'Section X' when introducing x_imu_L and x_lidar_L; this should reference Section IV-A.2.","section":"Section IV-A.4"},{"comment":"There are several typos and formatting issues: 'Corrsponding' in the footnote, 'assuption' in Assumption 1, 'recently works' in Section II-C, and 'MIVE' instead of 'MVIE' in Algorithm 1 and surrounding text.","section":"General"},{"comment":"The containment statistics table is difficult to read because the columns are packed without separators; please reformat the table and the surrounding text.","section":"Table IV"},{"comment":"The baseline comparisons for Autotrans involve a 'carefully tuned' additional safety margin of 0.4 m in the thin-pole experiments. This tuning should be described in more detail, including the tuning criterion, so that the comparison is reproducible and not perceived as favorable to the proposed method.","section":"Section VII-D"}],"recommendation":"major_revision","confidential_remarks":"The paper is a strong systems contribution with a convincing amount of experimental work. The revision should focus on aligning the safety claims with the formal content: either prove the trapezoid containment under stated assumptions or remove 'guarantee' from the abstract and contributions. I do not see grounds for rejection, because the perception and corridor-generation components appear sound and the experiments are extensive. However, the current overstatement of the safety guarantee is exactly the kind of claim that a careful reader will challenge, and it should be fixed before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: genuinely useful integration paper, but the safety language oversells what the math actually shows. The planner's whole-body guarantee is taut-cable only; the bent-cable NMPC envelope is an unproven containment step, not a theorem.\n\nWhat's new: MACIRI extends CIRI to multi-size convex-hull seeds with adaptive sphere inflation and a containment correction. That's a real generalization with clean convexity behind it. The outer-inner catenary estimator with the rotated plane and anti-degeneracy term is clever, and the diffeomorphic reparameterizations in the inner fit and the trajectory initialization are standard but well executed. The differential flatness derivation that keeps F_Q and F_L explicit is a legitimate robustness extension over the usual formulation. The full stack—LiDAR-IMU payload estimation, cable point-cloud filtering, corridor planning, NMPC with trapezoidal envelope—is assembled and tested in simulation and in real indoor/nighttime outdoor flights.\n\nWhere it gets soft. The corridor constraint (Eq. 56) binds only x_Q and x_L. The cable's safety follows from convexity only when the cable is straight, i.e., taut. The paper says this—it deliberately doesn't plan in bent mode—but the abstract and Section I still claim 'guarantee whole-body safety under cable bending.' That mismatch needs fixing. On the control side, the trapezoid containment B⊆T (Eq. 67) is asserted without proof. The catenary fit assumes a uniform resultant acceleration field. Real wind is nonuniform and time-varying, cable drag is distributed along the cable, so the actual bent shape can leave the fitted plane and potentially escape T. No theorem, simulation sweep, or hardware measurement bounds that excursion. If the envelope fails, the NMPC constraints apply to the wrong set and 'guarantee' drops to empirical robustness. That's a real gap, not a nitpick.\n\nThe baseline comparisons also need a side-eye: Autotrans's 0.4 m safety margin was tuned by you, and the claimed '1-3 orders of magnitude' speedup is against that specific configuration. The corridor method's qualitative advantage over ESDF-plus-cable-sampling is plausible; the specific number is not independently verified. Also, no code or commit hash, and hardware results are qualitative (no numeric logs). That's a reproducibility problem for a systems paper.\n\nBottom line: worthwhile work with sound engineering and a clean convexity core. It deserves a serious referee, but authors should either prove or bound the bent-cable excursion or soften the guarantee to 'empirically robust under tested disturbances.' I'd support conditional acceptance with major revision.","headline":"Genuinely useful full-stack suspended-load system, but the 'guarantee' holds only for the taut-cable planner; the bent-cable NMPC envelope is an unproven containment step.","tokens_in":30497,"tokens_out":4419,"would_cite":true,"duration_ms":46451,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A unified perception-planning-control stack makes cable-suspended drone transport autonomous, efficient, and safe around the clock, the paper claims.","keywords":["UAV suspended payload","whole-body perception","safe flight corridor","trajectory optimization","NMPC","cable bending","LiDAR-IMU fusion","aerial transport"],"falsifier":"Run the system under a wind field strong enough to bend the cable out of the fitted catenary plane while a motion-capture system tracks the true cable shape; if the true cable exits the trapezoidal envelope used in the NMPC at any time, the claimed whole-body safety guarantee is violated.","tokens_in":29429,"feed_emoji":"🚁","tokens_out":3453,"duration_ms":38749,"temperature":0.7,"pith_summary":"The paper claims that Acetrans is the first fully autonomous suspended-load framework that unifies perception, planning, and control for UAVs hauling cable-slung payloads. It argues that a LiDAR-IMU module can estimate both payload pose and cable shape under taut and bent conditions, a new corridor-generation algorithm (MACIRI) makes whole-body planning one to three orders of magnitude faster than ESDF-based baselines, and an NMPC with trapezoidal cable-bending envelopes maintains safety even when the cable deforms. If correct, this would make round-the-clock autonomous aerial delivery practical in forests, urban canyons, and indoor spaces where manual slung-load flight is too risky.","feed_headline":"Autonomous drone sling transport clears 0.6-m gaps and 1-mm poles","feed_subtitle":"A perception-planning-control pipeline keeps UAV, payload, and cable safe around the clock, even when the cable bends.","key_machinery":"The load-bearing object is the safe flight corridor produced by MACIRI, which accepts point, line, quadrilateral, or tetrahedral seeds and assigns each seed vertex its own obstacle-inflation radius. The convex-hull inclusion property—if all vertices satisfy the corridor's linear inequalities, every interior point does—allows the trajectory optimizer to guarantee whole-body safety while constraining only the quadrotor and payload, not the cable. At control time, a catenary-based estimator with a diffeomorphic reparameterization fits the bent cable in a plane, and the NMPC imposes constraints on just four trapezoid vertices that enclose the estimated cable, keeping the whole body inside the co","core_discovery":"Acetrans demonstrates that safe, efficient suspended-load transport can be achieved by jointly estimating cable and payload states with LiDAR and an end-tether IMU, planning taut-cable trajectories inside safe flight corridors whose convex-hull seeds carry different vertex sizes, and tracking the plan with an NMPC that encloses any bent cable in a conservative trapezoid. The paper reports 100% simulation success in narrow gaps down to 0.6 m and against poles as thin as 1 mm, with optimization speeds 1–3 orders of magnitude faster than Autotrans and Impactor, together with successful indoor and nighttime outdoor flights. Corridor inclusion lets the planner constrain only the quadrotor and pay","pith_inferences":["If the corridor-inclusion safety guarantee is the real source of the speed-up, the same design could transfer to other multi-body robots—such as UAVs towing a hose or manipulators carrying a chain-like load—by generating convex-hull seeds per link and constraining only link vertices.","Because the planner intentionally never triggers bent-cable trajectories, the bending-safety layer is essentially a reactive control patch; a stronger extension would let the planner explicitly reason about inevitable bending, for example by generating a hybrid corridor set that remains feasible when wind exceeds the taut assumption.","The trapezoidal envelope is fitted to a cable that is assumed to lie in a single plane under uniform wind; in real turbulence with out-of-plane bending, the true cable may leave that envelope. A concrete test is to fly in a gusty crosswind while a motion-capture system records the true cable, and check whether the envelope still contains it.","The reported simulation maps are randomly generated with Perlin noise; the closest real-world evidence is a 25-m night forest loop. Scaling to kilometer-length missions would reveal whether MACIRI's corridor generation and the taut-cable planner hold up against rare, strong wind gusts and localization drift."],"forward_implications":["Corridor-based planning avoids ESDF construction, so replanning runs in roughly 1–3 ms, enabling online replanning in large outdoor maps that would be too costly for dense signed-distance fields.","Whole-body safety for thin obstacles comes from convex-hull inclusion rather than dense cable sampling, yielding 100% success on 0.001-m poles without increasing computation time.","LiDAR-based perception removes the lighting dependence of vision-based slung-load estimation, supporting reliable day-and-night operation.","High-acceleration flight under taut-cable control can match the agility of bent-cable maneuvers while retaining full controllability, so the planner deliberately avoids bent-mode trajectories.","The NMPC with trapezoidal cable-bending envelopes provides, to the authors' knowledge, the first control-level obstacle avoidance under cable bending.","The full pipeline is validated in simulation and in indoor and nighttime outdoor flights, including a 25-m looped forest trajectory.","Corridor inclusion guarantees that the continuous trajectory—not just sampled points—remains collision-free, a stronger probabilistic safety assurance than ESDF-based sampling methods.","The corridor-based formulation reduces CPU and memory usage compared to ESDF-based baselines, making real-time performance feasible on an onboard NUC."],"fun_headline_variants":["Drone sling hauling clears 0.6-m gaps and 1-mm poles","Acetrans: autonomous corridor-based UAV transport with safety","Corridor-based UAV sling: efficient and safe transport","UAV sling system: 100% success in tight gaps and thin poles","Acetrans: efficient UAV transport with corridor planning"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The planning and whole-body safety guarantees are derived under the assumption that the cable stays taut throughout flight; if the cable bends in a way the catenary estimator does not capture (for example, out of the fitted plane), the planner's safety guarantee no longer holds and only a soft penalty in the controller remains.","fun_headline_variants_meta":{"raw":{"variants":["Drone sling hauling clears 0.6-m gaps and 1-mm poles","Acetrans: autonomous corridor-based UAV transport with safety","Corridor-based UAV sling: efficient and safe transport","UAV sling system: 100% success in tight gaps and thin poles","Acetrans: efficient UAV transport with corridor planning"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001548,"raw_usage":{"total_tokens":6039,"prompt_tokens":773,"completion_tokens":5266,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":517,"completion_tokens_details":{"reasoning_tokens":5188}},"tokens_in":517,"tokens_out":5266,"duration_ms":39497,"temperature":1.0,"reasoning_tokens":5188,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-04T17:51:24.277041+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the system under a wind field strong enough to bend the cable out of the fitted catenary plane while a motion-capture system tracks the true cable shape; if the true cable exits the trapezoidal envelope used in the NMPC at any time, the claimed whole-body safety guarantee is violated.","supporting_citations":[],"review_version":1}