Pith Number

pith:PNBRIR36

pith:2026:PNBRIR36RHL6GA5OI6FANIB25G

not attested not anchored not stored refs resolved

When Backdoors Meet Partial Observability: Attacking Real-World Reinforcement Learning

Haibo Hu, Jiawei Lian, Qingqing Ye, Tairan Huang, Yaxin Xiao, Yi Wang, Yulin Jin

A diffusion model learns visual triggers that activate backdoors in real robot RL policies despite varying uncontrollable sensors.

arxiv:2601.14104 v2 · 2026-01-20 · 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{PNBRIR36RHL6GA5OI6FANIB25G}

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

Experiments on a physical TurtleBot3 platform show that DGBA consistently outperforms prior RL backdoor attacks while preserving normal task performance.

C2weakest assumption

That a conditional diffusion model can learn a stochastic trigger distribution maintaining consistent attack activation across varying uncontrollable auxiliary states such as LiDAR and odometry signals.

C3one line summary

DGBA enables reliable backdoor attacks on real-world RL policies under partial observability by learning stochastic visual triggers via conditional diffusion and using advantage-based poisoning at critical states.

References

30 extracted · 30 resolved · 3 Pith anchors

[1] Turtlebot 3 as a robotics education platform 2019

[2] Stepping locomotion for a walking excavator robot using hierarchical reinforcement learning and action masking 2025

[3] Multi-agent inverse reinforcement learning in real world unstructured pedestrian crowds 2025

[4] Dames, and Mac Schwager 2025

[5] Choi, Fernando Casta˜neda, Won- suhk Jung, Bike Zhang, Claire J 2025

Receipt and verification

First computed	2026-05-18T03:10:11.297322Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

7b4314477e89d7e303ae478a06a03ae9abde9c3890d0549d21f781616c2d8855

Aliases

arxiv: 2601.14104 · arxiv_version: 2601.14104v2 · doi: 10.48550/arxiv.2601.14104 · pith_short_12: PNBRIR36RHL6 · pith_short_16: PNBRIR36RHL6GA5O · pith_short_8: PNBRIR36

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/PNBRIR36RHL6GA5OI6FANIB25G \
  | 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: 7b4314477e89d7e303ae478a06a03ae9abde9c3890d0549d21f781616c2d8855

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "4d9470df52c6b4a960ff7def8fd624362b528cf40e11905e2892606ce5715442",
    "cross_cats_sorted": [
      "cs.CV"
    ],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.RO",
    "submitted_at": "2026-01-20T16:03:51Z",
    "title_canon_sha256": "0954282a16ce738b881151c3bbe554046ed6c1689ad21950f917947bfe704bc3"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2601.14104",
    "kind": "arxiv",
    "version": 2
  }
}