Pith Number

pith:BF4S6654

pith:2025:BF4S6654AXYNQRTABTOED35P75

not attested not anchored not stored refs pending

Affordance-R1: Reinforcement Learning for Generalizable Affordance Reasoning in Multimodal Large Language Model

Hanqing Wang, Jiahao Yuan, Jiamin Wang, Mingyu Liu, Shaoyang Wang, Yifan Han, Yiming Zhong, Yuexin Ma, Zemin Yang, Zhiqing Cui

Reinforcement learning via GRPO with a custom affordance reward function produces zero-shot generalization and emergent test-time reasoning in multimodal models for robot affordance grounding.

arxiv:2508.06206 v5 · 2025-08-08 · cs.RO · cs.CV

Open paper page JSON Open Graph Bundle Merged state Verified badge What is a Pith Number?

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{BF4S6654AXYNQRTABTOED35P75}

Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge

Record completeness

1 Bitcoin timestamp

2 Internet Archive

3 Author claim open · sign in to claim

4 Citations open

5 Replications open

✓ Portable graph bundle live · download bundle · merged state

The bundle contains the canonical record plus signed events. A mirror can host it anywhere and recompute the same current state with the deterministic merge algorithm.

Claims

C1strongest claim

Trained exclusively via reinforcement learning with GRPO and without explicit reasoning data, Affordance-R1 achieves robust zero-shot generalization and exhibits emergent test-time reasoning capabilities.

C2weakest assumption

The custom affordance function containing format, perception, and cognition rewards will steer the GRPO optimization toward generalizable cognitive reasoning rather than overfitting to training distributions or reward specifics, as implied by the claim of emergent capabilities from RL-only training.

C3one line summary

Affordance-R1 applies GRPO-based reinforcement learning to multimodal LLMs for affordance grounding, using format-perception-cognition rewards and the ReasonAff dataset to achieve zero-shot generalization and emergent reasoning.

Formal links

2 machine-checked theorem links

Cited by

3 papers in Pith

OneThinker: All-in-one Reasoning Model for Image and Video

Affordance Agent Harness: Verification-Gated Skill Orchestration

Receipt and verification

First computed	2026-05-21T01:04:15.766254Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

09792f7bbc05f0d846600cdc41efafff6834f3c836aa21e98cfff82244b8febf

Aliases

arxiv: 2508.06206 · arxiv_version: 2508.06206v5 · doi: 10.48550/arxiv.2508.06206 · pith_short_12: BF4S6654AXYN · pith_short_16: BF4S6654AXYNQRTA · pith_short_8: BF4S6654

Agent API

Resolver JSON Graph JSON Events JSON Schema Signing key

Verify this Pith Number yourself

curl -sH 'Accept: application/ld+json' https://pith.science/pith/BF4S6654AXYNQRTABTOED35P75 \
  | jq -c '.canonical_record' \
  | python3 -c "import sys,json,hashlib; b=json.dumps(json.loads(sys.stdin.read()), sort_keys=True, separators=(',',':'), ensure_ascii=False).encode(); print(hashlib.sha256(b).hexdigest())"
# expect: 09792f7bbc05f0d846600cdc41efafff6834f3c836aa21e98cfff82244b8febf

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "f0fd336fe3d9bc8059b4480ebf62db307eacc3c5e3376a6b7d58645d1a2d4fa6",
    "cross_cats_sorted": [
      "cs.CV"
    ],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.RO",
    "submitted_at": "2025-08-08T10:39:04Z",
    "title_canon_sha256": "0d961a47fe42356b6afc94fe5f8ad41cd3877f2dd83f22bd7a975b64ddcfe849"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2508.06206",
    "kind": "arxiv",
    "version": 5
  }
}