{"id":"c434da1b-aa61-4836-bf75-1ff5ec2cfdb2","arxiv_id":"2507.00523","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":3.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A survey of edge robotics research from 2015 to 2025 that classifies recent work into seven application areas and identifies open challenges.","lead":"This paper reviews how edge computing, doing computation near robots instead of in distant cloud servers, is being applied in robotics. It organizes recent studies by application area and argues that edge robotics is promising but rarely tested under real-world conditions.","discovery_kind":"review","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central comprehensiveness claim rests on an unreported literature search with no inclusion/exclusion criteria, no article counts, and the paper's own admission that the survey is not exhaustive; absent a reproducible search protocol, the claim of being the first comprehensive edge robotics…","rationale":"The reader's weakest_assumption is also the most load-bearing condition for the abstract's claim. My stress-test confirms it and sharpens it: the search description in I.B is not merely undocumented in a cosmetic sense, it provides no way to verify that the 2015-2025 sample is representative; and Section VII's 'not exhaustive' sentence is a stated limitation that must be weighed, per the reviewing rule. I do not see a separate fatal flaw in the technical summaries: the paper's critical evaluations in Section V and challenge list in Section VI are largely consistent with the cited works, and the taxonomy (Table II and Fig. 1) is a reasonable organizing device. The conditional verdict is appropriate because the paper's contribution can be salvaged by either narrowing 'comprehensive' to 'structured overview' or adding the missing search protocol and counts; no deeper correctness issue forces rejection.","tokens_in":23432,"tokens_out":3804,"duration_ms":47041,"concrete_test":"Reproduce and expand the Section I.B search: run the four stated terms plus corrected 'multi-robot systems' and synonyms ('fog robotics', 'MEC robotics', 'cloudlet robotics', 'edge-assisted SLAM', 'edge-enabled navigation') across IEEE Xplore, ScienceDirect, ACM DL, Scopus, arXiv, and Web of Science for 2015-2025; record unique records screened and included. Then check whether any included papers address application categories in Table III (e.g., localization, navigation, latency minimization) that are absent from the reference list; if the expanded search surfaces more than a handful of such missed works in any category, the comprehensiveness claim fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The abstract's central claim is that no recent survey comprehensively examines edge-computing benefits and that this paper bridges that gap. The support for that claim is Section I.B's description of a literature review in IEEE Xplore, ScienceDirect, and ACM Digital Library using four search terms. This description is not a reproducible protocol: no exact query strings, no date filter details, no deduplication or screening rules, no counts of retrieved/screened/included papers, and no PRISMA-style flow. Section VII then states the survey 'is not exhaustive.' That internal admission directly undercuts the 'comprehensive' claim. Because the survey's novelty is defined against prior surveys it itself cites ([16]-[20], [44], [87]), the claim can only be evaluated if the search is shown to cover the field; currently it cannot be. If relevant work in arXiv-only venues, workshops, or adjacent terms like fog robotics and cloudlets is missing, the seven-way classification is not a comprehensive map but a selected overview. This is a load-bearing, falsifiable weakness, not a stylistic issue.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper is a survey of edge computing applied to robotics, covering work from approximately 2015 to 2025. It introduces and distinguishes cloud, fog, edge, MEC, mist, and cloudlet paradigms, then organizes the recent literature into seven application-driven categories: computational offloading, context awareness, localization, navigation, latency minimization, resource optimization, and energy minimization. For each category it summarizes representative papers, provides a comparison table of objectives (Table III), and closes with research gaps, future challenges, and a conclusion. The paper's central claim, stated in the abstract and Section I.B, is that no recent survey comprehensively examines the benefits of edge computing in robotics and that this paper fills that gap.","tokens_in":23615,"tokens_out":8153,"duration_ms":92712,"significance":"If the comprehensiveness claim were supported, the paper would offer a useful organizing framework for the rapidly growing edge-robotics literature: its seven-category application classification, the background taxonomy of computing paradigms, and the candid assessment of experimental limitations in Section V are all potentially valuable contributions. The paper also makes a useful attempt to separate what has been validated in real-world systems from what has only been evaluated in simulation or controlled testbeds. However, the central novelty claim is not currently verifiable because the literature search is not described reproducibly, the survey itself concedes that it is not exhaustive, and the comparison against prior surveys omits at least one directly overlapping recent survey. These issues are load-bearing rather than cosmetic, because the paper's contribution is defined in opposition to the comprehensiveness of earlier surveys.","major_comments":[{"comment":"The abstract's claims that 'there has been no recent survey that comprehensively examines these benefits' and that the paper 'provides a comprehensive evaluation' are not supported by the reported methodology. Section I.B names three databases and four search terms, but it gives no exact query strings, no screening or inclusion/exclusion criteria, no deduplication procedure, no counts of retrieved, screened, or included papers, and no PRISMA-style flow. Section VII then states that the survey 'is not exhaustive.' Since the paper's novelty is explicitly defined by its comprehensiveness relative to prior surveys, the authors must either provide a reproducible search and selection protocol with article counts, or reframe the contribution as a selected overview rather than a comprehensive survey.","section":"I.B and VII"},{"comment":"The positioning against prior surveys is incomplete in a directly relevant way. Reference [87], Chaari et al., 'Dynamic computation offloading for ground and flying robots: Taxonomy, state of art, and future directions' (Computer Science Review, 2022), is a recent survey that provides a taxonomy and state-of-the-art review of dynamic computation offloading for ground and flying robots, which substantially overlaps with Section IV.A and with the paper's stated scope. This reference is cited later in Section VI.B but is not discussed in Section I.B when the authors distinguish their work from existing surveys. The authors should explain how their survey goes beyond [87] in coverage, categorization, or critical assessment, and should position themselves explicitly against it.","section":"I.B and reference [87]"},{"comment":"The comparative table is not a reliable audit of the objectives claimed for each paper. For example, the row for Tahir et al. [49], [50] marks Localization and Navigation as addressed, but the surrounding text in Section IV.A describes utility-aware task offloading and decentralized multi-robot scheduling, with no mention of localization or navigation outcomes. Conversely, the row for Yin et al. [75] marks Minimizing Energy Consumption as not addressed, even though Section IV.E states that the paper's two resource-management schemes are explicitly 'minimizing and balancing energy usage' across slave robots. Because Table III is the paper's main comparative instrument, the authors should define a coding protocol for the checkmarks and ensure each mark is traceable to the cited work or to a specific statement in Section IV.","section":"Table III and Sections IV.A, IV.E"}],"minor_comments":[{"comment":"The article-organization list in Section I.C jumps from Section IV to Section VI and omits Section V ('Research Gaps and Limitations'); this should be corrected to match the actual structure and Figure 3.","section":"I.C"},{"comment":"The search term 'multi-edge servers in mutirobot systems' contains a typo; it should read 'multi-robot systems'.","section":"I.B"},{"comment":"The name 'Lui et al. [62]' should be 'Liu et al. [62]' to match the reference list and the corresponding authors of that work.","section":"IV.C and Table III"},{"comment":"The text refers to 'Qingqin et al. [71]', but the cited work is by 'L. Qingqing et al.'; the spelling should be consistent with the reference.","section":"IV.D"},{"comment":"The sentence 'it relies on multiple links for data transport, which is a significant disadvantage of fog computing' has an unclear antecedent; the contrast between edge and fog in this sentence should be rewritten for clarity.","section":"II.E"},{"comment":"The reference to 'A. J. Ben et al. [65]' is inaccurate; the author name is 'A. J. Ben Ali et al.', and it should be spelled consistently with Section IV.C.","section":"V and reference [65]"}],"recommendation":"major_revision","confidential_remarks":"The manuscript's main weakness is the gap between the 'comprehensive' promise in the abstract and the undocumented, non-exhaustive selection process described in Section I.B. If the authors are not willing to add a full search protocol, they should soften the novelty claim and present the paper as a structured survey of selected representative works. The Table III inconsistencies should be fixed regardless, because the comparative table is one of the paper's main contributions."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First, the practical take: this is a useful reference survey for anyone entering edge robotics, and Section V is genuinely better than the usual survey filler. But the abstract's 'comprehensive' claim is not supported, and the authors know it—the conclusion says 'not exhaustive.' That internal contradiction is the main thing to fix.\n\nWhat the paper does well: the seven application categories give newcomers a workable map. The summaries of roughly thirty systems are consistent with the original papers as far as I have checked. Table III's objective matrix is a simple but handy comparison tool. And Section V is the real contribution: it names concrete experimental weaknesses—idealized network assumptions, missing benchmarks, small testbeds, no energy accounting—rather than gesturing at 'future work.' That kind of candid critical review is rare in surveys and should be kept.\n\nThe soft spots: the literature search is not reproducible. One I.B paragraph says they used IEEE Xplore, ScienceDirect, and ACM DL with four terms, but gives no query strings, date filtering, screening, or counts. The 'first comprehensive survey' claim therefore cannot be verified, and the conclusion's 'not exhaustive' undercuts it. This is an easy fix: soften the claim or add a systematic protocol. Minor: the authors' own papers appear four times in the state-of-the-art table; that is not improper since they are central to multi-edge offloading, but it would be worth double-checking that no equally relevant outside work got crowded out. Also, one search term is misspelled ('mutirobot').\n\nWho benefits: graduate students and researchers wanting a structured entry into edge robotics. The taxonomy is not radical, but it is serviceable. The paper proposes no new algorithms or data, which is fine for a survey.\n\nRecommendation: this deserves a serious referee. I would accept it on condition that the authors either narrow the 'comprehensive' language or document the search protocol. The core content is honest and useful enough to warrant that effort.","headline":"A useful, candid survey of edge robotics that is undermined mainly by its own 'comprehensive' claim, which its methodology cannot support and its conclusion explicitly contradicts.","tokens_in":24130,"tokens_out":2296,"would_cite":true,"duration_ms":29835,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The first comprehensive survey of edge robotics from 2015 to 2025 organizes the field into seven application categories and a comparative objective table.","keywords":["edge computing","edge robotics","cloud robotics","fog robotics","multi-robot systems","computational offloading","simultaneous localization and mapping","latency minimization"],"falsifier":"A reader could rerun the search with additional databases (Scopus, Web of Science, Google Scholar) and broader terms such as 'edge intelligence' or 'MEC robotics,' then check whether any earlier survey already provides a comprehensive application-domain classification of edge robotics across 2015–2025; if such a survey exists, the paper's 'no recent comprehensive survey' premise is false. Alternatively, re-coding the same corpus with the paper's seven categories should reproduce the reported gap analysis; any major disagreement would show the taxonomy is not as stable as presented.","tokens_in":23219,"feed_emoji":"🤖","tokens_out":10018,"duration_ms":95881,"temperature":0.7,"pith_summary":"This survey aims to establish that edge robotics—running computation, storage, and intelligence on servers close to robots rather than in distant clouds—has become a distinct research domain that lacked an up-to-date comprehensive review. The authors argue that cloud robotics suffers from multi-hop latency that hurts time-sensitive tasks, and that edge computing answers that problem while adding mobility support, location awareness, and distributed parallel processing. To support this, they review work published between 2015 and 2025, gathered from IEEE Xplore, ScienceDirect, and the ACM Digital Library, and classify it into seven application categories: computational offloading, context awareness, localization, navigation, latency minimization, resource optimization, and energy minimization. The survey's comparative table shows that most systems target offloading and latency, while context awareness, security, handover, fault tolerance, and realistic multi-robot evaluation are underexplored. If accurate, this gives the community a shared map of what edge robotics can do and where the next work is needed.","feed_headline":"Survey sorts a decade of edge robotics into seven application classes","feed_subtitle":"A decade of edge-robotics papers, sorted by objective, shows what works and what remains untested in deployment.","key_machinery":"The load-bearing device is the seven-category application taxonomy realized in a comparison table that checks each surveyed system against the objectives it addresses: computational offloading, context awareness, localization, navigation, minimizing latency, resource optimization, and minimizing energy consumption. That table is what converts a collection of prototypes into evidence that the field clusters around a few goals and neglects others. Around it, the paper builds a conceptual taxonomy of computing paradigms—cloud, fog, edge, MEC, cloudlets, mist—that defines what counts as edge robotics and why it differs from cloud robotics and fog robotics. The paradigm taxonomy supplies the definitions; the application table supplies the map.","core_discovery":"The central claim is that edge robotics is a maturing but still young field whose literature can be systematically organized, and that this organization reveals a clear pattern. The paper positions edge robotics on a spectrum running from cloud robotics through fog robotics, Mobile/Multi-Access Edge Computing (MEC), and cloudlets to mist computing, and draws the operational line between edge and fog: edge systems can process data entirely at the network edge without involving a cloud, while fog systems sit between edge and cloud and forward preprocessed data upward. It then reviews state-of-the-art systems and codes each one against seven objectives, producing the field's first application-domain classification. The resulting pattern is that computational offloading and latency reduction dominate the literature, whereas security, context awareness, handover mechanisms, network-failure resilience, and fault tolerance receive little attention, and most experiments run under idealized network conditions or in small testbeds. The conclusion the paper draws is that edge robotics has real potential but remains in its early stages, with the open challenges laid out as a research agenda.","pith_inferences":["I would treat the seven-category table as a candidate coding standard for future surveys; the paper itself does not propose standardization, but if the community adopted it, cross-paper comparisons of objectives would become straightforward.","The observation that partial offloading often beats full offloading in the reviewed drone-navigation studies suggests a transferable design heuristic—offload compressed features rather than raw imagery—that the survey does not itself elevate to a principle.","The comprehensiveness claim could be stress-tested by extending the search to databases and terms the paper did not use; because inclusion and exclusion criteria are not reported, the representativeness of the corpus is the survey's most testable assumption.","A natural extension of the gap analysis is to add security, handover, and fault tolerance as explicit columns in the application table, which would make the under-explored objectives as visible as latency and resource use."],"forward_implications":["New edge-robotics papers can be positioned against a standard seven-category taxonomy, making it easier to see which objective combinations have already been tried.","The gap list in the survey translates directly into a research agenda: security, context awareness, handover, network-failure handling, fault tolerance, and heterogeneity are the named open problems.","Since the review finds that most evaluations assume stable wireless links or use small testbeds, the practical conclusion is that edge-robotics performance claims need validation under real-world, multi-robot, variable-bandwidth conditions.","The paradigm taxonomy gives deployment guidance: choose edge computing when low latency and fully local processing matter, fog when cloud mediation is acceptable, and cloud when scalability and centralized storage dominate.","Including MEC-based robotics inside edge robotics means the field's scope covers telecom-standardized edge servers inside radio access networks, not only local gateways."],"supporting_citations":[{"why":"Prior survey that experimentally evaluates end-to-end edge robotic systems but lacks a broad literature classification; the paper positions its comprehensiveness against this.","marker":"[16]"},{"why":"Cross-domain robotics review that covers taxonomy and human-machine interaction but not edge computing in multi-robot systems, marking the gap.","marker":"[17]"},{"why":"Domain-limited survey of edge computing in agriculture robotics, one of the prior field-specific reviews the paper distinguishes itself from.","marker":"[18]"},{"why":"Survey of edge computing for Industry 5.0 that addresses industrial robots in only one section, another example of non-comprehensive coverage.","marker":"[19]"},{"why":"Survey of edge-enabled healthcare service robots, showing existing reviews are confined to single application domains.","marker":"[20]"},{"why":"Foundational cloud-robotics architecture reference whose remote-server latency drawback motivates the turn to edge robotics.","marker":"[11]"},{"why":"Source of the term 'cloud robotics,' establishing the remote-computation framing that edge robotics extends.","marker":"[21]"}],"fun_headline_variants":["Edge robotics survey finds offloading focus, security gap","Seven objectives sort edge robotics; resilience is rare","Edge computing meets robotics: survey reveals blind spots","Edge robotics literature: latency wins, fault tolerance loses","Survey of edge robotics: young field, uneven coverage"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the literature search over IEEE Xplore, ScienceDirect, and the ACM Digital Library using four specified search terms captured a representative and complete sample of edge-robotics work from 2015 to 2025; if important works fall outside those databases or terms, the survey's comprehensiveness claim weakens.","fun_headline_variants_meta":{"raw":{"variants":["Edge robotics survey finds offloading focus, security gap","Seven objectives sort edge robotics; resilience is rare","Edge computing meets robotics: survey reveals blind spots","Edge robotics literature: latency wins, fault tolerance loses","Survey of edge robotics: young field, uneven coverage"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000467,"raw_usage":{"total_tokens":2327,"prompt_tokens":941,"completion_tokens":1386,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":557,"completion_tokens_details":{"reasoning_tokens":1312}},"tokens_in":557,"tokens_out":1386,"duration_ms":11386,"temperature":1.0,"reasoning_tokens":1312,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T21:12:06.144697+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A reader could rerun the search with additional databases (Scopus, Web of Science, Google Scholar) and broader terms such as 'edge intelligence' or 'MEC robotics,' then check whether any earlier survey already provides a comprehensive application-domain classification of edge robotics across 2015–2025; if such a survey exists, the paper's 'no recent comprehensive survey' premise is false. Alternatively, re-coding the same corpus with the paper's seven categories should reproduce the reported gap analysis; any major disagreement would show the taxonomy is not as stable as presented.","supporting_citations":[{"cited_title":"Edge robotics: Are we ready? an experimental evaluation of current vision and future directions,","cited_arxiv_id":null,"evidence_quote":"Prior survey that experimentally evaluates end-to-end edge robotic systems but lacks a broad literature classification; the paper positions its comprehensiveness against this."},{"cited_title":"Robotics 4.0–are we there yet?","cited_arxiv_id":null,"evidence_quote":"Cross-domain robotics review that covers taxonomy and human-machine interaction but not edge computing in multi-robot systems, marking the gap."},{"cited_title":"Edge computing and robotic appli- cations in modern agriculture,","cited_arxiv_id":null,"evidence_quote":"Domain-limited survey of edge computing in agriculture robotics, one of the prior field-specific reviews the paper distinguishes itself from."},{"cited_title":"Edge computing for industry 5.0: Fundamental, applications, and research challenges,","cited_arxiv_id":null,"evidence_quote":"Survey of edge computing for Industry 5.0 that addresses industrial robots in only one section, another example of non-comprehensive coverage."},{"cited_title":"Cognitive computing and wireless communications on the edge for healthcare service robots,","cited_arxiv_id":null,"evidence_quote":"Survey of edge-enabled healthcare service robots, showing existing reviews are confined to single application domains."},{"cited_title":"Cloud-enabled robots in: Ieee-ras international conference on humanoid robots,","cited_arxiv_id":null,"evidence_quote":"Source of the term 'cloud robotics,' establishing the remote-computation framing that edge robotics extends."}],"review_version":1}