{"id":"85616489-30d6-4520-a37b-2c948094947e","arxiv_id":"2606.28588","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":4.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper introduces an RCLF-QP controller with online adaptation and an extended state observer that maintains quadrotor stability under lumped disturbances and malicious cyberattacks where a baseline PD controller fails.","lead":"This paper develops a resilient quadratic program controller using Control Lyapunov Functions to stabilize quadrotors against actuator failures, wind disturbances, and unknown cyberattacks on control channels. Smart generalists might read it to see how safety systems for drones can be made more robust to both physical faults and deliberate interference.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.3","headline":"RCLF-QP stability claim for unbounded attacks implicitly requires the adaptation term to remain within actuator limits, but no feasibility analysis is given when attacks grow without bound.","rationale":"The reader's weakest assumption already flags the insertion of an adaptive compensator for unbounded attacks. The concrete concern is the missing feasibility argument under actuator saturation, which directly threatens the claimed stability preservation. This moves the verdict from UNVERDICTED to CONDITIONAL pending explicit checks on unbounded regimes; the simulation evidence remains useful for bounded attacks.","tokens_in":1806,"tokens_out":313,"duration_ms":40736,"concrete_test":"Re-run the high-fidelity simulations with attack signals whose amplitude grows linearly (e.g., attack = 0.5*t on a pseudo-control channel) while recording QP feasibility flag and position error; if the QP becomes infeasible or divergence occurs at finite t, the unbounded-attack guarantee fails.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central construction inserts a real-time adaptive resilient term into the CLF to cancel unknown unbounded attacks on pseudo-control channels. Because the attacks have no a-priori bound, the adaptation must generate arbitrarily large corrections. The QP formulation includes input constraints, yet the paper provides no proof that the QP remains feasible or that the CLF derivative stays negative definite once the required compensation exceeds available control authority. Simulations are reported to succeed, but without explicit attack-magnitude scaling or feasibility monitoring, they do not test the unbounded regime asserted in the abstract.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The paper develops a Resilient Control Lyapunov Function-based Quadratic Program (RCLF-QP) for quadrotors subject to partial actuator failures, lumped disturbances, and malicious unbounded cyberattacks on pseudo-control channels. It augments a conventional CLF with a real-time adaptive resilient compensational term, integrates this with an extended state observer, and formulates the result as a QP that preserves stability guarantees while incorporating input constraints. High-fidelity simulations are reported to show that the RCLF-QP prevents trajectory divergence in attack scenarios where a baseline PD controller fails.","tokens_in":1919,"tokens_out":493,"duration_ms":22466,"significance":"If the central claims hold, the construction supplies a systematic, extensible QP-based method for inserting online adaptation against unbounded attacks into CLF frameworks for underactuated systems without sacrificing the underlying Lyapunov guarantees. The explicit handling of cyberattacks on pseudo-controls, together with the observer, would be a useful addition to the fault-tolerant and resilient control literature for quadrotors.","major_comments":[{"comment":"The abstract and the RCLF-QP construction assert compensation for unknown and unbounded attacks via real-time adaptation, yet the QP formulation (with its input constraints) contains no feasibility analysis or proof that a feasible solution continues to exist once the required compensation exceeds available actuator authority; this directly affects the claim that stability is maintained for unbounded attacks.","section":"RCLF-QP formulation and stability section"},{"comment":"No explicit bound, scaling study, or monitoring of the adaptation term is provided to verify that the CLF derivative remains negative definite when attack magnitudes grow without bound and the compensator saturates the controls; the reported simulations therefore do not test the unbounded regime asserted in the abstract.","section":"Simulation results and attack model"}],"minor_comments":[{"comment":"The abstract refers to a 'high-fidelity environment' without naming the simulator, the precise attack injection model on the pseudo-control channels, or the magnitude scaling used in the reported trials.","section":"Abstract"},{"comment":"Notation for the resilient compensational term and its adaptation law should be introduced with an explicit equation number to allow direct reference in the stability argument.","section":"Controller design"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":{"model":"grok-4.3","summary":"We thank the referee for the constructive comments. We address each major comment point by point below, indicating planned revisions where appropriate.","responses":[{"response":"We agree that the manuscript lacks an explicit feasibility analysis for the QP under attacks whose compensation exceeds actuator limits. The RCLF-QP minimizes deviation from the nominal input while enforcing the CLF decrease condition and input bounds, but feasibility is not assured for arbitrarily large attacks. In the revision we will add a dedicated discussion of QP feasibility, state that the Lyapunov guarantees apply conditional on feasibility, and revise the abstract to refer to 'unknown attacks of large but compensable magnitude' rather than unbounded attacks.","revision_made":"yes","referee_comment":"[RCLF-QP formulation and stability section] The abstract and the RCLF-QP construction assert compensation for unknown and unbounded attacks via real-time adaptation, yet the QP formulation (with its input constraints) contains no feasibility analysis or proof that a feasible solution continues to exist once the required compensation exceeds available actuator authority; this directly affects the claim that stability is maintained for unbounded attacks."},{"response":"The reported high-fidelity simulations demonstrate failure of the baseline PD controller and success of RCLF-QP under chosen attack magnitudes, but do not include a scaling study or explicit monitoring of the adaptation term and CLF derivative near saturation. We will augment the simulation section with additional runs that vary attack magnitude, plot the adaptation term, and confirm negativity of the CLF derivative for the largest feasible attacks tested.","revision_made":"yes","referee_comment":"[Simulation results and attack model] No explicit bound, scaling study, or monitoring of the adaptation term is provided to verify that the CLF derivative remains negative definite when attack magnitudes grow without bound and the compensator saturates the controls; the reported simulations therefore do not test the unbounded regime asserted in the abstract."}],"tokens_in":1450,"tokens_out":443,"duration_ms":43001,"standing_objections":["A rigorous proof that the QP remains feasible (and thus stability is guaranteed) for truly unbounded attacks that exceed available actuator authority."]},"desk_editor":{"model":"grok-4.3","letter":"The paper extends a prior fault-tolerant CLF-QP controller for quadrotors that lose two opposing rotors. It inserts a resilient compensational term with real-time online adaptation into the CLF to cancel unknown and unbounded cyberattacks on the pseudo-control channels.\n\nThe QP structure is the part that works cleanly. It keeps the stability argument from the CLF while allowing extra constraints and objectives to be added without rewriting the core proof. Pairing the mechanism with the extended state observer for lumped disturbances and wind is a natural fit for the combined physical-plus-cyber scenario. The abstract states that high-fidelity simulations keep the vehicle on trajectory where plain PD feedback fails, which matches the stated motivation.\n\nThe soft spot is the unbounded attack case. The adaptation term must generate whatever correction is needed, yet the QP is subject to actuator limits. If the required input exceeds those limits, the QP can become infeasible and the negative-definiteness of the CLF derivative is no longer assured. The manuscript reports successful simulations but gives no scaling of attack magnitude, no feasibility monitoring, and no separate argument that the QP remains solvable when attacks grow without bound. That gap directly touches the central claim.\n\nThis is for control engineers who work on resilient UAV methods. A reader already using CLF-QP frameworks will see a usable incremental example. It deserves peer review because the construction is a legitimate extension with a practical target, even if the feasibility analysis for unbounded attacks needs to be added or clarified.","headline":"The paper adds an adaptive resilient term to an existing CLF-QP for quadrotors to handle cyberattacks, but leaves the unbounded-attack feasibility question open.","tokens_in":2442,"tokens_out":376,"would_cite":false,"duration_ms":29996,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.3","headline":"A resilient compensational term inserted into the Control Lyapunov Function counters unknown cyberattacks on quadrotor control channels.","keywords":["quadrotors","cyberattacks","resilient control","control Lyapunov function","quadratic program","extended state observer","fault-tolerant control"],"falsifier":"A high-fidelity simulation run in which the quadrotor trajectory diverges or the closed-loop system becomes unstable under an injected attack on the pseudo-control channels despite the RCLF-QP controller would falsify the central claim.","tokens_in":2700,"feed_emoji":"🛡️","tokens_out":637,"duration_ms":36977,"temperature":0.7,"pith_summary":"The paper develops a control method to keep quadrotors stable when facing unknown and unbounded cyberattacks on their pseudo-control signals, in addition to wind and aerodynamic disturbances. It modifies the standard Control Lyapunov Function by adding an adaptive compensation term that adjusts in real time to offset the attacks. This modified function is placed inside a quadratic program that optimizes the control inputs while preserving stability. The overall architecture also uses an extended state observer to estimate and reject lumped disturbances. High-fidelity simulations show the quadrotor stays on its intended trajectory in attack scenarios where a conventional proportional-derivative controller loses stability.","feed_headline":"Adaptive term in CLF stabilizes quadrotors under cyberattacks","feed_subtitle":"Real-time compensation in quadratic program prevents divergence where baseline PD control fails","key_machinery":"The Resilient Control Lyapunov Function-based Quadratic Program (RCLF-QP), which augments the standard CLF with an adaptive compensation term to restore stability guarantees under attacks.","core_discovery":"By designing a resilient compensational term with real-time online adaptation inside the conventional Control Lyapunov Function and embedding the result in a quadratic program, the RCLF-QP controller compensates for maliciously injected unknown and unbounded cyberattacks on the pseudo-control channels while an extended state observer handles lumped disturbances, thereby preventing trajectory divergence where baseline PD feedback fails.","pith_inferences":["The same compensation idea could be tested on other underactuated aerial or ground vehicles facing sensor or actuator attacks.","Hardware experiments on a physical quadrotor would reveal whether communication delays affect the real-time adaptation performance.","The QP structure naturally supports adding explicit safety constraints such as minimum altitude or obstacle avoidance alongside the attack compensation."],"forward_implications":["The quadratic-program framework permits new control objectives and constraints to be added without changing the underlying stability guarantees.","The integrated observer and RCLF-QP together mitigate both lumped external disturbances and adversarial cyberattacks.","The architecture extends prior fault-tolerant results for complete loss of two opposing rotors to also cover malicious attacks.","Stability is preserved even when attacks are unbounded, provided the adaptation term can be computed online."],"fun_headline_variants":["RCLF-QP compensates cyberattacks via adaptive CLF","QP with resilient CLF prevents quadrotor attack divergence","Online adaptation in CLF mitigates quadrotor cyberattacks","Resilient CLF-QP handles attacks where PD control fails"],"cache_read_input_tokens":2112,"weakest_assumption_plain":"A resilient compensational term with real-time online adaptation can be inserted into the conventional CLF to compensate for unknown and unbounded cyberattacks.","fun_headline_variants_meta":{"raw":{"variants":["RCLF-QP compensates cyberattacks via adaptive CLF","QP with resilient CLF prevents quadrotor attack divergence","Online adaptation in CLF mitigates quadrotor cyberattacks","Resilient CLF-QP handles attacks where PD control fails"]},"model":"grok-4.3","cost_usd":0.007579,"raw_usage":{"total_tokens":3496,"prompt_tokens":714,"num_sources_used":0,"completion_tokens":67,"cost_in_usd_ticks":75787000,"prompt_tokens_details":{"text_tokens":714,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":2715,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":714,"tokens_out":67,"duration_ms":41169,"temperature":1.0,"reasoning_tokens":2715,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-06-30T00:57:48.773290+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"A high-fidelity simulation run in which the quadrotor trajectory diverges or the closed-loop system becomes unstable under an injected attack on the pseudo-control channels despite the RCLF-QP controller would falsify the central claim.","supporting_citations":[],"review_version":1}