{"id":"c3a848c1-e785-4968-943a-0dfdb8ae5464","arxiv_id":"2606.11109","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":5.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":0,"one_line_summary":"EM-Fall integrates mmWave sensing with robotic mobility on humanoid robots to enable continuous fall detection across rooms, occlusions, and lighting conditions in real homes.","lead":"EM-Fall places mmWave radar on a humanoid robot that moves around a home to keep a person in view for fall detection, working without light and despite obstacles or pets. A smart generalist might read it to see how robot mobility can solve coverage and lighting problems that limit fixed sensors or wearables in elderly care.","discovery_kind":"new_application","skeptic_critique":{"model":"grok-4.3","headline":"Human-centered perception pipeline's handling of pet motion and multipath lacks demonstrated robustness to support reliable cross-room detection.","rationale":"The reader's weakest_assumption matches the load-bearing point exactly. Because the review was abstract-only and no quantitative results or method details are visible here, the information limit that produced UNVERDICTED remains unchanged.","tokens_in":1712,"tokens_out":234,"duration_ms":25070,"concrete_test":"On the in-home dataset, run an ablation removing the human-centered filtering and temporal modeling stages; if F1-score on pet-present or occluded sequences drops more than 15% relative to the full system, the pipeline's sufficiency is unsupported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim requires that mobility plus the pipeline maintains robust detection despite pet motion and multipath. The abstract states the pipeline exists but supplies no architecture details, loss functions, or ablation results isolating its contribution. With evaluation limited to four participants, it is unclear whether the lightweight temporal model generalizes beyond the collected falls or distinguishes pet trajectories from human pre-fall motion in unseen rooms.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The paper proposes EM-Fall, an embodied mmWave sensing framework deployed on a mobile humanoid robot for fall detection. The system uses robotic mobility to actively adjust the sensing viewpoint for improved observability across rooms and under occlusion. It incorporates a human-centered perception pipeline combined with lightweight temporal modeling to mitigate interference from pet motion and multipath artifacts. The framework is evaluated across eight real indoor environments with four participants, resulting in the construction of an in-home mmWave fall detection dataset. The authors claim that the embodied mobile sensing paradigm improves monitoring continuity and maintains robust fall detection performance under diverse environmental conditions.","tokens_in":1782,"tokens_out":370,"duration_ms":21680,"significance":"If the robustness claims hold, the work could provide a practical robot-assisted solution for elderly safety monitoring that addresses compliance, coverage, and lighting limitations of existing wearable and fixed-sensor approaches. The combination of mobility with mmWave sensing for day-and-night operation represents a potentially useful systems integration for residential environments.","major_comments":[{"comment":"Abstract / Evaluation description: The central claim that 'experimental results show that the embodied mobile sensing paradigm improves monitoring continuity and maintains robust fall detection performance under diverse environmental conditions' is unsupported by any quantitative metrics, baselines, error rates, dataset statistics (e.g., number of falls or pet-motion events), or ablation results isolating the contribution of the perception pipeline. With evaluation limited to four participants, it is impossible to assess whether the system generalizes or reliably distinguishes human falls from pet trajectories in unseen rooms.","section":"Abstract / Evaluation"}],"minor_comments":[],"recommendation":"uncertain","confidential_remarks":"The manuscript as presented supplies no implementation details, loss functions, or performance numbers despite asserting evaluation results; this is a basic requirement for assessing a systems paper claiming robustness."},"author_rebuttal":{"model":"grok-4.3","summary":"We thank the referee for the constructive comments. We agree that the abstract claim requires explicit quantitative support and will revise the manuscript to address this.","responses":[{"response":"We accept the point that the abstract's claim is not accompanied by supporting numbers. The manuscript describes evaluation across eight environments and four participants but does not report specific metrics, baselines, or ablations in the provided abstract text. In revision we will (1) add concrete statistics to the abstract (e.g., number of fall events, pet-motion instances, detection rates, and comparison to fixed-sensor baselines), (2) include ablation results isolating the perception pipeline and temporal modeling, and (3) rephrase the generalization language to reflect the limited participant count while noting the environmental diversity. These changes will make the claims directly traceable to the evaluation data.","revision_made":"yes","referee_comment":"[Abstract / Evaluation] Abstract / Evaluation description: The central claim that 'experimental results show that the embodied mobile sensing paradigm improves monitoring continuity and maintains robust fall detection performance under diverse environmental conditions' is unsupported by any quantitative metrics, baselines, error rates, dataset statistics (e.g., number of falls or pet-motion events), or ablation results isolating the contribution of the perception pipeline. With evaluation limited to four participants, it is impossible to assess whether the system generalizes or reliably distinguishes human falls from pet trajectories in unseen rooms."}],"tokens_in":1331,"tokens_out":305,"duration_ms":19713,"standing_objections":[]},"desk_editor":{"model":"grok-4.3","letter":"The main thing here is a systems integration: putting mmWave radar on a humanoid robot so it can move around and keep line-of-sight on a person even when they fall in another room or behind furniture. That combination of active viewpoint adjustment plus a human-centered pipeline for handling pets and multipath is the concrete new piece relative to fixed mmWave or wearable setups.\n\nWhat the work does cleanly is frame a practical home-monitoring problem—day/night, occlusion, low compliance with wearables—and show how mobility can address the spatial-coverage part. The abstract also mentions building a real in-home dataset across eight environments with four participants, which is the right direction for this kind of applied robotics paper.\n\nThe soft spot is that none of the performance claims can be checked from what is written. There are no detection rates, false-positive numbers, comparisons to static sensors, or ablation results on the temporal model. The stress-test note about pet motion and multipath is fair: the abstract asserts the pipeline handles them but gives no architecture details, loss functions, or evidence that it generalizes beyond the four people. With only four participants the generalization question is open.\n\nThis is the kind of paper that belongs in an applied robotics or human-robot interaction venue rather than a core sensing or learning conference. A reader working on robot-assisted eldercare or mmWave perception would get value from the system description and the dataset if the full paper supplies the missing metrics and code. It is worth sending to referees because the idea is grounded in a real deployment constraint and the authors appear to have run the experiments; the current text just does not let anyone judge whether the robustness claim holds.","headline":"The paper integrates mmWave sensing with a mobile humanoid robot for fall detection but the abstract supplies no numbers or baselines, leaving the performance claims uncheckable.","tokens_in":2312,"tokens_out":412,"would_cite":false,"duration_ms":10457,"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 humanoid robot with mmWave radar detects falls by moving to keep a clear view across rooms and at night.","keywords":["fall detection","mmWave sensing","humanoid robot","embodied sensing","elderly monitoring","robot mobility","indoor safety"],"falsifier":"A test in an additional home showing detection accuracy dropping below the reported level when a pet is active or when furniture prevents timely robot repositioning would challenge the robustness claim.","tokens_in":2620,"feed_emoji":"🤖","tokens_out":615,"duration_ms":35413,"temperature":0.7,"pith_summary":"This paper sets out to show that mounting millimeter-wave sensors on a mobile humanoid robot lets fall detection operate continuously in homes. The robot repositions itself to maintain a line of sight on the person even when furniture blocks the path or lights are off. A processing step that tracks how human motion changes before, during, and after an event helps separate real falls from pet movements or wall reflections. Readers might care because falls remain a leading cause of injury for older adults and existing wearables or fixed cameras often lose compliance or coverage in everyday spaces. If the approach holds, robot helpers could supply always-available safety checks without requiring people to wear devices or add permanent installations.","feed_headline":"Moving robot keeps mmWave view for day-night fall detection","feed_subtitle":"Repositioning maintains a clear radar line on the person despite obstacles, pets, and darkness.","key_machinery":"The embodied mobile sensing paradigm in which the robot actively adjusts its millimeter-wave sensing viewpoint to maintain observability, paired with a human-centered perception pipeline and lightweight temporal modeling of motion sequences.","core_discovery":"The central claim is that an embodied fall detection framework deployed on a mobile humanoid robot integrates millimeter-wave sensing with active viewpoint adjustment to maintain target observability across rooms and under occlusion. A human-centered perception pipeline combined with lightweight temporal modeling captures motion evolution to address interference from pet motion and multipath artifacts. Evaluation across eight real indoor environments with four participants shows that this mobile sensing paradigm improves monitoring continuity while maintaining robust fall detection performance under diverse conditions.","pith_inferences":["The same mobility and sensing combination could support monitoring of other motion-based events such as prolonged inactivity.","A single robot might eventually replace multiple fixed sensors by following optimized paths through the home."],"forward_implications":["The system maintains observability across rooms and under occlusion through robotic mobility.","It addresses interference from pet motion and multipath artifacts via the perception pipeline.","It delivers improved monitoring continuity compared with fixed sensing installations.","It maintains robust performance under diverse environmental conditions including poor lighting."],"fun_headline_variants":["Humanoid robot adjusts mmWave for continuous fall detection","Mobile robot mmWave maintains view despite occlusion and pets","Embodied mmWave on humanoid robot detects falls day and night","Robot mobility improves mmWave fall monitoring in homes","Humanoid robot repositions mmWave sensor for fall awareness"],"cache_read_input_tokens":64,"weakest_assumption_plain":"That robotic repositioning together with human-focused motion tracking over time can sufficiently filter pet movements and multipath effects to support reliable detection in varied home layouts.","fun_headline_variants_meta":{"raw":{"variants":["Humanoid robot adjusts mmWave for continuous fall detection","Mobile robot mmWave maintains view despite occlusion and pets","Embodied mmWave on humanoid robot detects falls day and night","Robot mobility improves mmWave fall monitoring in homes","Humanoid robot repositions mmWave sensor for fall awareness"]},"model":"grok-4.3","cost_usd":0.004444,"raw_usage":{"total_tokens":2220,"prompt_tokens":670,"num_sources_used":0,"completion_tokens":77,"cost_in_usd_ticks":44437000,"prompt_tokens_details":{"text_tokens":670,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":1473,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":670,"tokens_out":77,"duration_ms":10315,"temperature":1.0,"reasoning_tokens":1473,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-06-27T13:00:32.926621+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"A test in an additional home showing detection accuracy dropping below the reported level when a pet is active or when furniture prevents timely robot repositioning would challenge the robustness claim.","supporting_citations":[],"review_version":1}