{"id":"bb5bc29e-102f-42cf-892e-f52f0fb95a54","arxiv_id":"2502.00347","paper_version":1,"verdict":"REJECT","confidence":"HIGH","novelty_score":2.0,"correctness_risk":"high","formal_verification":"none","parameter_count":4,"one_line_summary":"An Arduino-based prototype combining MQ3 alcohol detection, IR eye-closure sensing, Bluetooth alerts, and automatic motor stop, with no quantitative validation of detection performance.","lead":"This paper describes a car safety prototype that combines an alcohol sensor, an eye-tracking infrared sensor, Bluetooth alerts, and automatic motor stop. The paper provides no measured evidence that the system detects drowsiness or intoxication reliably, so the central claim is unverified.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"IR sensor described as a heat-radiation motion detector cannot produce the sustained eye-closure signal the algorithm requires; hand-movement validation tests motion, not eyelid state.","rationale":"I read the paper as a prototype integration claim: off-the-shelf sensors plus Arduino plus Bluetooth plus relay. For the central claim to hold, the IR sensor must deliver a reliable sustained signal when the eyelid is closed. Section III.A.1 describes a heat-radiation motion detector with wavelength range 780 nm–50 μm and validation by hand movements; this is the behavior of a passive infrared (PIR) motion sensor. PIR sensors are intrinsically incapable of reporting a static occlusion as a persistent HIGH because they detect changes in the infrared scene, not static state. The algorithm's multi-stage persistence checks (2-second waits) cannot work with a signal that decays immediately after the eye closes. The hand-movement validation is consistent with motion detection and does not transfer to eye state. This is a stronger objection than 'the validation is weak' because it identifies a physical incompatibility between the described sensor and the required measurement. The reader's weakest_assumption captured the validation issue but not the sensor-principle mismatch, so I mark partial agreement. The rest of the system—MQ3 alcohol sensor, Bluetooth, relay—is standard, but the alcohol-detection path has similar calibration gaps (threshold 400, no ground-truth or false-alarm assessment), and the relay alone cannot implement 'progressive speed reduction' as claimed. Nevertheless, the eye-closure sensing is the load-bearing component for the drowsiness claim. I agree with REJECT: no code, data, wiring, or detection-accuracy results are shipped, and the limitations section concedes sensor accuracy and reliability issues.","tokens_in":9901,"tokens_out":5210,"duration_ms":55497,"concrete_test":"Ask the authors for the sensor part number and the glasses-mounted wiring diagram. Test the actual sensor with a seated subject: wear the glasses, look straight ahead, close eyes for 10 seconds, and log the digital output continuously. If the sensor is a PIR type (e.g., HC-SR501), the output will be a brief pulse at closure and at opening, not a sustained HIGH; the 2-second recheck logic will never fire. If the sensor is an active IR reflectance pair, repeat the test and compare against hand passes, head turns, and sunglasses to produce a confusion matrix for eye-closed vs. other states.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim depends on the IR sensor reporting prolonged eye closure as a sustained digital signal (EC=1) across repeated 2-second checks. Section III.A.1 specifies an infrared sensor detecting 780 nm–50 μm and 'changes in heat radiation' that is 'commonly used for motion detection'—characteristics of a passive infrared (PIR) sensor. A PIR sensor outputs a pulse only when the infrared scene changes; a closed eyelid that remains closed produces no ongoing change, so the signal would go LOW immediately after the eyelid settles. Algorithm 1's persistence checks at Wait 2s and Wait 2 more seconds cannot be satisfied. The validation described in III.A.1—'hand movements were used to simulate eye closure'—is exactly the kind of moving-source stimulus a PIR responds to, so it does not validate static eye-state detection. If the authors in fact used an active IR reflectance pair (IR LED + phototransistor), then the paper's physics description is wrong and the hand-movement test still fails to establish discrimination between eyelid closure and other occlusions. Either way, the core drowsiness detection claim is unsupported.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript proposes an Arduino Uno-based driver safety system that combines an MQ3 alcohol sensor and an infrared (IR) sensor mounted in glasses to detect intoxication and drowsiness, respectively. When alcohol or prolonged eye closure is detected, the system activates alarms, vibration, warning lights, Bluetooth alerts to a mobile phone, and, after persistence checks, gradually slows and stops the vehicle. The paper describes the hardware components, gives a pseudocode algorithm, presents proof-of-concept photos, and reports raw sensor readings over several days. The central claim is that this integrated low-cost system reliably detects driver impairment and drowsiness in real time.","tokens_in":10048,"tokens_out":5356,"duration_ms":53445,"significance":"If the central claims were supported, the system could be a useful low-cost addition to driver-safety research, and the paper addresses an important application with a clear component inventory, a formalized Algorithm 1, and an explicit limitations section. However, the evidence presented is qualitative or self-assigned: there are no accuracy, false-alarm, latency, or reliability measurements, and the IR sensor as described appears physically inconsistent with the required sustained eye-closure signal. The contribution as written is therefore not established, and the paper does not currently meet the evidentiary standard of a research article.","major_comments":[{"comment":"The IR sensor used for drowsiness detection is described as detecting 'changes in heat radiation' and as 'commonly used for motion detection,' i.e., a passive infrared (PIR) sensor, yet Algorithm 1 requires the eye-closure bit EC to remain 1 across an initial check and two subsequent 2-second waits (lines 13-21). A PIR sensor produces a transient pulse when the infrared scene changes, not a sustained output for a stationary closed eyelid, so the persistence checks cannot be satisfied by the stated sensor. The validation described in §III.A.1 ('hand movements were used to simulate eye closure') exercises motion detection, not static eyelid-state recognition, and does not establish discrimination between eyelid closure and other occlusions or head movements. If an active IR reflectance sensor was actually used, the paper must say so and provide a validation against eyelid state; as written, the core drowsiness-detection mechanism is unsupported.","section":"Alg. 1; §III.A.1"},{"comment":"The performance analysis reports raw MQ3 readings above a threshold of 400 and IR sensor readings over three or four days, but it does not compare these readings against ground-truth states of intoxication or drowsiness. No confusion matrix, true/false positive rates, false-alarm counts, or accuracy values are given. Consequently, the statements that 'our system surpassed it' and 'more efficiently identified eye movements' have no quantitative basis. This missing evaluation is the central evidence needed for the paper's reliability claim.","section":"§IV.A, Fig. 4"},{"comment":"Equations (1)-(8) are placeholder definitions (for example, vslow = f(Apersist) and data rate = data/time) rather than derived relationships with measured parameter values. In Algorithm 1, lines 9-10 execute '∆v ← 1; ∆v ← 0' consecutively, which collapses the claimed gradual slowdown and 10-15 s stop into an immediate stop, and no PWM/relay control law or timing calibration is provided. The automatic vehicle-stopping claim is therefore not backed by a testable control implementation.","section":"§III.C, Eqs. (1)-(8), Alg. 1"},{"comment":"The IoT versus non-IoT comparison is self-scored: the 0-100 ratings for Integration, Real-Time Data Acquisition, Safety Alert, Automation, and Instant Response are not tied to any measurement procedure, and the non-IoT data are said to come from literature, interviews, and prior studies without a scoring rubric. This figure cannot support the conclusion that the proposed system improves these metrics. The paper also does not compare the proposed system quantitatively against the other IoT systems listed in Table I.","section":"§IV.B, Fig. 5"},{"comment":"The 'real-time' claims are not backed by measured latency or throughput. Although Eq. (7) defines delay, no end-to-end Bluetooth delay, alarm activation time, or sensor-to-actuator response time is measured; §IV.C describes manual measurements and driver interviews rather than logged timings. The limitations section acknowledges sensor accuracy and environmental sensitivity, but it does not quantify how often false alarms occur or how the system behaves under variable lighting, vibration, or glasses movement. These are exactly the metrics needed to support the paper's reliability and real-time claims.","section":"§IV.C, §V.B, Eq. (7)"}],"minor_comments":[{"comment":"The conclusion and abstract contain garbled phrases such as 'the impact on-revision to reduce-threat recommendation' and 'driver alcoholism'; these should be corrected for clarity and professional tone.","section":"§VI, Abstract"},{"comment":"The axes, units, and the physical meaning of the sensor-value scale are not labeled, and no error bars, number of trials, or environmental conditions are reported, making the figures difficult to interpret.","section":"Fig. 4, Fig. 5"},{"comment":"Variable names are inconsistent: A is used for both the alcohol sensor reading and for alarm activation, while B is used for both Bluetooth notification and the eye-closure verification state; this makes the pseudocode harder to follow than necessary.","section":"Alg. 1"},{"comment":"Several background and methodology sentences are supported by the authors' own prior papers (e.g., [8], [9], [21], [23], [28], [31]-[34]) where a directly relevant external reference would be more appropriate; these citations should be replaced or supplemented.","section":"References"},{"comment":"The limitations bullet list mentions environmental sensitivity and user acceptance, but the Discussion section does not connect these limitations to the experimental results or propose a concrete test that would quantify their impact on detection performance.","section":"§V.B"}],"recommendation":"reject","confidential_remarks":"This manuscript reads as an undergraduate project report rather than a complete research contribution. The absence of quantitative evaluation and the physical inconsistency of the IR sensor with the required sustained eye-closure signal make the central claim unverifiable as written. I would not encourage resubmission in the current form; if the authors collect proper ground-truth data with a suitable eye-closure sensor, a substantially rewritten empirical paper could be reconsidered. There is also a pattern of self-citations in the background sections that should be curtailed in any future version."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my take on arXiv:2502.00347. The paper is a straightforward integration report: Arduino Uno plus MQ3 alcohol sensor, an IR sensor mounted in glasses, buzzer, vibration motor, relay, and Bluetooth. The write-up is clear and the algorithm is simple to follow. What's genuinely useful is the completeness of the hardware list and the explicit sequence of alerts and vehicle-stop logic. For an undergraduate design project, it's a reasonable proof of concept.\n\nThe problems are in the evidence, and one is fatal. The IR sensor is described as detecting 780 nm–50 µm heat-radiation changes and is 'commonly used for motion detection'—that's a passive infrared (PIR) sensor. PIRs output a pulse when the scene changes, then go low. The algorithm, however, requires EC=1 to persist across a 2-second wait and then a second 2-second wait before slowing the vehicle. A PIR cannot produce that sustained signal; a closed eyelid that stays closed creates no further scene change. The validation used hand movements to simulate eye closure, which is exactly the moving-source stimulus a PIR responds to, so it tests motion, not eyelid state. If the authors actually used an active IR reflectance pair, the physics description is wrong and the hand-movement test still doesn't discriminate eyelid closure from other occlusions. Either way, the core drowsiness detection claim is unsupported.\n\nThe rest of the soft spots are proportionate: no ground-truth accuracy, no false-alarm counts, no latency numbers, the IoT comparison in Fig. 5 is self-scored, and Eqs. (1)–(8) are definitions rather than derivations. No code, wiring diagrams, or raw data are shipped. The limitations section does concede sensor accuracy and reliability, which is honest but doesn't fix the lack of quantitative support. Citation-wise, a large fraction of background references point to Anichur Rahman's own SDN/blockchain work that is only tangentially related to driver safety; this reads as padding.\n\nWhat's genuinely novel? The specific combination of alcohol sensing, eye monitoring, Bluetooth alerts, vibration, and relay-based auto-stop in one Arduino system. Their Table I shows prior systems with overlapping features, so the novelty is combination, not component.\n\nWho is this for? A reader interested in low-cost driver-safety prototypes for low- and middle-income countries might find the design sketch useful as a starting point. As a research contribution it does not hold up: the central measurement is not validated, and the sensor physics contradicts the algorithm. I would not send this to a serious referee in its current form. If the authors ran a controlled driving study with ground truth, false-alarm rates, and an IR sensor that actually measures eyelid state, it could become a modest empirical paper.","headline":"A clearly written prototype report whose central drowsiness-detection claim is undercut by the IR sensor's physics and by validation that tests motion, not eyelid state.","tokens_in":10675,"tokens_out":3157,"would_cite":false,"duration_ms":29277,"reading_group":"no","serious_thinker":"no","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"An Arduino system that detects drunk and drowsy drivers and automatically slows and stops the vehicle.","keywords":["IoT","drowsiness detection","alcohol detection","Arduino Uno","IR sensor","MQ3 sensor","driver safety","real-time monitoring"],"falsifier":"A controlled road test in which a driver wears the IR-sensor glasses while a video camera records the face as ground truth: if the sensor triggers alerts during normal driving with eyes open (during head turns, sunlight, or bumps) or misses prolonged eye closure, the central claim that the system reliably detects drowsiness fails.","tokens_in":9650,"feed_emoji":"🚗","tokens_out":2357,"duration_ms":26097,"temperature":0.7,"pith_summary":"This paper proposes a low-cost, Arduino-based safety system that combines an MQ3 alcohol sensor with an infrared (IR) eye-closure sensor to monitor a driver for intoxication and drowsiness in real time. When alcohol is detected, the system sounds an alarm and flashes a red light, then slows and stops the motor within 10 to 15 seconds if the alcohol persists. When the IR sensor finds the eyes closed for more than two seconds, it triggers vibration, an alarm, and a red light, and it stops the vehicle if the eyes remain closed. The system also sends alerts to the driver's mobile phone via Bluetooth. If the approach works in real vehicles, it offers an inexpensive way to reduce accidents caused by impaired or sleepy drivers, particularly in low- and middle-income countries where such systems are scarce.","feed_headline":"Cheap sensor rig can stop a car when the driver nods off","feed_subtitle":"An Arduino, an alcohol sensor, and an eye-closure detector combine to warn, slow, and halt the vehicle.","key_machinery":"The load-bearing mechanism is the sensor-and-relay control loop managed by the Arduino Uno: the MQ3 sensor acts as a chemiresistor whose resistance drops in the presence of alcohol, and the IR sensor (mounted in driver's glasses) detects changes in heat radiation that the paper interprets as eyelid closure. A relay module connected to the motor enables the microcontroller to physically reduce speed and cut power. The verification timers—20 seconds for alcohol rechecking, two seconds for eye-closure confirmation—are what let the system distinguish a transient signal from a sustained risk before it triggers vehicle stoppage. Bluetooth (HC-06) carries alerts and commands to a mobile app, completing the 'IoT-enabled' monitoring loop.","core_discovery":"The central claim is that a driver-monitoring unit built from off-the-shelf components—an Arduino Uno, an MQ3 alcohol sensor, an IR sensor mounted in glasses, a relay module, a buzzer, LEDs, and an HC-06 Bluetooth module—can reliably detect both alcohol impairment and prolonged eye closure, warn the driver, and take direct control of the vehicle by gradually slowing and then stopping the motor. The paper presents an algorithm that first gives a short verification window (20 seconds for alcohol, two seconds for eye closure) before escalating to automated vehicle stoppage, so that brief events do not trigger harsh interventions. It also reports a comparison of IoT and non-IoT systems on metrics like integration, real-time data acquisition, safety alerts, automation, and instant response, claiming the IoT approach scores higher on all. The contribution is presented as the first integration of dual-detection sensing, wireless mobile alerts, and automatic speed reduction and stoppage in a single low-cost prototype.","pith_inferences":["The paper's validation of the IR sensor with hand movements rather than real eyelid closure means the deployment risk is not the sensor hardware but the transfer of the simulated test to a moving, vibrating, variably lit vehicle; a field test that compares sensor output with video-graded eyelid closure would settle this quickly.","The claimed comparison scores for IoT versus non-IoT systems appear to be drawn from a small, informal sample plus literature, so the quantitative performance metrics should be read as illustrative rather than statistically established.","A natural extension the paper does not develop is replacing the IR sensor with a camera-based blink classifier (for example, eye-aspect-ratio tracking) to reduce false alarms from head turns and glasses slippage, while keeping the same Arduino control and relay logic.","If the dual-detection design is adopted in real fleets, an external safety question arises that the paper does not address: how the system's automatic stoppage interacts with surrounding traffic, since a sudden halt on a highway could itself cause a collision."],"forward_implications":["If the alcohol sensor reads above its threshold, the system activates the alarm and red light for 20 seconds, then rechecks; a second positive reading triggers progressive speed reduction and motor stop within 10 to 15 seconds.","If the IR sensor detects closed eyes, the system verifies for two seconds, then activates vibration, alarm, and red light; if the eyes remain closed for another two seconds, it slows and stops the vehicle.","All alerts and vehicle-state messages are transmitted to the driver's phone over Bluetooth, giving the driver a real-time view of their eye status and alcohol readings.","The system's automatic speed reduction and stoppage would apply to the host vehicle's motor via the relay module, meaning a single hardware design could be ported to cars, buses, trucks, and fleet vehicles.","Because the system uses standard Arduino components, the per-unit cost is low enough that the design could be adopted in low- and middle-income countries where road fatalities are highest."],"supporting_citations":[{"why":"Supplies the background that IR sensors sense changes in heat radiation and are cost-effective, which the paper uses to justify the eye-closure detection approach.","marker":"[21]"},{"why":"Provides the MQ3 sensor's operating specification (5V, 800mW, 25-500 ppm detection range) that grounds the alcohol-detection threshold claim.","marker":"[22]"},{"why":"Defines the Arduino Uno's ATmega328P capabilities and I/O count, which the system's control logic depends on.","marker":"[23]"},{"why":"Gives the HC-06 Bluetooth module's protocol and data-rate characteristics that support the real-time alert transmission claim.","marker":"[24]"},{"why":"Explains relay module voltage ratings and switching function, which is the mechanism the paper relies on to slow and stop the motor.","marker":"[25]"},{"why":"The cited prior Arduino-based anti-sleep device that the paper explicitly extends by adding alcohol detection, Bluetooth alerts, and automatic vehicle stoppage.","marker":"[7]"}],"fun_headline_variants":["Drowsy driver? This Arduino rig slows and stops your car","IoT rig detects alcohol and eye closure, halts vehicle","Sensor glasses and alcohol sniffer auto-stop car when drowsy","Car halts itself when driver dozes or drinks, thanks to smart sensors","Arduino-based driver monitor slows and stops for drowsiness and alcohol"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The IR sensor was validated only by using hand movements to simulate eye closure, so the whole system depends on the unproven assumption that real eyelid closure in a moving, vibrating, and variably lit vehicle produces the same sensor response without false alarms from head motion, glasses slippage, or ambient infrared sources.","fun_headline_variants_meta":{"raw":{"variants":["Drowsy driver? This Arduino rig slows and stops your car","IoT rig detects alcohol and eye closure, halts vehicle","Sensor glasses and alcohol sniffer auto-stop car when drowsy","Car halts itself when driver dozes or drinks, thanks to smart sensors","Arduino-based driver monitor slows and stops for drowsiness and alcohol"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000675,"raw_usage":{"total_tokens":3054,"prompt_tokens":913,"completion_tokens":2141,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":529,"completion_tokens_details":{"reasoning_tokens":2051}},"tokens_in":529,"tokens_out":2141,"duration_ms":16301,"temperature":1.0,"reasoning_tokens":2051,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T19:20:04.340229+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A controlled road test in which a driver wears the IR-sensor glasses while a video camera records the face as ground truth: if the sensor triggers alerts during normal driving with eyes open (during head turns, sunlight, or bumps) or misses prolonged eye closure, the central claim that the system reliably detects drowsiness fails.","supporting_citations":[{"cited_title":"Towards a blockchain-sdn-based secure architecture for cloud computing in smart industrial iot,","cited_arxiv_id":null,"evidence_quote":"Supplies the background that IR sensors sense changes in heat radiation and are cost-effective, which the paper uses to justify the eye-closure detection approach."},{"cited_title":"Mq-3 alcohol sensor arduino interface tutorial,","cited_arxiv_id":null,"evidence_quote":"Provides the MQ3 sensor's operating specification (5V, 800mW, 25-500 ppm detection range) that grounds the alcohol-detection threshold claim."},{"cited_title":"On the icn-iot with federated learning integration of communication: Concepts, security-privacy issues, applications, and future perspectives,","cited_arxiv_id":null,"evidence_quote":"Defines the Arduino Uno's ATmega328P capabilities and I/O count, which the system's control logic depends on."},{"cited_title":"Hc-06 bluetooth module pinout & datasheet,","cited_arxiv_id":null,"evidence_quote":"Gives the HC-06 Bluetooth module's protocol and data-rate characteristics that support the real-time alert transmission claim."},{"cited_title":"What is a relay module and what does it do?","cited_arxiv_id":null,"evidence_quote":"Explains relay module voltage ratings and switching function, which is the mechanism the paper relies on to slow and stop the motor."},{"cited_title":"Developing an arduino based anti-sleep device for driver,","cited_arxiv_id":null,"evidence_quote":"The cited prior Arduino-based anti-sleep device that the paper explicitly extends by adding alcohol detection, Bluetooth alerts, and automatic vehicle stoppage."}],"review_version":1}