{"paper":{"title":"Learning Dynamic Rope Manipulation Using Task-Level Iterative Learning Control","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Task-level iterative learning control lets robots master dynamic flying knots from one human demonstration and a simplified rope model.","cross_cats":[],"primary_cat":"cs.RO","authors_text":"Chris Atkeson, Krishna Suresh","submitted_at":"2026-02-24T19:13:25Z","abstract_excerpt":"We introduce a Task-Level Iterative Learning Control method for dynamic manipulation of ropes. We demonstrate this method on a non-planar rope manipulation task called the flying knot. Using a single human demonstration and a simplified rope model, the method learns directly on hardware without reliance on large amounts of demonstration data or massive amounts of simulation. At each iteration, the algorithm inverts a model of the robot and rope by solving a quadratic program to propagate task-space errors into action updates. We evaluate performance across 7 different kinds of ropes, including"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Using a single human demonstration and a simplified rope model, the method learns directly on hardware without reliance on large amounts of demonstration data or massive amounts of simulation. Learning achieves a 100% success rate within 10 trials on all ropes. Furthermore, the method can successfully transfer between most rope types in 2--5 trials.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The simplified rope model is sufficient to allow quadratic program inversion to accurately propagate task-space errors into effective action updates for the dynamic flying knot task.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Task-level ILC learns flying knot rope manipulation from one demo, achieving 100% success within 10 trials on 7 rope types with 2-5 trial transfers.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Task-level iterative learning control lets robots master dynamic flying knots from one human demonstration and a simplified rope model.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"8ef27b1eb34055f61f337555bfedd9a46f8f83863909c0bdc7acc444161c1801"},"source":{"id":"2602.21302","kind":"arxiv","version":2},"verdict":{"id":"87f6c342-73f1-437d-b4f4-ac6c90e479f5","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T19:41:16.455300Z","strongest_claim":"Using a single human demonstration and a simplified rope model, the method learns directly on hardware without reliance on large amounts of demonstration data or massive amounts of simulation. Learning achieves a 100% success rate within 10 trials on all ropes. Furthermore, the method can successfully transfer between most rope types in 2--5 trials.","one_line_summary":"Task-level ILC learns flying knot rope manipulation from one demo, achieving 100% success within 10 trials on 7 rope types with 2-5 trial transfers.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The simplified rope model is sufficient to allow quadratic program inversion to accurately propagate task-space errors into effective action updates for the dynamic flying knot task.","pith_extraction_headline":"Task-level iterative learning control lets robots master dynamic flying knots from one human demonstration and a simplified rope model."},"references":{"count":51,"sample":[{"doi":"","year":2006,"title":"Using inaccurate models in reinforcement learning","work_id":"218aa5d7-38f3-4e8d-a9bb-1ff705c31c6e","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.1109/robot","year":1988,"title":"The enhanced compact qp method for redundant manipulators using practical inequality constraints","work_id":"d203a59f-a6de-4624-8a99-b6b970c2bd77","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"A regret minimization approach to iterative learning control","work_id":"fff1819c-3545-40fa-9722-528824638926","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"URL https://proceedings.mlr.press/v139/ agarwal21b.html","work_id":"f17c1ef8-a029-4acb-a153-19c1e93eeac7","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Chae H. An, Christopher G. Atkeson, and John Holler- bach.Model-Based Control of a Robot Manipulator. Artificial Intelligence Series. MIT Press. ISBN 978-0- 262-51157-5","work_id":"762f56f6-92df-4011-a334-af66a1aa8a7e","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":51,"snapshot_sha256":"b570041887e374b3e3550647299eeb8c3cb9ba75904d9c27a4c7f4b3247cb144","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"0c73b4abab8cd1f17cfb7335c926ff8e35dbf29f1d4b35516d6540020d5d2fea"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}