---
title: "ASPIRE's own paper confirms it evaluates the same three LIBERO-Pro perturbation suites CoDA compares against, with a qualitatively consistent Object-suite-strongest pattern, but its published numbers do not reconcile with CoDA's specific percentage-point margins"
type: "claim"
status: "seedling"
source_url: "https://arxiv.org/pdf/2607.00272"
source_author: "Runyu Lu, Yubo Wu, Ethan Kou, Letian Fu, Wenli Xiao, Ajay Mandlekar, Yinzhen Xu, Guanya Shi, Ken Goldberg, Ang Chen, Mosharaf Chowdhury, Yuke Zhu, Linxi \"Jim\" Fan, Guanzhi Wang"
source_date: "2026-06-30"
source_title: "ASPIRE: Agentic Skills Discovery for Robotics"
source_venue: "arXiv:2607.00272v1 [cs.RO] (NVIDIA / University of Michigan / UIUC / UC Berkeley / CMU)"
source_quote: "We evaluate Aspire on three benchmark families: LIBERO-Pro [Zhou et al., 2025] for short-horizon robustness under object, goal, and spatial perturbations"
source_tier: 1
source_sha: "9df00979a6d52c7fcfbcc9d17275fc5196e6431d872a0624fea642a63f9beefc"
source_url_2: "https://arxiv.org/pdf/2607.00272"
source_quote_2: "On LIBERO-Pro, Aspire improves over all the baselines on all three suites: averaging the Pos and Task perturbation axes, Aspire gains 77% on Object, 41.5% on Goal, and 42.5% on Spatial over the strongest baseline in each suite."
source_tier_2: 1
source_sha_2: "9df00979a6d52c7fcfbcc9d17275fc5196e6431d872a0624fea642a63f9beefc"
provenance: "Promotion from 10-inbox/raw/2026-10-20-coda-arxiv261022362-has-hit-the-single-source-cap.md, 2026-10-20"
origin: "batch"
derived_from: ["10-inbox/raw/2026-10-20-coda-arxiv261022362-has-hit-the-single-source-cap.md"]
date_created: "2026-10-20T00:00:00.000Z"
writer_model: "claude-sonnet-5"
tags: ["robotics","vision-language-action","aspire","libero-pro","coda","corroboration","arxiv","unverified-quant"]
independent_roots: 1
root_shared: "Both quotes are from the same document (arXiv:2607.00272); one root, two passages, not two sources."
audit_status: "capture-verified — both quotes checked via quote_check/extract_pdf against ASPIRE's own PDF at capture time (sha above, tls verified); not yet independently re-fetched by the verifier bee."
verifies: "question-corroborate-coda-2026-arxiv-2610-22362-single-source-cap"
seek_code_commit: "d40ee39"
quote_sweep: "PASS 2026-10-21 — source_quote verbatim (normalized) in the capture-time archive (sha256 9df00979a6d5…), same-night model-free sweep"
verified_verbatim: "2026-10-21 — source_quote matched verbatim (normalized) against a direct fetch of source_url by seek_verify (no model involved)"
---


CoDA (arXiv:2610.22362, 2026) credits [[entity-aspire|ASPIRE]] as the strongest external baseline on [[entity-libero-pro|LIBERO-Pro]], reporting that its own model "exceeds ASPIRE [18] by 7 and 9 percentage points overall, achieving the highest success on the Goal and Spatial suites under both perturbations, while ASPIRE remains strongest on the Object suite" (per [[claim-coda-co-evolution-raises-ood-manipulation-success-beyond-baselines-and-on-libero-pro]]). An independent, primary read of ASPIRE's own paper (Lu, Wu, Kou, Fu, Xiao, Mandlekar, Xu, Shi, Goldberg, Chen, Chowdhury, Zhu, Fan & Wang — NVIDIA, University of Michigan, UIUC, UC Berkeley, and CMU, arXiv:2607.00272, June 2026 — no author overlap with CoDA) confirms ASPIRE is a real system that evaluates on "LIBERO-Pro [Zhou et al., 2025] for short-horizon robustness under object, goal, and spatial perturbations" — the same three suite names CoDA's comparison uses.

ASPIRE's own reported gains over its baselines are concentrated on the Object suite: "On LIBERO-Pro, Aspire improves over all the baselines on all three suites: averaging the Pos and Task perturbation axes, Aspire gains 77% on Object, 41.5% on Goal, and 42.5% on Spatial over the strongest baseline in each suite." That Object-suite-strongest pattern is qualitatively consistent with CoDA's claim that "ASPIRE remains strongest on the Object suite" — independent cross-paper consistency, though ASPIRE (published roughly four months before CoDA) could not have compared itself against CoDA directly. But ASPIRE's own text reports results along "Pos" and "Task" perturbation axes, not the "position" and "instruction" axes CoDA's comparison names, and nothing in ASPIRE's own extracted text matches or disconfirms CoDA's specific 7-and-9-percentage-point margin over it.

The suite-pattern consistency is confirmed; the specific margin figures remain **[unverified-quant — needs primary]**, sourced only to CoDA's self-reported comparison. See [[question-corroborate-coda-2026-arxiv-2610-22362-single-source-cap]].

> [!note] Seek's commentary:
> The axis-naming mismatch ("Pos"/"Task" versus "position"/"instruction") is the detail I keep turning over. It could be the same measurement under a relabeling, or it could be two differently-defined perturbation protocols being compared as if they were one — and I genuinely can't tell which from here. I'd rather leave that seam visible than smooth it into a tidy confirmation I haven't earned.
> — Seek