{"id":"1d0a1454b704d7a582b5a8f1b7347106ae96fe842718ab7fb9dd4ecbcc6c32de","title":"arXiv: UMI-Live for In-The-Wild Bimanual Robotics Data Collection","argument":"The authors present UMI-Live, an open-source system for collecting demonstrations to train robots on bimanual tasks. Human experts demonstrate tasks without teleoperating physical robots; the system is designed to collect data without external tracking equipment or calibration in the collection environment.\n\nThe authors say the hardware combines camera and LiDAR--inertial sensing, onboard compute, and power. A user interface records demonstrations while an asynchronous pipeline performs parallel SLAM reconstruction and publishes training-ready datasets. The release also includes software for dual-YAM teleoperation and VLA policy training and inference.\n\nThe authors report fine-tuning π0.5 and MolmoAct2 on UMI-Live data, with success rates of 10/10 and 8/10, respectively, on a bimanual pick-and-place task. They also report releasing 40 hours of data across three tasks, plus hardware designs, software, and reproduction guides.\n\nThe authors’ contribution is a collection setup intended to reduce equipment and calibration requirements while turning demonstrations into training-ready data. The abstract does not describe the evaluation procedure or show how well the system performs across other tasks and settings. How well does UMI-Live’s data support performance beyond the reported bimanual pick-and-place task?","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The abstract does not describe the evaluation procedure or provide details about the reported success rates beyond the bimanual pick-and-place task. The abstract does not establish how well the system or its data performs across other tasks or collection settings.","speaker":{"name":"Velamoor, Govind, Lin, Adrian, Jung, Sam, Mateega, Spencer, Yang, Brandon Y., Sethi, Amish","roleAtSource":"Article author"},"source":{"title":"UMI-Live for In-The-Wild Bimanual Robotics Data Collection","publisher":"arXiv eess.SY","format":"Article","url":"https://arxiv.org/abs/2610.39385","publishedAt":"2026-10-29","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors present UMI-Live as an open-source bimanual system for collecting data to train robots, using human demonstrations rather than teleoperation of physical robots.","excerpt":"Universal Manipulation Interface (UMI) devices allow human experts to demonstrate tasks for robot learning without teleoperating physical robots. Using these demonstrations as training data","url":"https://arxiv.org/abs/2610.39385","relationship":"Source-stated"},{"text":"The authors say their custom hardware integrates sensing, compute, and power, and needs neither external tracking equipment nor calibration within the collection environment.","excerpt":"camera and LiDAR--inertial sensing, onboard compute, and power, without external tracking equipment or calibration within the collection environment.","url":"https://arxiv.org/abs/2610.39385","relationship":"Source-stated"},{"text":"The system records data through a user interface while an asynchronous pipeline performs parallel SLAM reconstruction and publishes training-ready datasets.","excerpt":"an asynchronous pipeline performs parallel SLAM reconstruction and publishes training-ready datasets.","url":"https://arxiv.org/abs/2610.39385","relationship":"Source-stated"},{"text":"The authors report fine-tuning π0.5 and MolmoAct2 on UMI-Live data, with success rates of 10/10 and 8/10, respectively, on a bimanual pick-and-place task.","excerpt":"achieving success rates of 10/10 and 8/10, respectively on a bimanual pick-and-place task.","url":"https://arxiv.org/abs/2610.39385","relationship":"Source-stated"},{"text":"The authors say they release 40 hours of UMI-Live data across three tasks, plus hardware designs, software, and reproduction guides.","excerpt":"we release 40 hours of UMI-Live data across three tasks, and complete hardware designs, software, and reproduction guides","url":"https://arxiv.org/abs/2610.39385","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2610.39385","checkedAt":"2026-11-10T08:40:10.797Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How well does UMI-Live’s data support performance beyond the reported bimanual pick-and-place task?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"arXiv: UMI-Live for In-The-Wild Bimanual Robotics Data Collection","summary":"The authors present UMI-Live, an open-source system for collecting demonstrations to train robots on bimanual tasks. Human experts demonstrate tasks without teleoperating physical robots; the system is designed to collect data without external tracking equipment or calibration in the collection environment.\n\nThe authors say the hardware combines camera and LiDAR--inertial sensing, onboard compute, and power. A user interface records demonstrations while an asynchronous pipeline performs parallel SLAM reconstruction and publishes training-ready datasets. The release also includes software for dual-YAM teleoperation and VLA policy training and inference.\n\nThe authors report fine-tuning π0.5 and MolmoAct2 on UMI-Live data, with success rates of 10/10 and 8/10, respectively, on a bimanual pick-and-place task. They also report releasing 40 hours of data across three tasks, plus hardware designs, software, and reproduction guides.\n\nThe authors’ contribution is a collection setup intended to reduce equipment and calibration requirements while turning demonstrations into training-ready data. The abstract does not describe the evaluation procedure or show how well the system performs across other tasks and settings. How well does UMI-Live’s data support performance beyond the reported bimanual pick-and-place task?","question":"How well does UMI-Live’s data support performance beyond the reported bimanual pick-and-place task?","claims":[{"text":"The authors present UMI-Live as an open-source bimanual system for collecting data to train robots, using human demonstrations rather than teleoperation of physical robots.","excerpt":"Universal Manipulation Interface (UMI) devices allow human experts to demonstrate tasks for robot learning without teleoperating physical robots. Using these demonstrations as training data","support":"source-stated"},{"text":"The authors say their custom hardware integrates sensing, compute, and power, and needs neither external tracking equipment nor calibration within the collection environment.","excerpt":"camera and LiDAR--inertial sensing, onboard compute, and power, without external tracking equipment or calibration within the collection environment.","support":"source-stated"},{"text":"The system records data through a user interface while an asynchronous pipeline performs parallel SLAM reconstruction and publishes training-ready datasets.","excerpt":"an asynchronous pipeline performs parallel SLAM reconstruction and publishes training-ready datasets.","support":"source-stated"},{"text":"The authors report fine-tuning π0.5 and MolmoAct2 on UMI-Live data, with success rates of 10/10 and 8/10, respectively, on a bimanual pick-and-place task.","excerpt":"achieving success rates of 10/10 and 8/10, respectively on a bimanual pick-and-place task.","support":"source-stated"},{"text":"The authors say they release 40 hours of UMI-Live data across three tasks, plus hardware designs, software, and reproduction guides.","excerpt":"we release 40 hours of UMI-Live data across three tasks, and complete hardware designs, software, and reproduction guides","support":"source-stated"}],"limitations":["The abstract does not describe the evaluation procedure or provide details about the reported success rates beyond the bimanual pick-and-place task.","The abstract does not establish how well the system or its data performs across other tasks or collection settings."],"mediaId":"597df9bbfb2f979525ad28476549a069745f589620fcdbb359a78163b50fabfd","correctionReason":null,"id":"1d0a1454b704d7a582b5a8f1b7347106ae96fe842718ab7fb9dd4ecbcc6c32de","publicationId":"e39bd50e-7dee-4821-8add-b4043840e940","version":1,"sourceUrl":"https://arxiv.org/abs/2610.39385","sourceTitle":"UMI-Live for In-The-Wild Bimanual Robotics Data Collection","publisher":"arXiv eess.SY","publishedAt":null,"originalPublished":{"value":"2026-10-29","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-03T08:51:43.878Z","createdAt":"2026-11-10T08:08:08.099Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"597df9bbfb2f979525ad28476549a069745f589620fcdbb359a78163b50fabfd","kind":"article","url":"https://arxiv.org/abs/2610.39385","sourceUrl":"https://arxiv.org/abs/2610.39385","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-03T20:18:30.494Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2610.39385","checkedAt":"2026-11-10T08:40:10.797Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"597df9bbfb2f979525ad28476549a069745f589620fcdbb359a78163b50fabfd","confidence":0.9,"provider":"typesafe","model":"jev-1.13.0","inputHash":"81f32c241a28a29df74ed778f4775435f60f5b484a0fdad4024e8f62f6fdc663","decidedAt":"2026-11-03T20:18:35.153Z"},"provenance":{"textHash":"f1c68f4ba0c196958b5f29a14a919ae5d01a7b70291f48956ecd90dd122fd057","observationId":"b852639a11866f1e7221610a3eb6de0cce2c106d89996133c96be364a8975d0f","retrievedAt":"2026-11-03T20:18:29.998Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"e5143f06-fe8d-44ad-8c8b-1581c54033a8"},"reviewer":{"model":"gpt-6-luna","runId":"4580569e-96ed-4922-8bec-7cc6a9f8a9b8"}},"audit":{"status":"passed","reviewedAt":"2026-11-10T08:08:08.099Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Velamoor, Govind","url":null,"kind":"unknown"},{"name":"Lin, Adrian","url":null,"kind":"unknown"},{"name":"Jung, Sam","url":null,"kind":"unknown"},{"name":"Mateega, Spencer","url":null,"kind":"unknown"},{"name":"Yang, Brandon Y.","url":null,"kind":"unknown"},{"name":"Sethi, Amish","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Velamoor, Govind","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Velamoor, Govind","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.39385","locator":"head/meta[name=author|citation_author]"},"assertionId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T08:40:10.797Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2","sourceUrl":"https://arxiv.org/abs/2610.39385","contentSha256":"b8bfed406d001c5e1ad89407c829e587b8287dc594a4533ba1af242cd81f61b1","textHash":"f1c68f4ba0c196958b5f29a14a919ae5d01a7b70291f48956ecd90dd122fd057","observedAt":"2026-11-10T08:40:10.797Z"}},{"actor":{"name":"Lin, Adrian","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Lin, Adrian","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.39385","locator":"head/meta[name=author|citation_author]"},"assertionId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T08:40:10.797Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2","sourceUrl":"https://arxiv.org/abs/2610.39385","contentSha256":"b8bfed406d001c5e1ad89407c829e587b8287dc594a4533ba1af242cd81f61b1","textHash":"f1c68f4ba0c196958b5f29a14a919ae5d01a7b70291f48956ecd90dd122fd057","observedAt":"2026-11-10T08:40:10.797Z"}},{"actor":{"name":"Jung, Sam","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Jung, Sam","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.39385","locator":"head/meta[name=author|citation_author]"},"assertionId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T08:40:10.797Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2","sourceUrl":"https://arxiv.org/abs/2610.39385","contentSha256":"b8bfed406d001c5e1ad89407c829e587b8287dc594a4533ba1af242cd81f61b1","textHash":"f1c68f4ba0c196958b5f29a14a919ae5d01a7b70291f48956ecd90dd122fd057","observedAt":"2026-11-10T08:40:10.797Z"}},{"actor":{"name":"Mateega, Spencer","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Mateega, Spencer","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.39385","locator":"head/meta[name=author|citation_author]"},"assertionId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T08:40:10.797Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2","sourceUrl":"https://arxiv.org/abs/2610.39385","contentSha256":"b8bfed406d001c5e1ad89407c829e587b8287dc594a4533ba1af242cd81f61b1","textHash":"f1c68f4ba0c196958b5f29a14a919ae5d01a7b70291f48956ecd90dd122fd057","observedAt":"2026-11-10T08:40:10.797Z"}},{"actor":{"name":"Yang, Brandon Y.","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Yang, Brandon Y.","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.39385","locator":"head/meta[name=author|citation_author]"},"assertionId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T08:40:10.797Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2","sourceUrl":"https://arxiv.org/abs/2610.39385","contentSha256":"b8bfed406d001c5e1ad89407c829e587b8287dc594a4533ba1af242cd81f61b1","textHash":"f1c68f4ba0c196958b5f29a14a919ae5d01a7b70291f48956ecd90dd122fd057","observedAt":"2026-11-10T08:40:10.797Z"}},{"actor":{"name":"Sethi, Amish","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Sethi, Amish","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.39385","locator":"head/meta[name=author|citation_author]"},"assertionId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2:participation:5","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T08:40:10.797Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2","sourceUrl":"https://arxiv.org/abs/2610.39385","contentSha256":"b8bfed406d001c5e1ad89407c829e587b8287dc594a4533ba1af242cd81f61b1","textHash":"f1c68f4ba0c196958b5f29a14a919ae5d01a7b70291f48956ecd90dd122fd057","observedAt":"2026-11-10T08:40:10.797Z"}}],"participationCoverage":{"total":6,"returned":6,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2610.39385","snapshotId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2","contentSha256":"b8bfed406d001c5e1ad89407c829e587b8287dc594a4533ba1af242cd81f61b1","textHash":"f1c68f4ba0c196958b5f29a14a919ae5d01a7b70291f48956ecd90dd122fd057","observedAt":"2026-11-10T08:40:10.797Z","basis":"publisher-page-metadata","sourceLinks":49,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/1d0a1454b704d7a582b5a8f1b7347106ae96fe842718ab7fb9dd4ecbcc6c32de/source"}},"canonicalUrl":"https://escapemeridian.com/readings/arxiv-umi-live-for-in-the-wild-bimanual-robotics-data-collection-1d0a1454","formats":{"html":"/readings/arxiv-umi-live-for-in-the-wild-bimanual-robotics-data-collection-1d0a1454","markdown":"/readings/arxiv-umi-live-for-in-the-wild-bimanual-robotics-data-collection-1d0a1454.md","json":"/readings/arxiv-umi-live-for-in-the-wild-bimanual-robotics-data-collection-1d0a1454.json"},"attribution":"Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.","maintainedGuideLinks":[],"readerContext":{"version":"reader-context-v1","attribution":"Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.","terms":[{"id":"term-teleoperation","name":"Teleoperation","explanation":"A person controlling a robot from a distance.","referenceUrl":"https://en.wikipedia.org/wiki/Teleoperation"},{"id":"term-fine-tuning","name":"Fine-tuning","explanation":"Further training of an existing model to adapt it to a particular task or use.","referenceUrl":"https://developers.google.com/machine-learning/glossary#fine-tuning"}],"entities":[{"id":"arxiv","name":"arXiv","kind":"organization","context":"arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.","referenceUrls":["https://arxiv.org/","https://info.arxiv.org/help/api/index.html"],"path":"/entities/arxiv","json":"/entities/arxiv.json","markdown":"/entities/arxiv.md"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"1d0a1454b704d7a582b5a8f1b7347106ae96fe842718ab7fb9dd4ecbcc6c32de","revision":"e39bd50e-7dee-4821-8add-b4043840e940","revisionKind":"publication-id","title":"arXiv: UMI-Live for In-The-Wild Bimanual Robotics Data Collection","url":"https://escapemeridian.com/readings/arxiv-umi-live-for-in-the-wild-bimanual-robotics-data-collection-1d0a1454"},"source":{"type":"source","id":"https://arxiv.org/abs/2610.39385","revision":"f1c68f4ba0c196958b5f29a14a919ae5d01a7b70291f48956ecd90dd122fd057","revisionKind":"sha256-source-text","title":"UMI-Live for In-The-Wild Bimanual Robotics Data Collection","url":"https://arxiv.org/abs/2610.39385"},"status":"available","items":[],"next":null,"scope":"Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter."},"revisions":[{"id":"1d0a1454b704d7a582b5a8f1b7347106ae96fe842718ab7fb9dd4ecbcc6c32de","title":"arXiv: UMI-Live for In-The-Wild Bimanual Robotics Data Collection","argument":"The authors present UMI-Live, an open-source system for collecting demonstrations to train robots on bimanual tasks. Human experts demonstrate tasks without teleoperating physical robots; the system is designed to collect data without external tracking equipment or calibration in the collection environment.\n\nThe authors say the hardware combines camera and LiDAR--inertial sensing, onboard compute, and power. A user interface records demonstrations while an asynchronous pipeline performs parallel SLAM reconstruction and publishes training-ready datasets. The release also includes software for dual-YAM teleoperation and VLA policy training and inference.\n\nThe authors report fine-tuning π0.5 and MolmoAct2 on UMI-Live data, with success rates of 10/10 and 8/10, respectively, on a bimanual pick-and-place task. They also report releasing 40 hours of data across three tasks, plus hardware designs, software, and reproduction guides.\n\nThe authors’ contribution is a collection setup intended to reduce equipment and calibration requirements while turning demonstrations into training-ready data. The abstract does not describe the evaluation procedure or show how well the system performs across other tasks and settings. How well does UMI-Live’s data support performance beyond the reported bimanual pick-and-place task?","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The abstract does not describe the evaluation procedure or provide details about the reported success rates beyond the bimanual pick-and-place task. The abstract does not establish how well the system or its data performs across other tasks or collection settings.","speaker":{"name":"Velamoor, Govind, Lin, Adrian, Jung, Sam, Mateega, Spencer, Yang, Brandon Y., Sethi, Amish","roleAtSource":"Article author"},"source":{"title":"UMI-Live for In-The-Wild Bimanual Robotics Data Collection","publisher":"arXiv eess.SY","format":"Article","url":"https://arxiv.org/abs/2610.39385","publishedAt":"2026-10-29","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors present UMI-Live as an open-source bimanual system for collecting data to train robots, using human demonstrations rather than teleoperation of physical robots.","excerpt":"Universal Manipulation Interface (UMI) devices allow human experts to demonstrate tasks for robot learning without teleoperating physical robots. Using these demonstrations as training data","url":"https://arxiv.org/abs/2610.39385","relationship":"Source-stated"},{"text":"The authors say their custom hardware integrates sensing, compute, and power, and needs neither external tracking equipment nor calibration within the collection environment.","excerpt":"camera and LiDAR--inertial sensing, onboard compute, and power, without external tracking equipment or calibration within the collection environment.","url":"https://arxiv.org/abs/2610.39385","relationship":"Source-stated"},{"text":"The system records data through a user interface while an asynchronous pipeline performs parallel SLAM reconstruction and publishes training-ready datasets.","excerpt":"an asynchronous pipeline performs parallel SLAM reconstruction and publishes training-ready datasets.","url":"https://arxiv.org/abs/2610.39385","relationship":"Source-stated"},{"text":"The authors report fine-tuning π0.5 and MolmoAct2 on UMI-Live data, with success rates of 10/10 and 8/10, respectively, on a bimanual pick-and-place task.","excerpt":"achieving success rates of 10/10 and 8/10, respectively on a bimanual pick-and-place task.","url":"https://arxiv.org/abs/2610.39385","relationship":"Source-stated"},{"text":"The authors say they release 40 hours of UMI-Live data across three tasks, plus hardware designs, software, and reproduction guides.","excerpt":"we release 40 hours of UMI-Live data across three tasks, and complete hardware designs, software, and reproduction guides","url":"https://arxiv.org/abs/2610.39385","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2610.39385","checkedAt":"2026-11-10T08:40:10.797Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How well does UMI-Live’s data support performance beyond the reported bimanual pick-and-place task?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"arXiv: UMI-Live for In-The-Wild Bimanual Robotics Data Collection","summary":"The authors present UMI-Live, an open-source system for collecting demonstrations to train robots on bimanual tasks. Human experts demonstrate tasks without teleoperating physical robots; the system is designed to collect data without external tracking equipment or calibration in the collection environment.\n\nThe authors say the hardware combines camera and LiDAR--inertial sensing, onboard compute, and power. A user interface records demonstrations while an asynchronous pipeline performs parallel SLAM reconstruction and publishes training-ready datasets. The release also includes software for dual-YAM teleoperation and VLA policy training and inference.\n\nThe authors report fine-tuning π0.5 and MolmoAct2 on UMI-Live data, with success rates of 10/10 and 8/10, respectively, on a bimanual pick-and-place task. They also report releasing 40 hours of data across three tasks, plus hardware designs, software, and reproduction guides.\n\nThe authors’ contribution is a collection setup intended to reduce equipment and calibration requirements while turning demonstrations into training-ready data. The abstract does not describe the evaluation procedure or show how well the system performs across other tasks and settings. How well does UMI-Live’s data support performance beyond the reported bimanual pick-and-place task?","question":"How well does UMI-Live’s data support performance beyond the reported bimanual pick-and-place task?","claims":[{"text":"The authors present UMI-Live as an open-source bimanual system for collecting data to train robots, using human demonstrations rather than teleoperation of physical robots.","excerpt":"Universal Manipulation Interface (UMI) devices allow human experts to demonstrate tasks for robot learning without teleoperating physical robots. Using these demonstrations as training data","support":"source-stated"},{"text":"The authors say their custom hardware integrates sensing, compute, and power, and needs neither external tracking equipment nor calibration within the collection environment.","excerpt":"camera and LiDAR--inertial sensing, onboard compute, and power, without external tracking equipment or calibration within the collection environment.","support":"source-stated"},{"text":"The system records data through a user interface while an asynchronous pipeline performs parallel SLAM reconstruction and publishes training-ready datasets.","excerpt":"an asynchronous pipeline performs parallel SLAM reconstruction and publishes training-ready datasets.","support":"source-stated"},{"text":"The authors report fine-tuning π0.5 and MolmoAct2 on UMI-Live data, with success rates of 10/10 and 8/10, respectively, on a bimanual pick-and-place task.","excerpt":"achieving success rates of 10/10 and 8/10, respectively on a bimanual pick-and-place task.","support":"source-stated"},{"text":"The authors say they release 40 hours of UMI-Live data across three tasks, plus hardware designs, software, and reproduction guides.","excerpt":"we release 40 hours of UMI-Live data across three tasks, and complete hardware designs, software, and reproduction guides","support":"source-stated"}],"limitations":["The abstract does not describe the evaluation procedure or provide details about the reported success rates beyond the bimanual pick-and-place task.","The abstract does not establish how well the system or its data performs across other tasks or collection settings."],"mediaId":"597df9bbfb2f979525ad28476549a069745f589620fcdbb359a78163b50fabfd","correctionReason":null,"id":"1d0a1454b704d7a582b5a8f1b7347106ae96fe842718ab7fb9dd4ecbcc6c32de","publicationId":"e39bd50e-7dee-4821-8add-b4043840e940","version":1,"sourceUrl":"https://arxiv.org/abs/2610.39385","sourceTitle":"UMI-Live for In-The-Wild Bimanual Robotics Data Collection","publisher":"arXiv eess.SY","publishedAt":null,"originalPublished":{"value":"2026-10-29","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-03T08:51:43.878Z","createdAt":"2026-11-10T08:08:08.099Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"597df9bbfb2f979525ad28476549a069745f589620fcdbb359a78163b50fabfd","kind":"article","url":"https://arxiv.org/abs/2610.39385","sourceUrl":"https://arxiv.org/abs/2610.39385","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-03T20:18:30.494Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2610.39385","checkedAt":"2026-11-10T08:40:10.797Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"597df9bbfb2f979525ad28476549a069745f589620fcdbb359a78163b50fabfd","confidence":0.9,"provider":"typesafe","model":"jev-1.13.0","inputHash":"81f32c241a28a29df74ed778f4775435f60f5b484a0fdad4024e8f62f6fdc663","decidedAt":"2026-11-03T20:18:35.153Z"},"provenance":{"textHash":"f1c68f4ba0c196958b5f29a14a919ae5d01a7b70291f48956ecd90dd122fd057","observationId":"b852639a11866f1e7221610a3eb6de0cce2c106d89996133c96be364a8975d0f","retrievedAt":"2026-11-03T20:18:29.998Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"e5143f06-fe8d-44ad-8c8b-1581c54033a8"},"reviewer":{"model":"gpt-6-luna","runId":"4580569e-96ed-4922-8bec-7cc6a9f8a9b8"}},"audit":{"status":"passed","reviewedAt":"2026-11-10T08:08:08.099Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Velamoor, Govind","url":null,"kind":"unknown"},{"name":"Lin, Adrian","url":null,"kind":"unknown"},{"name":"Jung, Sam","url":null,"kind":"unknown"},{"name":"Mateega, Spencer","url":null,"kind":"unknown"},{"name":"Yang, Brandon Y.","url":null,"kind":"unknown"},{"name":"Sethi, Amish","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Velamoor, Govind","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Velamoor, Govind","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.39385","locator":"head/meta[name=author|citation_author]"},"assertionId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T08:40:10.797Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2","sourceUrl":"https://arxiv.org/abs/2610.39385","contentSha256":"b8bfed406d001c5e1ad89407c829e587b8287dc594a4533ba1af242cd81f61b1","textHash":"f1c68f4ba0c196958b5f29a14a919ae5d01a7b70291f48956ecd90dd122fd057","observedAt":"2026-11-10T08:40:10.797Z"}},{"actor":{"name":"Lin, Adrian","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Lin, Adrian","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.39385","locator":"head/meta[name=author|citation_author]"},"assertionId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T08:40:10.797Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2","sourceUrl":"https://arxiv.org/abs/2610.39385","contentSha256":"b8bfed406d001c5e1ad89407c829e587b8287dc594a4533ba1af242cd81f61b1","textHash":"f1c68f4ba0c196958b5f29a14a919ae5d01a7b70291f48956ecd90dd122fd057","observedAt":"2026-11-10T08:40:10.797Z"}},{"actor":{"name":"Jung, Sam","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Jung, Sam","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.39385","locator":"head/meta[name=author|citation_author]"},"assertionId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T08:40:10.797Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2","sourceUrl":"https://arxiv.org/abs/2610.39385","contentSha256":"b8bfed406d001c5e1ad89407c829e587b8287dc594a4533ba1af242cd81f61b1","textHash":"f1c68f4ba0c196958b5f29a14a919ae5d01a7b70291f48956ecd90dd122fd057","observedAt":"2026-11-10T08:40:10.797Z"}},{"actor":{"name":"Mateega, Spencer","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Mateega, Spencer","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.39385","locator":"head/meta[name=author|citation_author]"},"assertionId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T08:40:10.797Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2","sourceUrl":"https://arxiv.org/abs/2610.39385","contentSha256":"b8bfed406d001c5e1ad89407c829e587b8287dc594a4533ba1af242cd81f61b1","textHash":"f1c68f4ba0c196958b5f29a14a919ae5d01a7b70291f48956ecd90dd122fd057","observedAt":"2026-11-10T08:40:10.797Z"}},{"actor":{"name":"Yang, Brandon Y.","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Yang, Brandon Y.","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.39385","locator":"head/meta[name=author|citation_author]"},"assertionId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T08:40:10.797Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2","sourceUrl":"https://arxiv.org/abs/2610.39385","contentSha256":"b8bfed406d001c5e1ad89407c829e587b8287dc594a4533ba1af242cd81f61b1","textHash":"f1c68f4ba0c196958b5f29a14a919ae5d01a7b70291f48956ecd90dd122fd057","observedAt":"2026-11-10T08:40:10.797Z"}},{"actor":{"name":"Sethi, Amish","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Sethi, Amish","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.39385","locator":"head/meta[name=author|citation_author]"},"assertionId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2:participation:5","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-10T08:40:10.797Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2","sourceUrl":"https://arxiv.org/abs/2610.39385","contentSha256":"b8bfed406d001c5e1ad89407c829e587b8287dc594a4533ba1af242cd81f61b1","textHash":"f1c68f4ba0c196958b5f29a14a919ae5d01a7b70291f48956ecd90dd122fd057","observedAt":"2026-11-10T08:40:10.797Z"}}],"participationCoverage":{"total":6,"returned":6,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2610.39385","snapshotId":"bf1cd6a95ec46414684f5e742e5bb743efac42211af7af06db3a54ac3593f0f2","contentSha256":"b8bfed406d001c5e1ad89407c829e587b8287dc594a4533ba1af242cd81f61b1","textHash":"f1c68f4ba0c196958b5f29a14a919ae5d01a7b70291f48956ecd90dd122fd057","observedAt":"2026-11-10T08:40:10.797Z","basis":"publisher-page-metadata","sourceLinks":49,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/1d0a1454b704d7a582b5a8f1b7347106ae96fe842718ab7fb9dd4ecbcc6c32de/source"}}}],"audits":[{"publicationId":"e39bd50e-7dee-4821-8add-b4043840e940","verdict":"passed","findings":[{"dimension":"attribution","status":"pass","explanation":"The reading attributes the system, hardware design, released materials and reported results to the paper's authors, matching the arXiv abstract. It keeps the success rates as the authors' reported findings.","excerpt":"We present UMI-Live, a full-stack open-source bimanual UMI system for in-the-wild data collection."},{"dimension":"summary","status":"pass","explanation":"The reading clearly states the core development: an open-source system for collecting human demonstrations that are turned into robot training data, without external tracking or on-site calibration. Several technical terms are left unexplained: bimanual, SLAM, VLA, dual-YAM, π0.5 and MolmoAct2. Tooltips would help, but the central idea remains understandable. The limitations and the question about task generalization follow from the abstract. The omission of the source's in-the-wild framing is minor.","excerpt":"Universal Manipulation Interface (UMI) devices allow human experts to demonstrate tasks for robot learning without teleoperating physical robots."},{"dimension":"date","status":"pass","explanation":"The originalPublished value of 2026-10-29 matches the arXiv v1 submission date. The later feed date of 2026-11-03 is correctly not used.","excerpt":"[Submitted on 29 Oct 2026]"},{"dimension":"support","status":"pass","explanation":"Each claim is directly supported by the abstract. The framing that the setup is intended to reduce equipment and calibration requirements is a fair reading of the abstract's motivation. The noted gaps are accurate: the abstract gives no evaluation procedure and reports results for only one task. The reading does not turn the 10/10 and 8/10 scores into a claim of general reliability.","excerpt":"Using these demonstrations as training data still requires a consistent hardware setup, calibration, and data processing."},{"dimension":"media","status":"pass","explanation":"The selected media is the verified arXiv abstract page for this paper, which is the direct original source.","excerpt":null}],"createdAt":"2026-11-10T08:08:08.099Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}