History / Review Cross Embodiment

Add Review-Single-Checkpoint-Multi-Robot: deploy-side cross-embodiment (one frozen checkpoint, many robots) New page separating the DEPLOYMENT question (one unchanged checkpoint controls multiple physical robots at inference) from the training-data cut in Review-Cross-Embodiment. Organized by three tiers: - Tier 2 (zero-shot to unseen robot): LAP-3B (actions-as-language, first substantial zero-shot to unseen), Green-VLA, Gemini Robotics 1.5, DreamZero/DYNA-2 (WAM), Contact-Anchored Policies, One-Hand. - Tier 1 (routed seen-robot generalist): RT-X, CrossFormer, RDT-1B, UniAct, GR00T N1, pi0.5->pi0.7, Motus. - Tier 3 (contrast, needs per-robot fit): Octo, HPT, X-VLA; plus MergeVLA (merge specialists into one checkpoint). Includes a routing-mechanism taxonomy, honest limits (AnyBody, unreplicated 2026 zero-shot claims, 'single checkpoint != nothing per robot'), and a design guide. Cross-linked from Review-Cross-Embodiment, Reviews, Home. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Heungwoo committed Aug 12, 2026

Escape pipe in all in-table wikilinks wiki-wide (202 links, 17 files) GitHub-wiki table cells read a wikilink's separator | as a column delimiter, splitting the cell and breaking the link. Escape to \| in every table-row wikilink (Home nav, RSS-2026-Papers, topic surveys). Prose wikilinks left as plain | (render correctly outside tables). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Heungwoo committed Aug 11, 2026

Weave ICML 2026 evidence into deep-dive surveys; revise three verdicts 14 State-of-the-Field sections gain ICML 2026 findings from the 99-paper index: recipes and latent-action supervision (VLANeXt, From-Pixels-to-Tokens, XR-1), MoT dual-systems and shortcut counters, the 9-paper efficiency cluster (Reflex 50Hz, GridS -76% FLOPs, XPU profile, latent reasoning -90%), reward/critic and model-based RL (VLAC, VLAW +39.2%), memory (HiMe/SOMA/CAPS), world models (DreamDojo 44kh, LAC-WM, dWorldEval), dexterous (DexMachina/DECO/Tabero/CTSRL), cross-embodiment (OXE-AugE, latent motion codes), evaluation (LIBERO-Gen, VLA-Arena, FixBench, TRAP). Verdicts revised: forgetting milder than assumed; discrete-token verdict scoped to robot-action auxiliaries; WM-evaluator action gap first crack. Home synced. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Heungwoo committed Aug 5, 2026

Decision map: prominent deep-dive links + uniform detail-page template Home fold-outs now lead with a heading-level "Deep dive ->" link and compress trend/approaches/limitations into a labeled 3-row table. The 12 detail-page State-of-the-Field sections are rewritten to one template (Verdict quote + Trend + Approaches-and-trade-offs + optional Established-findings + Limitations, dated Aug 2026); the three standalone surveys get matching headers with structure legends. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Heungwoo committed Aug 5, 2026

Survey-depth topic pages: 12 State-of-the-Field updates + 3 new surveys Each decision-map topic's detail page now carries a dated July-2026 survey section: trend arc through the latest venues, approach taxonomy with definitions and trade-offs, and current limitations. Three previously page-less topics get dedicated surveys: Human-Video Transfer (emergence/decoupling/synthesis fork + decision guide), VLA Evaluation (indictment + 2026 toolkit + emerging norms), Real-Time Execution (RTC->Legato arc + approach comparison). Home fold-outs link the full surveys; Reviews catalog updated. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Heungwoo committed Jul 27, 2026

Update review theses where ICRA 2026 shifts the message (not just content) - Cross-Embodiment Q1: fold in Galaxea G0 counter-evidence — single-embodiment data consistency is a third lever alongside scale-for-coverage / architecture- for-extrapolation (raw cross-embodiment scale is not automatically better) - VLM-Action trends: add trend #7 — by ICRA 2026 the mechanism set has stabilized (no 8th coupling primitive); frontier moved from inventing wirings to deploying frozen ones (FD-VLA, G0, FPO) - System-0/1/2 TL;DR: dual-system is now the label-free systems-community default (Galaxea G0, DualVLN) Other reviews' theses are reinforced (not changed) by ICRA — left as-is. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Heungwoo committed Jun 1, 2026

Integrate ICRA 2026 into cross-paper in-depth reviews Wove ICRA 2026 developments into 9 cross-paper reviews (drawing only from the already-verified ICRA topic/per-paper pages — no new web claims), mapping ICRA papers onto each review's existing taxonomy: - VLA-Architecture, Dexterous-Manipulation, RL, VLA-Memory, Goal-Image-Conditioning, System-0-1-2, Cross-Embodiment, VLM-Action-Connection, WAM-vs-VLA-Robustness Six of these had zero ICRA content before. 0 dangling wikilinks. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Heungwoo committed Jun 1, 2026

Source-verification audit: fix fabrications, rename PI slugs, link hygiene, add foundational pages - Re-verified all 195 pages vs original sources; removed 47 confirmed fabricated tables/numbers, restored 19 false-positive deletions (full-PDF re-check) - Renamed PI tech-report slugs ICLR-2026-pi07/pi06/RECAP -> PI-pi07/PI-pi06/PI-RECAP (these are PI technical reports, not ICLR 2026 papers); updated 171 wikilinks - Fixed 60 broken wikilinks -> 0 dangling across the wiki - Added foundational pages: OpenVLA, ReKep, AgiBot World Colosseo, RoboBrain 2.0; linked from Home + sidebar + venue indexes Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Heungwoo committed Jun 1, 2026

Integrate NeurIPS 2025 papers into in-depth review pages Review-VLA-Architecture.md: - Add 13 NeurIPS 2025 papers to §3 paper table: Knowledge Insulation (Spotlight), ChatVLA-2, DreamVLA, VLA-OS, Fast-in-Slow, ThinkAct, Chain-of-Action, Real-Time Chunking, ReinFlow, VideoVLA, CogVLA, BridgeVLA, DynaGuide. - Extend category exemplar lists: · B (Flow matching): Knowledge Insulation + RTC + ReinFlow · E (World-model / VAM): DreamVLA + VideoVLA + SAMPO + OSVI-WM + RLVR-World · F (Hierarchical / dual-system / MoE): Fast-in-Slow, ChatVLA-2, ThinkAct, VLA-OS · G (Reasoning / CoT): Chain-of-Action, ThinkAct, Robot-R1, DreamVLA · I (Small / efficient): CogVLA · J (Sensory-augmented): BridgeVLA (3D → 2D heatmap I/O) - Update Trends 2 and 3 to highlight NeurIPS 2025 as the conference that cemented dual-system (Trend 2) and deepened world-model-in-VLA (Trend 3). RL.md: - Update header: now covers CoRL 2025 + NeurIPS 2025 + ICLR 2026. - New §3a "Making flow-matching policies RL-trainable (log-prob-exact)": ReinFlow + DSRL ancestor. - Extend §7 (reasoning-token RL) with ThinkAct + Robot-R1. - New §9 (Empirical studies & human-preference RL): What-Can-RL-Bring (PPO > DPO/GRPO empirical) + APO (binary-signal RLHF from HRI). - Update reading order: insert What-Can-RL-Bring as step 2, ReinFlow as step 5, ThinkAct as step 6. Review-Cross-Embodiment.md: - Add Grasp2Grasp (NeurIPS 2025, Princeton) to §3 paper table. - Extend Category C (embodiment-invariant latents) with Grasp2Grasp as a grasp-level Schrödinger-Bridge exemplar. All three reviews now reflect that NeurIPS 2025 sits between CoRL 2025 and ICLR 2026 in the lineage — not just chronologically but substantively, contributing key published formalisms (KI, RTC, ReinFlow) and algorithm comparisons (What-Can-RL-Bring) that ICLR 2026 builds on. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Heungwoo committed Apr 17, 2026

Add in-depth cross-paper review of cross-embodiment VLA training Covers 27 papers across ICLR 2026, CoRL 2025, and the π series, plus key 2024-2026 precursors (OXE/RT-X, Octo, CrossFormer, HPT, RDT-1B, UniAct, GR00T N1, Gemini Robotics 1.5) and the AnyBody benchmark. Page structure: - Why cross-embodiment matters; π0.7 UR5e zero-shot laundry (85.6% progress, 80% SR) matching expert teleop, vs AnyBody showing novel- morphology extrapolation still fails - Seven-category taxonomy: A. Unified tokenized action space (RT-X, Octo, RDT-1B, FAST) B. Soft-prompt / body-conditioning (X-VLA, HPT) C. Embodiment-invariant latents (UniVLA, UniAct, X-Sim, UniSkill, TraceVLA, XR-1 UVMC) D. Human video / wearable bridging (DexUMI, EgoDex, Visual-Imit- Humanoid, ImMimic) E. Morphology-aware architecture (CrossFormer, HPT stems, GR00T, WholeBodyVLA) F. Scale + prompt expansion (pi0.5 -> pi0.6 -> pi0.7, Gemini Robotics 1.5) G. World-model-mediated (DreamGen, Cosmos Policy, Ctrl-World) - Per-paper deep-dives with arXiv links, mechanisms, pros/cons - Cross-axis comparison (cheapest new-robot extension, most diverse morphology span, most data-efficient, strongest zero-shot result) - Trends: 2023 no-sharing -> 2024 Group A dominance -> 2025 Groups C and D bloom -> late 2025/2026 Group F crowns production -> ICLR 2026 synthesizes B+C+G - Open questions: does scale alone solve it; sim-to-real vs real-to- sim; universal action space; strategy transfer; benchmark maturity; open vs closed ecosystem - Practical decision guide for shipping cross-embodiment VLAs Cross-links added in X-VLA and X-Sim summary pages; sidebar and Home updated to surface the new review. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Heungwoo committed Apr 17, 2026