{"id":"1c945e1a-fe2c-41fa-9c94-4a7f8bc33b7b","arxiv_id":"2507.16258","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Animal encounters with autonomous vehicles, delivery robots, and drones cluster into five concern areas, physical impact, behavioral effects, accessibility, ethics and regulation, and urban disturbance, with design and policy directions for coexistence.","lead":"This paper maps how animals interact with self-driving cars, delivery robots, and drones by combining a 45-paper review, real YouTube and Reddit encounters, and 8 expert interviews. It organizes the problems into five areas, physical harm, behavior changes, service-animal access, ethics and rules, and city disruption, and then proposes design and policy directions.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Ethnographic data were coded with the same five themes they are said to confirm, so the claimed multi-method convergence is partly constructed.","rationale":"The reader's weakest assumption concerned the reliability of user-generated YouTube and Reddit evidence. That is a real risk, but the literature alone already supports the Physical Impact and Behavioural Effects themes, so even if some clips are inaccurate or cherry-picked, the central claim does not collapse. The more load-bearing issue is methodological: the five themes were developed from the scoping review and then applied to the ethnography, so the ethnography cannot serve as an independent corroboration of those themes. The reader did note this in their rationale point (3), but did not make it the primary weakest assumption. My read therefore agrees partially with the reader's framing. The concern does not overturn the paper; it is an honest, well-documented qualitative synthesis, and the expert interviews provide some additional grounding. But it does mean the abstract's 'surfaces five key areas' overstates the evidentiary strength of the multi-method design. The verdict should remain CONDITIONAL, with a request that the authors either (a) report an independent inductive coding of the ethnographic corpus or (b) soften the claim so that the five themes are attributed to the literature with ethnographic illustrations, not to independent cross-source emergence.","tokens_in":23521,"tokens_out":4474,"duration_ms":48051,"concrete_test":"Have two independent coders, blind to the paper's five themes, perform open coding on the 39 YouTube videos and 11 Reddit discussions listed in Tables 7 and 8, then group the resulting codes into their own thematic categories. Compare the emergent categories with the paper's five themes. If the independent grouping diverges substantially—for example, if items about visual misrepresentation or legal accountability do not fall under Physical Impact or Ethics and Regulations, or if a new dominant category such as 'human responsibility for companion animals' appears—then the cross-source convergence in Section 4.1 is partly an artifact of the codebook, and the abstract's claim that the analysis 'surfaces' these five areas should be qualified.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim, stated in the abstract as 'Our analysis surfaces five key areas of concern,' depends on convergence between the scoping review and the online ethnography. But the five themes were not independently derived in both corpora. In Section 3.1, papers were 'subsequently grouped under five thematic categories' after inductive analysis; in Section 3.2, the videos and Reddit posts were coded by applying 'the same five thematic categories from the scoping review.' The ethnographic material was therefore sorted into a pre-existing frame, so the apparent agreement between literature and real-world encounters is partly by construction. This does not by itself falsify the thematic map, but it removes the ethnography as an independent confirmation that these five categories are the natural clustering of the domain. If the ethnographic items were coded openly, different or additional categories might emerge, which would alter the design and policy emphasis. The paper is transparent about the procedure, which is good, but the abstract's phrasing invites a stronger reading than the method can support.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports a multi-method study of how animals interact with autonomous mobility systems, including AVs, drones, delivery robots, and lawn-mower robots. It combines a scoping review of 45 articles, an online ethnography of 39 YouTube videos and 11 Reddit discussions, and 8 expert interviews. The authors identify five thematic areas of concern—Physical Impact, Behavioural Effects, Accessibility Concerns, Ethics and Regulations, and Urban Disturbance—and derive design and policy directions for multispecies coexistence. The central contribution is a thematic map of the problem space plus a set of recommendations for more animal-aware design and regulation.","tokens_in":23538,"tokens_out":4040,"duration_ms":41488,"significance":"If the thematic map is accepted, the paper provides a useful synthesis for HCI, automotive UI, and robotics researchers. Its strengths are the documented search logs (Appendix A), the full enumerated list of videos and Reddit threads (Tables 7 and 8), explicit screening criteria, and a candid limitations section. The framing of animals as relational actors rather than obstacles is a constructive contribution. The main risks are methodological: the ethnography is coded with the same themes derived from the literature, and the incident reports are unverified user-generated content, so the multi-method convergence asserted in the abstract is weaker than the paper's presentation suggests.","major_comments":[{"comment":"The five themes are derived inductively from the scoping review, and the ethnography is then coded with 'the same five thematic categories from the scoping review' (Section 3.2, Data Analysis). The correspondence between the literature and the online ethnography is therefore partly by construction, yet the abstract and Section 5 present the multi-method design as surfacing and converging on five areas of concern. To support the convergence claim, the ethnographic items would need to be coded openly or by coders blind to the literature themes, with inter-coder agreement reported; alternatively, the ethnography should be explicitly presented as an illustrative extension rather than an independent confirmation. As written, the abstract's phrase 'Our analysis surfaces five key areas of concern' overstates the independence of the ethnography.","section":"3.1, 3.2, and Abstract"},{"comment":"Claims such as 'Collisions were documented in two Tesla FSD videos in which stationary deer were not detected at night' and the Reddit account of a Tesla striking a dog while in FSD mode rest on unverified user-generated footage and posts. The paper does not triangulate these accounts with manufacturer data, system logs, or official incident reports, and Section 6 acknowledges that the corpus is user-curated and may skew toward unusual events. Because Physical Impact is one of the five headline themes, these incidents need to be either verified against independent sources or explicitly treated as anecdotal and illustrative; the current Results section gives them evidentiary weight that the source material cannot support.","section":"4.1.1, Ethnography paragraph"},{"comment":"The expert interview analysis is described as inductive thematic analysis, but the third interview phase presented 'selected findings from the scoping review and online ethnography' to the experts, and Figure 3 maps expert insights onto the same five categories. This makes it difficult to know whether the experts independently generated the five themes or were responding to them. Please clarify whether the expert themes were elicited before or after exposure to the five categories, and consider reporting the interview guide or at least the order of questions so that readers can assess the degree to which the expert data are independent confirmation.","section":"3.3 and 4.2, Expert interviews"}],"minor_comments":[{"comment":"The text reports 681 Google Scholar results, but Table 4 sums to 426; the stated grand total of 1,231 corresponds to 805 + 426, so the 681 figure appears to be a typographical or arithmetic error.","section":"Section 3.1 vs Appendix A.2, Table 4"},{"comment":"The acknowledgments name 'Paul Schmidt', but the author list and expert profiles give 'Paul Schmitt'; the spelling should be made consistent.","section":"Acknowledgments"},{"comment":"The heading reads 'Google Scholar keyword search eesults' and should read 'results'.","section":"Appendix A.2, Table 4 heading"},{"comment":"The sentence beginning 'While this paper does not argue for a shift to animal-centred design, but rather, the need...' is a run-on and should be rewritten for clarity.","section":"Section 5.2, first paragraph"},{"comment":"The screening process is described as conducted by the first author and then reviewed by the first and second authors, but no inter-coder reliability or conflict-resolution procedure is reported; a brief statement on how disagreements were resolved would strengthen the trustworthiness of the thematic synthesis.","section":"Section 3.1, Screening process"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is well within scope for AutomotiveUI, and the transparency of the appendices is a genuine strength. The two main concerns—the partly constructed convergence between the scoping review and the ethnography, and the reliance on unverified user-generated incident reports—are fixable within the paper's scope through reframing and more careful epistemic claims. I do not see a need for rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nRead the Tran et al. paper on animal interaction with autonomous mobility. It is a useful and unusually transparent piece of qualitative synthesis. It ships its homework: full search logs in Appendix A, complete lists of YouTube videos and Reddit threads, expert profiles, and an explicit limitations section. The genuinely new content is the catalogued ethnographic corpus and the eight expert interviews, and they feed a sensible set of design and policy directions. The five themes (Physical Impact, Behavioural Effects, Accessibility, Ethics and Regulations, Urban Disturbance) line up with the empirical studies the paper cites, such as Rasmussen on hedgehogs and Väätäjä on dogs, so they are not invented from nothing.\n\nThe main soft spot is structural. The scoping review produced the five themes inductively; the ethnography was then coded with those same five themes. The claimed multi-method convergence is therefore partly built into the method. The authors are transparent about doing this, but the abstract's 'our analysis surfaces five key areas' invites a stronger reading than the analysis supports. That is a moderate problem, not a fatal one. The taxonomy still works as an analytic frame; it just is not an independently confirmed clustering.\n\nA few specific fixes. The 'European Union’s Green Smart Directive' is cited to Hassenzahl et al. 2022, whose title makes clear it is speculative design fiction, not a real directive; presenting it as policy shaping garden robot design is a factual error. The 85 dB noise and ultrasound claims rest on human-oriented citations, not animal studies. The internal numbers are inconsistent (681 vs 426 Google Scholar results; '40 plus 7' vs 45 papers). And the ethnographic incidents are unverified, user-generated clips; the paper acknowledges selection bias, but should be careful not to imply confirmed system behaviors.\n\nAll of this is correctable. The paper deserves a serious referee and a request for revision, not a desk reject. It gives HCI and ACI researchers a structured map of an overlooked problem and a reasonable starting point for design work. I would bring it to reading group and, if I worked in this area, I would cite it. Recommend: accept with revisions.","headline":"A transparent, useful mapping of an overlooked intersection; the five-theme convergence is partly built into the method, and a couple of factual slips need fixing.","tokens_in":24222,"tokens_out":3353,"would_cite":true,"duration_ms":32530,"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":"Animal encounters with autonomous vehicles, drones, and robots cluster into five shared areas of concern—physical impact, behavioral effects, accessibility, ethics and regulation, and urban disturbance—that current design and policy…","keywords":["animals","wildlife","autonomous vehicles","delivery robots","drones","animal-computer interaction","more-than-human design","multispecies coexistence"],"falsifier":"Compare the animal collisions and near-misses described in the paper's 39 videos and 11 Reddit discussions against official incident data for the same vehicle models and periods (e.g., NHTSA standing general orders for autonomous vehicles, Waymo incident reports, or Tesla FSD disengagement records). If logged animal encounters are far rarer or qualitatively different—for example, if no stationary-deer night collisions appear in any official record—the Physical Impact theme would be substantially weakened.","tokens_in":23184,"feed_emoji":"🐾","tokens_out":5302,"duration_ms":51270,"temperature":0.7,"pith_summary":"This paper argues that animals interacting with autonomous mobility systems—self-driving cars, delivery robots, drones, and lawn-mower robots—are affected in five recurring ways: physical impact (collisions and detection failures), behavioural effects (stress, avoidance, aggression), accessibility concerns (especially for guide dogs and their handlers), ethics and regulations (human-life prioritisation and standards gaps), and urban disturbance (ripple effects on public spaces). To establish this, the authors combine a scoping review of 45 papers, an online ethnography of 39 YouTube videos and 11 Reddit discussions, and interviews with eight experts. The intended contribution is to reframe animals as relational actors rather than edge-case obstacles, and to give designers and policymakers a validated thematic map for multispecies coexistence. If the paper is right, animal welfare becomes a standing input to autonomous-system design and regulation rather than an afterthought.","feed_headline":"New study maps five ways autonomous systems affect animals","feed_subtitle":"A 45-paper review, 50 real-world encounters, and 8 expert interviews converge on one taxonomy for animal-friendly design.","key_machinery":"The carrying mechanism is the multi-method triangulation built around a single five-category thematic framework. A scoping review of 45 papers supplies the academic landscape, an online ethnography of 39 YouTube videos and 11 Reddit discussions supplies real-world encounters and user discussions, and eight expert interviews (ecologists, animal behaviourists, and mobility and ACI researchers) interpret and extend both. The five themes—Physical Impact, Behavioural Effects, Accessibility Concerns, Ethics and Regulations, Urban Disturbance—act as the analytic grid into which all evidence is coded, allowing the authors to compare literature against observed practice and to attach design and policy directions to each area.","core_discovery":"The paper's central claim is that the scattered evidence on animal–autonomous mobility interaction converges into five key areas of concern that cut across system types and species. Physical Impact covers collisions and failures to detect, documented in the literature (e.g., hedgehogs injured by robotic lawn mowers) and in user-generated footage (e.g., two Tesla FSD videos showing stationary deer not detected at night). Behavioural Effects covers stress, avoidance, and aggression, such as dogs reacting fearfully to delivery robots and drones provoking defensive attacks from wildlife. Accessibility Concerns centres on service animals: guide dogs disoriented by silent electric vehicles and unpredictable robots, and robotaxi pet policies that exclude emotional support animals. Ethics and Regulations captures the priority given to human life in codes such as the German Ethics Code, the absence of animal protections in safety standards like EN ISO 13482, and speciesist bias in AI. Urban Disturbance covers how robot-triggered animal reactions ripple into blocked sidewalks and disrupted shared spaces. These five themes, layered with expert insights, are offered as a foundation for design and policy to support multispecies coexistence.","pith_inferences":["Beyond the paper's claims, the five-theme map could function as a checklist for AV safety cases, e.g., reporting animal near-misses as a standard metric in deployment reporting; the paper implies but does not state this.","The inconsistent visualisation of animals (a deer shown as a bird or pedestrian) suggests a testable UI hypothesis: that the way an AV explains its animal detection affects human trust and takeover decisions, which could be studied with driving simulators.","The paper's framework implies that perception datasets should be audited for species coverage; since its own data are Northern-Hemisphere centric, trials with Australian species such as wombats and kangaroos would test whether the five themes generalise.","If experts' 'habituation is species-specific' observation holds, longitudinal field studies of the same dog populations encountering delivery robots over weeks would be a natural next step that the paper does not conduct."],"forward_implications":["AV perception and planning should treat animals as species-specific actors, not static obstacles, because documented failures include stationary deer missed at night and animals visualised as wrong object classes on in-vehicle displays.","External vehicle and robot communication needs to be designed for animal senses; silent electric vehicles already confuse guide dogs judging speed and distance, and high-frequency signals may be ineffective or aversive.","Deployment guidelines for drones and delivery robots should account for behavioural stress and defensive aggression observed across many species, with habituation treated as uncertain and species-dependent.","Safety standards and regulations for autonomous mobility need explicit animal-welfare provisions, since the current reference standard for personal care robots treats animals as generic safety-related objects without specific protections.","Service-animal accessibility requires relational inclusivity: robotaxi pet policies, public stigma, and animals' unease with driverless cabs all need to be addressed for human–animal partnerships to use these systems equally."],"supporting_citations":[{"why":"Frames the responsibility agenda that motivates studying technology's risks to animals.","marker":"[45]"},{"why":"Supplies the main empirical evidence of physical impact: robotic lawn mowers injuring hedgehogs.","marker":"[53]"},{"why":"Provides survey data on dogs' behavioural reactions to delivery robots.","marker":"[67]"},{"why":"Shows that YouTube footage can complement scientific evidence on drone threats to wildlife.","marker":"[54]"},{"why":"Documents how silent vehicles confuse guide dogs, grounding the accessibility theme.","marker":"[32]"},{"why":"Offers observational evidence of sidewalk disturbances involving delivery robots and pets.","marker":"[72]"},{"why":"Supports the ethics theme by showing speciesist bias in AI training data.","marker":"[22]"},{"why":"Identifies the safety-standard gap that ignores animals as bystanders.","marker":"[58]"}],"fun_headline_variants":["Five risks when robots share streets with animals","Robot meets wildlife: five collision zones","Multispecies streets: five animal-robot conflict areas","Service dogs, deer, and delivery bots: five concerns"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The paper's real-world evidence for physical and behavioural impacts rests on unverified YouTube clips and Reddit posts that were never checked against system logs, manufacturer data, or the videographers, so the severity and representativeness of those impacts depend on the accuracy of user-generated accounts.","fun_headline_variants_meta":{"raw":{"variants":["Five risks when robots share streets with animals","Robot meets wildlife: five collision zones","Multispecies streets: five animal-robot conflict areas","Service dogs, deer, and delivery bots: five concerns"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000167,"raw_usage":{"total_tokens":1256,"prompt_tokens":944,"completion_tokens":312,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":560,"completion_tokens_details":{"reasoning_tokens":252}},"tokens_in":560,"tokens_out":312,"duration_ms":3950,"temperature":1.0,"reasoning_tokens":252,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T15:15:14.925853+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compare the animal collisions and near-misses described in the paper's 39 videos and 11 Reddit discussions against official incident data for the same vehicle models and periods (e.g., NHTSA standing general orders for autonomous vehicles, Waymo incident reports, or Tesla FSD disengagement records). If logged animal encounters are far rarer or qualitatively different—for example, if no stationary-deer night collisions appear in any official record—the Physical Impact theme would be substantially weakened.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides survey data on dogs' behavioural reactions to delivery robots."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the main empirical evidence of physical impact: robotic lawn mowers injuring hedgehogs."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Shows that YouTube footage can complement scientific evidence on drone threats to wildlife."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Offers observational evidence of sidewalk disturbances involving delivery robots and pets."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Identifies the safety-standard gap that ignores animals as bystanders."}],"review_version":1}