{"id":"871d64dc-cd0f-4382-aac1-8520c7df495f","arxiv_id":"2606.22289","paper_version":1,"verdict":"ACCEPT","confidence":"LOW","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Short sequences of legitimate MAVLink messages can manipulate ArduPilot's PID, EKF, and failsafe parameters to degrade stability and induce crashes in both simulation and hardware.","lead":"This paper shows that attackers can crash ArduPilot drones by sending ordinary MAVLink commands that tweak PID gains, EKF filters, and failsafe settings. A smart generalist might read it to see how trusted drone control links can be turned into weapons without breaking any code.","discovery_kind":"unclear","skeptic_critique":{"model":"grok-4.3","headline":"No significant objection identified","rationale":"Reader's weakest assumption matches the explicit attack vector; hardware validation supplies the concrete evidence needed to support an ACCEPT verdict. No additional load-bearing gap is visible from the provided description.","tokens_in":1718,"tokens_out":251,"duration_ms":16598,"concrete_test":"Re-execute the six MAVLink sequences from the hardware-validation section on an identical Pixhawk 2.4.8 + ArduCopter 4.x firmware image while logging EKF innovation, PID state, and attitude error; confirm that at least one combined sequence produces the reported crash within the same number of messages.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim rests on the observation that well-formed MAVLink parameter-set commands are accepted and applied without runtime stability or estimator-health checks. The manuscript supplies SITL traces plus Pixhawk 4.8 hardware runs showing the six described sequences produce the claimed degradation in attitude, rate, and EKF health, culminating in loss of control when effects are combined. No internal inconsistency appears between the attack construction and the reported outcomes; the absence of validation is both the stated attack surface and the demonstrated mechanism.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The paper claims that short sequences of well-formed MAVLink parameter-set commands can exploit ArduPilot's multi-layer controllers (PID gains, EKF configuration, failsafe assumptions) to degrade attitude stability, angular-rate behavior, trajectory tracking, and estimator health, ultimately causing loss of control and crashes when effects are combined. The central results are direct experimental observations from SITL simulation and Pixhawk 2.4.8 hardware runs demonstrating six specific attacks.","tokens_in":1796,"tokens_out":359,"duration_ms":25459,"significance":"If the results hold, the work is significant for highlighting practical security gaps in input-parameter handling and command trust within a widely deployed open-source flight controller. The combination of SITL traces and hardware validation on real Pixhawk hardware, together with the absence of fitted parameters or circular derivations, provides concrete, falsifiable evidence of controller sensitivity to legitimate commands.","major_comments":[{"comment":"Evaluation section: the abstract states concrete degradation metrics, but full data tables, run counts, success rates, and exclusion criteria are not visible; this is load-bearing for assessing reproducibility of the hardware validation and the claim that combined effects produce crashes.","section":"Evaluation section"}],"minor_comments":[{"comment":"Abstract: 'inwhich' is a typographical error and should be 'in which'.","section":"Abstract"},{"comment":"The six attacks are described at a high level; adding a table summarizing the exact MAVLink message sequences, targeted parameters, and observed effects per attack would improve clarity without altering the central claim.","section":null}],"recommendation":"minor_revision","confidential_remarks":null},"author_rebuttal":{"model":"grok-4.3","summary":"Thank you for the opportunity to respond to the referee's report. We appreciate the referee's recognition of the significance of our work and the recommendation for minor revision. We address the single major comment below.","responses":[{"response":"We agree with the referee that additional details on the experimental data would improve the reproducibility of our results. The current manuscript presents key observations from SITL and hardware experiments but does not include comprehensive tables summarizing all runs. In the revised version, we will expand the Evaluation section to include tables reporting the number of experimental runs performed for each attack, success rates, observed degradation metrics with statistics where applicable, and any exclusion criteria applied during hardware validation. This will provide stronger support for the claims that combined effects can lead to crashes.","revision_made":"yes","referee_comment":"[Evaluation section] Evaluation section: the abstract states concrete degradation metrics, but full data tables, run counts, success rates, and exclusion criteria are not visible; this is load-bearing for assessing reproducibility of the hardware validation and the claim that combined effects produce crashes."}],"tokens_in":1253,"tokens_out":240,"duration_ms":19204,"standing_objections":[]},"desk_editor":{"model":"grok-4.3","letter":"The core contribution is a set of six concrete attack sequences that tweak PID gains, EKF parameters, and failsafe settings through ordinary MAVLink messages. These sequences are shown to affect attitude, rates, trajectory tracking, and estimator health, with combined effects leading to loss of control.\n\nThe work does a clean job of moving from simulation to real hardware on a Pixhawk 2.4.8. The SITL traces and hardware runs give direct evidence that the controller applies the changes without runtime stability checks, which is the stated attack surface. That experimental grounding is the strongest part.\n\nA minor soft spot is that the manuscript does not appear to test whether later ArduPilot versions added any parameter validation or rate limiting that might blunt these sequences. The results are tied to the version under test, so the practical window of exposure is not fully mapped. No other load-bearing issues stand out.\n\nThis is useful for anyone working on UAV security or open-source autopilot hardening. Readers who need examples of how parameter trust can be exploited will find the catalog and the hardware confirmation directly applicable.\n\nI would send it to peer review. The experimental evidence is there and the topic matters for deployed systems.","headline":"The paper catalogs six MAVLink parameter attacks on ArduPilot that degrade control and cause crashes in both simulation and Pixhawk hardware, with the experiments appearing to match the claims.","tokens_in":2275,"tokens_out":324,"would_cite":false,"duration_ms":16958,"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":"Short sequences of legitimate MAVLink commands can degrade ArduPilot UAV stability and cause crashes by changing controller parameters.","keywords":["ArduPilot","MAVLink","UAV security","flight controller attacks","PID gains","EKF","failsafe","control-aware attacks"],"falsifier":"Running the described sequences of MAVLink parameter commands in ArduPilot SITL simulation or on Pixhawk hardware and observing whether attitude stability, angular rates, trajectory tracking, and estimator health degrade as claimed, or if the vehicle crashes.","tokens_in":2640,"feed_emoji":"🚁","tokens_out":671,"duration_ms":26409,"temperature":0.7,"pith_summary":"This paper establishes that an adversary can use well-formed MAVLink messages to alter PID gains, EKF settings, and failsafe parameters in ArduPilot flight controllers. These changes exploit the controller's sensitivity to parameter values and update rates, leading to degraded attitude stability, trajectory tracking, and estimator performance. When effects are combined, the UAV enters unsafe states and crashes. The demonstrations in simulation and on Pixhawk hardware show that no malformed commands are needed, only trusted ones. A sympathetic reader would care because current UAV systems may lack checks against such parameter manipulations that affect closed-loop behavior.","feed_headline":"Legitimate MAVLink messages crash ArduPilot drones","feed_subtitle":"Short command sequences alter PID and EKF parameters to degrade stability and force crashes in simulation and hardware tests.","key_machinery":"The six attacks that modify PID gains, alter EKF estimation configuration, and violate failsafe assumptions to exploit controller sensitivities to parameter values and update frequency.","core_discovery":"The paper claims that six specific attacks, each using short sequences of MAVLink parameter-change commands, can force ArduPilot into unsafe operating conditions by modifying interactions among its multi-layer controllers. These attacks target PID gains, EKF estimation configuration, and failsafe assumptions, and their combined effects result in loss of control and vehicle crashes, as validated in both SITL simulation and Pixhawk hardware experiments.","pith_inferences":["Other flight controllers that accept MAVLink parameter commands without stability checks may share similar vulnerabilities.","Adding runtime validation of parameter effects on closed-loop dynamics could block these attacks but would change how controllers process trusted inputs.","The same parameter-manipulation approach could be tested on different UAV hardware platforms to determine if the issues extend beyond ArduPilot.","Security for UAVs may need to treat parameter updates as potential control inputs rather than benign configuration."],"forward_implications":["Modifying PID gains degrades attitude stability and angular-rate behavior.","Altering EKF estimation configuration affects estimator health.","Violating failsafe assumptions allows the vehicle to enter unsafe operating conditions.","Combining multiple parameter changes causes trajectory tracking failure and vehicle crashes.","These outcomes occur from short sequences of well-formed MAVLink messages without code changes."],"fun_headline_variants":["MAVLink alters ArduPilot PID and EKF to crash","Short MAVLink commands destabilize ArduPilot flight","ArduPilot crashes from MAVLink parameter sequences","Hardware tests show MAVLink causing ArduPilot crashes"],"cache_read_input_tokens":2112,"weakest_assumption_plain":"The ArduPilot flight controller accepts and immediately applies MAVLink parameter-change commands without additional runtime validation of their effect on closed-loop stability or estimator health.","fun_headline_variants_meta":{"raw":{"variants":["MAVLink alters ArduPilot PID and EKF to crash","Short MAVLink commands destabilize ArduPilot flight","ArduPilot crashes from MAVLink parameter sequences","Hardware tests show MAVLink causing ArduPilot crashes"]},"model":"grok-4.3","cost_usd":0.007334,"raw_usage":{"total_tokens":3286,"prompt_tokens":651,"num_sources_used":0,"completion_tokens":63,"cost_in_usd_ticks":73340500,"prompt_tokens_details":{"text_tokens":651,"audio_tokens":0,"image_tokens":0,"cached_tokens":64},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":2572,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":651,"tokens_out":63,"duration_ms":20073,"temperature":1.0,"reasoning_tokens":2572,"cache_read_input_tokens":64,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-06-26T10:53:17.876408+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"Running the described sequences of MAVLink parameter commands in ArduPilot SITL simulation or on Pixhawk hardware and observing whether attitude stability, angular rates, trajectory tracking, and estimator health degrade as claimed, or if the vehicle crashes.","supporting_citations":[],"review_version":1}