{"id":"c718761c-b4b8-4965-b3cd-c0b6f149788d","arxiv_id":"2611.05155","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Asynchronous Robot Policies decouple slow semantic reasoning from fast action generation, and keep LIBERO success stable even when the vision-language model runs 10 times slower.","lead":"This paper proposes an asynchronous robot-policy architecture that lets a large vision-language model fill a semantic memory slowly while a small vision-action model generates fast motor commands, and reports better success and speed than synchronous vision-language-action baselines on Franka manipulation tasks. The result matters because it offers a path to keep large foundation models in the control loop for reactive manipulation without inheriting their latency.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek/deepseek-v4-flash","headline":"The dynamic-task gains are attributed to asynchronous decoupling, but no control isolates asynchrony from staleness augmentation: the decisive cell - a synchronous policy with the same delayed conditioning (TRTC/VLASH-style) - is never run.","rationale":"I take the central claim to be causal and architectural: removing the VLM from the perception-action loop (asynchronous decoupling) is what enables high-frequency reactive control, and this decoupling is what produces the 3-5x dynamic-task gains and the VLM-rate robustness. Three conditions must hold: (A) a fast expert conditioned on real-time vision plus a stale memory is competent; (B) scaling the VLM still buys capability; (C) the dynamic gains are causally due to decoupling rather than to a separable training component. (A) is supported: the Tab. 2-3 margins are large enough to survive sampling noise, and the task-identity-vs-position decomposition is coherent - the stale memory supplies instruction and object identity, which are stable across a perturbation, while real-time vision supplies the new pose. (B) is fragile: Tab. 1 shows only a few-pp gains from 1B to 8B concentrated on the Long suite, against a replicated π0 that is 14 pp below the published π0 (94.2% still beats every ARP variant), and the abstract's 18% headline (b7, b33) is not reproducible from any table; the capability half of the trade-off claim therefore rests on a weak, unvalidated baseline and an untraceable number. (C) is the least secure and is my primary concern: the dynamic comparisons confound decoupling with staleness training. Credit where due: Tab. 4 is a genuinely informative internal control and establishes necessity of staleness training (the d_t ≡ 0 row collapses at 0.5 Hz); the default-row result (81.8% at 0.5 Hz) is the paper's strongest evidence; the latency math is definitional since the VLM is off the fast path; and the authors honestly flag the open lower-frequency limit (b255) and the lexical-proxy caveat (b283), which are acknowledged scope limits rather than my objection. My objection is the missing 2x2 cell: sync architecture plus staleness augmentation, exactly the TRTC/VLASH recipe the authors cite as prior work. If that cell matches ARP's dynamic numbers, the decoupling story reduces to training-time latency compensation; if it does not, the central claim is confirmed. This is a runnable experiment the revision should include, alongside seed-level variance and the code release promised in the reproducibility statement. CONDITIONAL remains the right verdict; my read sharpens the conditions rather than changing the verdict.","tokens_in":19614,"tokens_out":25197,"duration_ms":180977,"concrete_test":"Train the same π0-1B architecture used in the replicated baseline with ARP's staleness/latency augmentation - sample d_t from U{1,...,40} (Eq. 3) and condition the flow-matching objective on staleness and execution delay in the style of Training-Time RTC (Black et al., 2025) or VLASH (Tang et al., 2025) - keeping data, seeds, and the Sec. B/C benchmark protocols identical. Evaluate on LIBERO-Perturb and Dynamic Bump exactly as in Tab. 2-3. If this synchronous augmented policy reaches within ~10 pp of ARP-1B (51.5% / 44.7%), the reactivity gains are attributable to training-time augmentation rather than to asynchronous decoupling, and the central claim requires revision; if it remains near the un-augmented π0-1B levels (9.5% / 12.7%), the decoupling is the operative mechanism.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The headline reactivity evidence (Tab. 2-3: ARP-1B 51.5% vs π0-1B 9.5% on LIBERO-Perturb; 44.7% vs 12.7% on Dynamic Bump) is presented as proof that asynchronous decoupling breaks the reactivity-capability trade-off. But every one of these comparisons varies two factors at once: the asynchronous architecture (VLM moved off the action path) and training-time staleness/latency augmentation (Eq. 3, Sec. 3.4), which the π0-1B baseline does not receive. The paper establishes that staleness training is necessary - Tab. 4 shows the async interface alone (d_t ≡ 0) collapses at slow refresh - but it never runs the complementary cell: a synchronous policy trained with the same staleness distribution and delay conditioning, i.e., the recipe of the cited Training-Time RTC (Black et al., 2025) and VLASH (Tang et al., 2025). The 'Vision-Gated VLA' ablation in Sec. 4.5 does not fill this gap: it drops the semantic memory (removing the augmentation) rather than keeping augmentation while removing asynchrony. If a sync-plus-augmentation π0-1B matches ARP's 51.5%/44.7%, the dynamic gains are caused by training-time latency compensation, which is prior art, not by the decoupling the paper claims as the operative contribution; if it stays near 9.5%/12.7%, decoupling is confirmed. Secondary but relevant: the abstract's '18% higher success in quasi-static setups' (b7, b33) does not match any ARP-vs-baseline comparison in Tab. 1 (largest relative gain is roughly 8%), and the replicated π0 (80.2%) sits 14 pp below the published π0 (94.2%), so the capability half of the trade-off claim rests on an unvalidated baseline and an untraceable headline.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek/deepseek-v4-flash","summary":"This paper proposes Asynchronous Robot Policies (ARP), a two-loop architecture for vision-language-action control. A high-capacity VLM runs as a low-frequency 'semantic expert' that writes a compact, fixed-width memory (a projected KV-cache of its internal activations, Eq. 2), while a lightweight 'vision-action expert' runs at a higher control frequency, generating actions from the current observation, the latest (possibly stale) memory, and explicit staleness and execution-delay conditioning (Eq. 1). To make the expert robust to delayed semantic context, training samples memory staleness from a mixture distribution with point mass at zero and a uniform band up to d_max (Eq. 3). The paper evaluates ARP on LIBERO (quasi-static), two new dynamic benchmarks (LIBERO-Perturb with a 5 cm mid-episode object displacement; Dynamic Bump with velocity impulses), VLM rate/compute throttling sweeps, and DROID-scale training with RoboLab simulation and real Franka deployment. The central claims are that ARP matches or improves quasi-static success, achieves 3-5x the dynamic success of a pi0-style baseline, tolerates a 10x reduction in VLM refresh rate with negligible loss, and thereby breaks the reactivity-capability trade-off of monolithic VLA pipelines.","tokens_in":20113,"tokens_out":18391,"duration_ms":137618,"significance":"ARP addresses a genuine deployment bottleneck: the latency of large VLMs inside the control loop. The design is clean, with the memory interface making the action-loop latency independent of backbone scale by construction (Sec. 3.1), and the paper has real strengths: a controlled LIBERO protocol, a careful staleness-distribution ablation with 500 episodes per cell (Tab. 4), deterministic seeding for bit-identical dynamic comparisons (App. C), two reusable dynamic benchmarks, DROID-scale training, and real-robot evaluation. The most solid finding is that the breadth of the training-time staleness band, rather than the cached-context interface alone, is what enables slow-refresh robustness (Tab. 4). The main weakness is attribution: the dynamic gains are presented as evidence for asynchronous decoupling, but no experiment isolates decoupling from training-time staleness augmentation, which the paper itself identifies as prior art. With one decisive control experiment and corrected headline numbers, the contribution is publishable.","major_comments":[{"comment":"The central attribution claim is confounded. Every dynamic comparison (ARP-1B vs pi0-1B at 51.5% vs 9.5% and 44.7% vs 12.7%) varies two factors at once: the asynchronous architecture and the training-time staleness/latency augmentation of Eq. (3), which the pi0 baseline does not receive. Tab. 4 establishes that the augmentation is necessary (the d_t=0 row collapses at 2.5-0.5 Hz), but no experiment runs the complementary cell: a synchronous pi0-style policy trained with the same staleness mixture and delay conditioning, i.e., the recipe of the cited Training-Time RTC (Black et al., 2025) and VLASH (Tang et al., 2025). The 'Vision-Gated VLA' ablation does not fill this gap because it drops the semantic memory rather than keeping the augmentation while removing asynchrony. If sync-plus-augmentation matches ARP's dynamic numbers, the gains are attributable to training-time latency compensation, which is prior art, and the claim that asynchronous decoupling itself 'breaks the reactivity-capability trade-off' (Sec. 1, Fig. 1b) would need to be substantially re-scoped. I recommend running this control, or directly benchmarking the cited TRTC/VLASH recipes, and reporting the outcome.","section":"Sec. 4.3 (Tabs. 2-3), Sec. 3.4 (Eq. 3), Sec. 4.5","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The headline '18% higher success rate and 20% less time need for execution in quasi-static setups' is not supported by any single comparison in the paper. In Tab. 1 the largest ARP-vs-pi0 relative gain on the average is roughly 8-11% (ARP-8B/14B vs pi0-1B-replicated); the closest cells to 18% are the Long-suite comparison (62.4 vs 51.8, +20.5% relative) and ARP-1B vs Diffusion Policy (+17.1%), neither of which is a clean 'conventional VLA baseline' average. The 20% time reduction appears only in the DROID/RoboLab evaluation (16.8 s vs 20.9 s, Fig. 3), where ARP's success is below pi0.5, and no execution-time table is provided for the LIBERO quasi-static experiments. The abstract's '3-5x' also overstates the p=0.5 cell of Tab. 2 (46.7 vs 20.0, i.e., 2.3x). These numbers should be reconciled with the tables or re-specified with their exact basis.","section":"Abstract and Sec. 1 (b7, b33) vs. Tabs. 1, 6 and Fig. 3","authors_acknowledge":null,"authors_acknowledge_p":null}],"minor_comments":[{"comment":"The 'ARP vs. Vision-Gated VLA' paragraph is ambiguous about what is ablated; it reads as if it refers to the d_t=0 row of Tab. 4, but that row retains the semantic-memory channel, so please state explicitly whether the memory is removed or merely trained at zero lag.","section":"Sec. 4.5","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"Please report the number of evaluation episodes for the dynamic benchmarks and add standard errors, confidence intervals, or seed variance for the success rates; the real-robot protocols use only 10 trials per task, and the figure-only reporting there makes the margins hard to assess.","section":"Tabs. 2-3 and Sec. D.2","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The latency list '64.0ms, 64.0ms, 90.5ms, 125.ms' lacks the model-size labels (presumably 1B, 2B, 8B, 14B) and contains a typo ('125.ms'); please also define the 'nominal' VLM frequency against which the 10x reduction is computed.","section":"Sec. A.1","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"Please specify the shape of KV_tau: the text says KV caches across L intermediate layers are tapped, but the projected memory is in R^{S x D}, so the layer index L does not appear in the memory dimension.","section":"Sec. 3.2, Eq. (2)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The annotations '+1.8, +7.0, +10.0, +18.2' are unexplained, the axis is labeled 'Success rate (%)' while the annotated values appear to be deltas, and the relationship to the frequency columns of Tab. 4 should be stated explicitly.","section":"Fig. 7","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The claim that ARP is 'the first truly asynchronous robotic control framework' is too strong given that AsyncVLA and AR-VLA also place a local action expert outside the VLM path; please state the precise architectural criterion that distinguishes ARP from those systems.","section":"Sec. 1 and Sec. 2.2","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The '10x lower frequency without performance loss' claim is an interpolation result, since 0.5 Hz refresh at a 20 Hz control rate produces staleness up to the d_max=40 training bound; the envelope caveat of Sec. A.2 (b255) should be echoed in the main text where the claim is made.","section":"Sec. 4.4 vs. Sec. A.2","authors_acknowledge":null,"authors_acknowledge_p":null}],"recommendation":"major_revision"},"author_rebuttal":null,"desk_editor":{"model":"deepseek/deepseek-v4-flash","letter":"Colleague,\n\nARP is worth a serious look. It decouples a slow VLM from a fast vision-action expert, and the genuinely new piece is end-to-end training of the fast expert on stale KV-cache memory plus explicit conditioning on staleness and execution delay. The strongest evidence is Tab. 4: training with no staleness collapses at low VLM refresh, while broad staleness sampling keeps success near 80% even at 0.5 Hz. That is a clean ablation, and it earns the paper its frequency-robustness claim. The citation pattern also looks fine; the relevant prior work on asynchronous VLAs and training-time latency compensation is cited.\n\nThe soft spots are real but not fatal. The dynamic-task result (ARP 44.7–51.5% vs π0 9.5–12.7%) varies two things at once: asynchrony and staleness-augmented training. The paper never runs a synchronous policy trained with the same staleness and delay conditioning—the recipe in Training-Time RTC and VLASH. The “Vision-Gated VLA” ablation drops the memory, so it does not fill that cell either. Until that control is run, the headline “asynchrony is what makes dynamic tasks work” is not isolated; the gains could be mostly training-time latency compensation, which is prior art. The method section is appropriately careful, but the introduction’s claim that ARP “breaks the reactivity-capability trade-off” goes further than the reported experiments support.\n\nTwo smaller issues. The abstract’s “18% higher success” in quasi-static setups does not match any ARP-vs-baseline comparison in Tab. 1; the largest gap is around 8%. And the replicated π0-1B baseline sits about 14 points below the published π0 result, so the capability side of the trade-off rests on a baseline the authors did not validate. These are fixable with a corrected abstract and baseline calibration. Also, no code or checkpoints are released, and the success numbers are single-point estimates without error bars; the real-robot evaluation is small.\n\nOverall: the mechanism is plausible, the staleness ablation is honest and informative, and the dynamic gains are large enough to be worth independent replication. I would send it to review, asking for the missing synchronous-plus-augmentation control, variance reporting, baseline validation, and code before treating the reactivity claim as established. Revise, not reject.","headline":"A well-ablated, plausible async VLA architecture; the dynamic reactivity claim needs a synchronous-plus-staleness control and a cleaned-up headline before it is established.","tokens_in":20688,"tokens_out":3337,"duration_ms":25791},"pith_extraction":{"msc":["68T40","68T45","68T07"],"pacs":[],"model":"deepseek/deepseek-v4-flash","headline":"This paper claims that a robot policy can keep a large vision-language model's semantic power without inheriting its latency, by letting a lightweight vision-action expert react at control frequency on a memory the VLM writes…","keywords":["asynchronous robot policies","vision-language-action models","semantic memory","KV-cache compression","reactive robot control","robot manipulation","flow matching","staleness-aware training"],"falsifier":"Run ARP with the VLM frozen after a single memory write at episode start, or at refresh rates well below 0.5 Hz, on LIBERO-Perturb, and additionally displace the target object twice within one refresh interval; if success collapses once the number or scale of mid-interval scene changes exceeds a threshold that shrinks as refresh rate drops, the memory-sufficiency assumption fails. The gap between the frozen-memory and normal-refresh success rates would quantify how much of ARP's reactivity comes from the memory stream rather than from the expert's own visual servoing.","tokens_in":21075,"feed_emoji":"🤖","tokens_out":15074,"duration_ms":99188,"temperature":0.7,"pith_summary":"Standard vision-language-action (VLA) robot policies run the large model and the action decoder sequentially inside the control loop, so every gain in semantic capability arrives with a latency penalty that degrades real-time reactivity. This paper proposes Asynchronous Robot Policies (ARP), which runs the VLM on its own slow clock and has it periodically write a compact semantic memory, while a lightweight vision-action expert generates motor commands at control frequency, conditioned on that memory, on fresh sensor input, and on how stale the memory is. The claim is that the perception-to-action latency then collapses to the expert's time alone, so reactivity stops growing with model size. On Franka manipulation tasks the paper reports an 18% success-rate gain with 20% shorter execution in quasi-static settings, 3-5 times baseline success under mid-episode perturbations, and stable performance when the VLM runs ten times slower than nominal. If true, this matters because generalist robot policies could scale their reasoning backbone without giving up physical agility or closed-loop robustness.","feed_headline":"Robot policy splits slow AI reasoning from fast reflexes","feed_subtitle":"A lightweight expert reacts while a large model thinks, lifting dynamic-task success 3-5x.","key_machinery":"The load-bearing object is the rate-decoupling semantic memory $M_\\tau = \\mathrm{LNLinear}(KV_\\tau) + P + E_{\\mathrm{stale}}(d_t)$: a learned linear projection of the VLM's layer-wise key-value cache, plus a positional embedding and a sine embedding of the memory's age, whose dimensions stay fixed no matter how large the VLM backbone grows. Because its width never depends on the backbone, this memory is the uniform interface that lets the fast Vision-Action Expert read high-level task context without ever invoking the VLM; the expert cross-attends to the memory in every Transformer layer, while per-token adaptive normalization conditions the flow-matching trajectory on both the noise timestep and the latency pair $(d_t, \\ell_t)$ made of memory staleness and execution delay. The second essential mechanism is training-time staleness augmentation: sampling memory age from a mixture of a point mass at zero lag and a uniform band of stale horizons, which the paper's ablations show is required for the expert to act correctly on delayed semantics.","core_discovery":"The central claim is that a robot policy does not have to wait for its slow semantic model. ARP factors the policy into two asynchronous processes: a high-capacity VLM that periodically encodes the latest observation and instruction into a fixed-width memory matrix drawn from its key-value cache activations, and a Vision-Action Expert that, at every control step, cross-attends to the most recent memory while processing real-time images and proprioception. Because the fast loop's path contains no operation whose cost scales with the VLM backbone, the paper argues, the effective perception-to-action latency is reduced to the expert's inference time alone, independent of VLM parameter count. To make an outdated memory usable, the expert is trained on temporally mismatched observation-memory pairs, with staleness and execution delay supplied as explicit conditioning and with action-prefix conditioning for smooth replanning. Across LIBERO, LIBERO-Perturb, a dynamic-bump benchmark, and DROID-trained Franka evaluations, the paper reports that ARP outperforms $\\pi_0$-style synchronous baselines in quasi-static success and execution time, by wide margins under perturbations, and stays accurate when the VLM is slowed tenfold.","pith_inferences":["The two-clock pattern is a general design principle: any embodied agent pairing a slow semantic reasoner, such as an LLM, a world model, or a video predictor, with a fast controller could use a fixed-width memory interface to keep the reactive loop independent of the reasoner's cost; navigation, driving, and humanoid control are plausible targets beyond tabletop manipulation.","If the memory width stays fixed while the backbone scales, ARP's advantage should widen with scale because the marginal deployment cost of better semantics approaches zero; the paper's parameter-area plot points that way but does not measure the rate of improvement.","A testable explanation of the staleness-training bonus is that it pressures the VLM to write temporally invariant, anticipatory features; one could probe this by checking whether memory features taken from old frames predict future observations better than features from a synchronously trained VLM.","The open lower limit on refresh rate is probably set by the rate of goal-relevant scene change rather than by wall-clock time, which would argue for scheduling VLM refreshes from scene-change statistics instead of a fixed clock."],"forward_implications":["Because the fast loop's latency is the expert's time alone, backbone size can grow from 1B to 14B in these experiments, with rising LIBERO success, without lengthening the perception-to-action loop, so semantic capability and physical reactivity stop trading off against each other.","A single ARP policy keeps near-constant success while the VLM refresh frequency is cut from 5 Hz to 0.5 Hz, which implies the same robot can run its largest model at a tenth of the nominal inference frequency, saving deployment compute and energy.","Synchronous baselines collapse when the scene changes mid-episode, with 9.5% versus 51.5% success after a 5 cm target displacement and 12.7% versus 44.7% under physical bumps, so the paper identifies asynchronous execution, not better perception, as what restores closed-loop recovery.","Staleness-aware training is necessary for these gains: without it the policy falls to 2.0% success at a 0.5 Hz refresh, while a wide staleness band keeps 81.8-86.5%, and the same training also lifts fully synchronized performance.","Asynchronous action-prefix conditioning adds 5.6 percentage points of success, shortens episodes by 15.3%, and reduces jerk on DROID-trained evaluations, which the paper presents as what makes the architecture deployable on physical robots rather than only in simulation."],"supporting_citations":[{"why":"Defines $\\pi_0$, the flow-matching VLA policy that serves as the primary architectural baseline; ARP is compared against replicated $\\pi_0$-1B and $\\pi_0$-8B variants on LIBERO.","marker":"[Black et al., 2024]"},{"why":"LIBERO, the benchmark supplying the quasi-static task suites and the setting for the 5 cm target-displacement perturbation protocol.","marker":"[Liu et al., 2023]"},{"why":"DROID, the large in-the-wild manipulation dataset on which ARP is trained to test scalability to generalist policies.","marker":"[Khazatsky et al., 2024]"},{"why":"RoboLab, the photorealistic evaluation framework where DROID-trained ARP is compared against generalist baselines on 120 tasks.","marker":"[Yang et al., 2026]"},{"why":"Flow matching, the objective used by the Vision-Action Expert to generate action-trajectory chunks.","marker":"[Lipman et al., 2024]"},{"why":"AsyncVLA, the closest prior asynchronous architecture pairing a local expert with a remote VLA, from which ARP distinguishes itself by process-level asynchronous execution and end-to-end training.","marker":"[Hirose et al., 2026]"},{"why":"AR-VLA, the autoregressive action-expert work that decouples VLM cache from action generation and motivates ARP's separation of the two loops.","marker":"[Hu et al., 2026]"},{"why":"ROS 2, whose publisher-subscriber model is the stated conceptual basis for ARP's decoupled slow and fast loops.","marker":"[Macenski et al., 2022]"},{"why":"InternVL, the VLM family whose frozen vision encoder and 1B-14B LLM backbones supply ARP's semantic expert.","marker":"[Chen et al., 2024b;a]"}],"fun_headline_variants":["Asynchronous robot policy decouples slow semantic reasoning from fast action","Robot policy runs fast reflexes while the VLM thinks asynchronously","Slow VLM updates memory while lightweight expert controls robot in real time","Asynchronous robot policies lift dynamic-task success 3-5x over synchronous baselines","Asynchronous robot policy lets large model think without slowing the action loop"],"cache_read_input_tokens":0,"weakest_assumption_plain":"The scheme stands or falls on whether the compressed memory written by the slow VLM preserves enough spatial and task semantics for the fast expert to act correctly for the whole interval until the next refresh; the dynamic benchmarks exercise that assumption over short horizons, but long episodes, repeated mid-interval changes, and very slow refresh rates, whose lower limit the paper leaves open, rest on it unverified.","fun_headline_variants_meta":{"raw":{"variants":["Asynchronous robot policy decouples slow semantic reasoning from fast action","Robot policy runs fast reflexes while the VLM thinks asynchronously","Slow VLM updates memory while lightweight expert controls robot in real time","Asynchronous robot policies lift dynamic-task success 3-5x over synchronous baselines","Asynchronous robot policy lets large model think without slowing the action loop"]},"model":"deepseek/deepseek-v4-flash","effort":"low","cost_usd":0.000588,"raw_usage":{"cost":0.0005876416,"is_byok":false,"cost_details":{"upstream_inference_cost":0.0005876416,"upstream_inference_prompt_cost":3.56356e-05,"upstream_inference_completions_cost":0.000552006},"total_tokens":4171,"prompt_tokens":1138,"completion_tokens":3033,"prompt_tokens_details":{"audio_tokens":0,"video_tokens":0,"cached_tokens":933,"cache_write_tokens":0},"completion_tokens_details":{"audio_tokens":0,"image_tokens":0,"reasoning_tokens":2951}},"tokens_in":1138,"tokens_out":3033,"cost_source":"openrouter_usage","duration_ms":20488,"image_pages":[],"temperature":1.0,"billed_attempts":1,"reasoning_tokens":2951,"transport_provider":"openrouter","cache_read_input_tokens":0,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-11-04T09:21:31.923253+00:00","model_set":{"reader":"deepseek/deepseek-v4-flash"},"pipeline_version":"pith-pipeline@v1.4.1-daily-deepseek-or-medium","input_kind":null,"input_chars":null,"input_total_chars":null,"passes_missing":null,"published_standing":null,"published_confidence":null,"objection_audit":null,"run_record":null,"basis":{"kind":"not_recorded","source":null,"chars_given":null,"chars_total":null},"receipt":null,"falsifier":"Run ARP with the VLM frozen after a single memory write at episode start, or at refresh rates well below 0.5 Hz, on LIBERO-Perturb, and additionally displace the target object twice within one refresh interval; if success collapses once the number or scale of mid-interval scene changes exceeds a threshold that shrinks as refresh rate drops, the memory-sufficiency assumption fails. The gap between the frozen-memory and normal-refresh success rates would quantify how much of ARP's reactivity comes from the memory stream rather than from the expert's own visual servoing.","supporting_citations":[{"cited_title":"Science Robotics , volume =","cited_arxiv_id":null,"evidence_quote":"ROS 2, whose publisher-subscriber model is the stated conceptual basis for ARP's decoupled slow and fast loops."}],"lane_accuracy":{"status":"not yet measured","lane":{"model":"deepseek/deepseek-v4-flash","input_kind":null},"url":"https://pith.science/accuracy"},"review_version":1}