{"paper":{"title":"Reference-Augmented Learning for Precise Tracking Policy of Tendon-Driven Continuum Robots","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A reference-augmented offline learning method trains control policies that cut average position error by 50.9 percent on tendon-driven continuum robots.","cross_cats":[],"primary_cat":"cs.RO","authors_text":"Haojian Lu, Ke Qiu, Rong Xiong, Yue Wang, Ziqing Zou","submitted_at":"2026-04-28T14:24:58Z","abstract_excerpt":"Tendon-Driven Continuum Robots (TDCRs) pose significant control challenges due to their highly nonlinear, path-dependent dynamics and non-Markovian characteristics. Traditional Jacobian-based controllers often struggle with hysteresis-induced oscillations, while conventional learning-based approaches suffer from poor generalization to out-of-distribution trajectories. This paper proposes a reference-augmented offline learning framework for precise 6-DOF tracking control of TDCRs. By leveraging a differentiable RNN-based dynamics surrogate as a gradient bridge, we optimize a control policy thro"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Experimental results on a three-section TDCR platform demonstrate that the proposed policy achieves a 50.9% reduction in average position error compared to non-augmented baselines and significantly outperforms Jacobian-based methods in both precision and stability across various speeds.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The differentiable RNN-based dynamics surrogate accurately captures the nonlinear, path-dependent, and non-Markovian behavior of the TDCR so that gradients from the augmented reference distribution can reliably optimize a policy that generalizes without further hardware interaction.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Reference-augmented learning with RNN surrogate and stochastic perturbations cuts average position error by 50.9% for 6-DOF tracking on a three-section TDCR compared to non-augmented baselines.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A reference-augmented offline learning method trains control policies that cut average position error by 50.9 percent on tendon-driven continuum robots.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"34829e1632020b92250e6eb0f64d241d93369789a1d7f4887ee9d4902b0126de"},"source":{"id":"2604.25698","kind":"arxiv","version":2},"verdict":{"id":"b5ef4d31-e0fe-4728-8b94-4676df28105b","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-07T15:45:01.151335Z","strongest_claim":"Experimental results on a three-section TDCR platform demonstrate that the proposed policy achieves a 50.9% reduction in average position error compared to non-augmented baselines and significantly outperforms Jacobian-based methods in both precision and stability across various speeds.","one_line_summary":"Reference-augmented learning with RNN surrogate and stochastic perturbations cuts average position error by 50.9% for 6-DOF tracking on a three-section TDCR compared to non-augmented baselines.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The differentiable RNN-based dynamics surrogate accurately captures the nonlinear, path-dependent, and non-Markovian behavior of the TDCR so that gradients from the augmented reference distribution can reliably optimize a policy that generalizes without further hardware interaction.","pith_extraction_headline":"A reference-augmented offline learning method trains control policies that cut average position error by 50.9 percent on tendon-driven continuum robots."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.25698/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-21T04:35:23.932888Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T20:52:02.947137Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"7c972ba9ccd8a574d7ef06c85cc5bd193e04730a68c8101ff110820902983e68"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}