{"id":"ec525db3626e03336483c8549fe1aead05ff3cb433b354b911e986905a3b3d59","title":"arXiv eess.SY: Authors propose an onboard vision-language navigation stack for aerial robots","argument":"The authors of “VLN on the Fly” propose an onboard stack that separates grounding, planning, and control rather than combining them in one end-to-end policy. Their design uses a quantized VLM to map an instruction to an image cell, depth to turn that target into a 3D goal, a B-spline planner to produce a trajectory, and a pretrained reinforcement learning policy to track it.\n\nIn 15 onboard flights involving three everyday referents in a controlled indoor volume, the stack reached the target in 13 trials, with 5.72 cm mean goal error and 39.3% average GPU utilization. In six additional cluttered-environment trials, it tracked collision-free trajectories under onboard perception gating. These results describe a limited set of flights; the abstract does not establish performance in other environments or broader operational reliability. The authors argue that keeping stages inspectable preserves observability and safety checks that end-to-end policies give up. As an Escape Meridian reading, the abstract leaves performance beyond the described trials unknown.\n\nHow would the stack perform in environments beyond the controlled indoor volume and the six cluttered-environment trials described in the abstract?","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The abstract reports a limited number of flights and does not establish performance outside the described controlled indoor volume and cluttered-environment trials. The source page provides the abstract, not the full paper's methods or results; this reading is limited to the claims in that abstract.","speaker":{"name":"Tayar, Marco S., Tommaselli, Felipe, Capezzuto, Gianluca, Saraiva, Pedro Antonio Rabelo, de Freitas, Pedro H. V., Kido, Lucas, Sonego, Guilherme, Godoy, Ricardo V., Becker, Marcelo","roleAtSource":"Article author"},"source":{"title":"VLN on the Fly: An Onboard Vision-Language Navigation Stack for Aerial Robots","publisher":"arXiv eess.SY","format":"Article","url":"https://arxiv.org/abs/2609.20191","publishedAt":"2026-08-07","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors propose an onboard aerial-robot navigation stack with grounding, planning, and control as separate stages.","excerpt":"We propose VLN on the Fly, an onboard stack that keeps grounding, planning, and control as separate, inspectable stages.","url":"https://arxiv.org/abs/2609.20191","relationship":"Source-stated"},{"text":"The stack uses a quantized VLM, depth, a B-spline planner, and a pretrained reinforcement learning policy in sequence.","excerpt":"A quantized VLM grounds an instruction to a coarse image cell, depth lifts it to a 3D goal, a fast B-spline planner returns a feasible trajectory, and a pretrained reinforcement learning policy tracks it to motor commands across quadrotors.","url":"https://arxiv.org/abs/2609.20191","relationship":"Source-stated"},{"text":"Across 15 onboard flights in a controlled indoor volume, the stack reached the target in 13 trials, with 5.72 cm mean goal error and 39.3% average GPU utilization.","excerpt":"Across 15 onboard flights over three everyday referents in a controlled indoor volume, the stack reaches the target in 13 of 15 trials with 5.72 cm mean goal error and 39.3% average GPU utilization.","url":"https://arxiv.org/abs/2609.20191","relationship":"Source-stated"},{"text":"In six additional cluttered-environment trials, the stack tracked collision-free trajectories under onboard perception gating.","excerpt":"In 6 additional cluttered-environment trials, the stack tracks collision-free trajectories under onboard perception gating.","url":"https://arxiv.org/abs/2609.20191","relationship":"Source-stated"},{"text":"The abstract reports these results in a controlled indoor volume and a further six cluttered-environment trials; performance beyond those settings is not established there.","excerpt":"Across 15 onboard flights over three everyday referents in a controlled indoor volume","url":"https://arxiv.org/abs/2609.20191","relationship":"Escape Meridian interpretation"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2609.20191","checkedAt":"2026-11-05T11:22:07.040Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How would the stack perform in environments beyond the controlled indoor volume and the six cluttered-environment trials described in the abstract?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"arXiv eess.SY: Authors propose an onboard vision-language navigation stack for aerial robots","summary":"The authors of “VLN on the Fly” propose an onboard stack that separates grounding, planning, and control rather than combining them in one end-to-end policy. Their design uses a quantized VLM to map an instruction to an image cell, depth to turn that target into a 3D goal, a B-spline planner to produce a trajectory, and a pretrained reinforcement learning policy to track it.\n\nIn 15 onboard flights involving three everyday referents in a controlled indoor volume, the stack reached the target in 13 trials, with 5.72 cm mean goal error and 39.3% average GPU utilization. In six additional cluttered-environment trials, it tracked collision-free trajectories under onboard perception gating. These results describe a limited set of flights; the abstract does not establish performance in other environments or broader operational reliability. The authors argue that keeping stages inspectable preserves observability and safety checks that end-to-end policies give up. As an Escape Meridian reading, the abstract leaves performance beyond the described trials unknown.\n\nHow would the stack perform in environments beyond the controlled indoor volume and the six cluttered-environment trials described in the abstract?","question":"How would the stack perform in environments beyond the controlled indoor volume and the six cluttered-environment trials described in the abstract?","claims":[{"text":"The authors propose an onboard aerial-robot navigation stack with grounding, planning, and control as separate stages.","excerpt":"We propose VLN on the Fly, an onboard stack that keeps grounding, planning, and control as separate, inspectable stages.","support":"source-stated"},{"text":"The stack uses a quantized VLM, depth, a B-spline planner, and a pretrained reinforcement learning policy in sequence.","excerpt":"A quantized VLM grounds an instruction to a coarse image cell, depth lifts it to a 3D goal, a fast B-spline planner returns a feasible trajectory, and a pretrained reinforcement learning policy tracks it to motor commands across quadrotors.","support":"source-stated"},{"text":"Across 15 onboard flights in a controlled indoor volume, the stack reached the target in 13 trials, with 5.72 cm mean goal error and 39.3% average GPU utilization.","excerpt":"Across 15 onboard flights over three everyday referents in a controlled indoor volume, the stack reaches the target in 13 of 15 trials with 5.72 cm mean goal error and 39.3% average GPU utilization.","support":"source-stated"},{"text":"In six additional cluttered-environment trials, the stack tracked collision-free trajectories under onboard perception gating.","excerpt":"In 6 additional cluttered-environment trials, the stack tracks collision-free trajectories under onboard perception gating.","support":"source-stated"},{"text":"The abstract reports these results in a controlled indoor volume and a further six cluttered-environment trials; performance beyond those settings is not established there.","excerpt":"Across 15 onboard flights over three everyday referents in a controlled indoor volume","support":"interpretation"}],"limitations":["The abstract reports a limited number of flights and does not establish performance outside the described controlled indoor volume and cluttered-environment trials.","The source page provides the abstract, not the full paper's methods or results; this reading is limited to the claims in that abstract."],"mediaId":"380420548dfdf52be91ded8caebbe01827c9738096b6d97c35b69cf92ab0e0e3","correctionReason":null,"id":"ec525db3626e03336483c8549fe1aead05ff3cb433b354b911e986905a3b3d59","publicationId":"6d89c36b-fb16-4b53-8eb3-61c709b06fe1","version":1,"sourceUrl":"https://arxiv.org/abs/2609.20191","sourceTitle":"VLN on the Fly: An Onboard Vision-Language Navigation Stack for Aerial Robots","publisher":"arXiv eess.SY","publishedAt":null,"originalPublished":{"value":"2026-08-07","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-10-27T04:27:08.136Z","createdAt":"2026-11-05T10:18:25.660Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"380420548dfdf52be91ded8caebbe01827c9738096b6d97c35b69cf92ab0e0e3","kind":"article","url":"https://arxiv.org/abs/2609.20191","sourceUrl":"https://arxiv.org/abs/2609.20191","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-10-30T01:48:54.046Z","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/2609.20191","checkedAt":"2026-11-05T11:22:07.040Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"380420548dfdf52be91ded8caebbe01827c9738096b6d97c35b69cf92ab0e0e3","confidence":0.83,"provider":"typesafe","model":"jev-1.13.0","inputHash":"ecaa22ea48fe71d417ab4a9f752b684f685d038c76150124ac2cd0626ed03044","decidedAt":"2026-10-30T01:48:55.916Z"},"provenance":{"textHash":"6866c05de5800017cff735ce5f741ee58ba8eae629f906b31386e08ab4fbbf27","observationId":"4ff3d66e9f4f922e6344c244c187ca13dfb467c1284aee0082d755b02d5bcfda","retrievedAt":"2026-10-30T01:48:53.748Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"828ea604-ea50-4846-8a29-d2de33487fc2"},"reviewer":{"model":"gpt-6-luna","runId":"5076b4c9-858b-4580-8a18-2b252294d51d"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-05T10:18:25.660Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Tayar, Marco S.","url":null,"kind":"unknown"},{"name":"Tommaselli, Felipe","url":null,"kind":"unknown"},{"name":"Capezzuto, Gianluca","url":null,"kind":"unknown"},{"name":"Saraiva, Pedro Antonio Rabelo","url":null,"kind":"unknown"},{"name":"de Freitas, Pedro H. V.","url":null,"kind":"unknown"},{"name":"Kido, Lucas","url":null,"kind":"unknown"},{"name":"Sonego, Guilherme","url":null,"kind":"unknown"},{"name":"Godoy, Ricardo V.","url":null,"kind":"unknown"},{"name":"Becker, Marcelo","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Tayar, Marco S.","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Tayar, Marco S.","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2609.20191","locator":"head/meta[name=author|citation_author]"},"assertionId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-05T11:22:07.040Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422","sourceUrl":"https://arxiv.org/abs/2609.20191","contentSha256":"4ab3e3068143c8b53569b3081c88bd8fc80b8be17c51eb1707febc8c3bb48903","textHash":"6866c05de5800017cff735ce5f741ee58ba8eae629f906b31386e08ab4fbbf27","observedAt":"2026-11-05T11:22:07.040Z"}},{"actor":{"name":"Tommaselli, Felipe","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Tommaselli, Felipe","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2609.20191","locator":"head/meta[name=author|citation_author]"},"assertionId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-05T11:22:07.040Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422","sourceUrl":"https://arxiv.org/abs/2609.20191","contentSha256":"4ab3e3068143c8b53569b3081c88bd8fc80b8be17c51eb1707febc8c3bb48903","textHash":"6866c05de5800017cff735ce5f741ee58ba8eae629f906b31386e08ab4fbbf27","observedAt":"2026-11-05T11:22:07.040Z"}},{"actor":{"name":"Capezzuto, Gianluca","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Capezzuto, Gianluca","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2609.20191","locator":"head/meta[name=author|citation_author]"},"assertionId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-05T11:22:07.040Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422","sourceUrl":"https://arxiv.org/abs/2609.20191","contentSha256":"4ab3e3068143c8b53569b3081c88bd8fc80b8be17c51eb1707febc8c3bb48903","textHash":"6866c05de5800017cff735ce5f741ee58ba8eae629f906b31386e08ab4fbbf27","observedAt":"2026-11-05T11:22:07.040Z"}},{"actor":{"name":"Saraiva, Pedro Antonio Rabelo","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Saraiva, Pedro Antonio Rabelo","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2609.20191","locator":"head/meta[name=author|citation_author]"},"assertionId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-05T11:22:07.040Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422","sourceUrl":"https://arxiv.org/abs/2609.20191","contentSha256":"4ab3e3068143c8b53569b3081c88bd8fc80b8be17c51eb1707febc8c3bb48903","textHash":"6866c05de5800017cff735ce5f741ee58ba8eae629f906b31386e08ab4fbbf27","observedAt":"2026-11-05T11:22:07.040Z"}},{"actor":{"name":"de Freitas, Pedro H. V.","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"de Freitas, Pedro H. V.","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2609.20191","locator":"head/meta[name=author|citation_author]"},"assertionId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-05T11:22:07.040Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422","sourceUrl":"https://arxiv.org/abs/2609.20191","contentSha256":"4ab3e3068143c8b53569b3081c88bd8fc80b8be17c51eb1707febc8c3bb48903","textHash":"6866c05de5800017cff735ce5f741ee58ba8eae629f906b31386e08ab4fbbf27","observedAt":"2026-11-05T11:22:07.040Z"}},{"actor":{"name":"Kido, Lucas","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Kido, Lucas","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2609.20191","locator":"head/meta[name=author|citation_author]"},"assertionId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422:participation:5","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-05T11:22:07.040Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422","sourceUrl":"https://arxiv.org/abs/2609.20191","contentSha256":"4ab3e3068143c8b53569b3081c88bd8fc80b8be17c51eb1707febc8c3bb48903","textHash":"6866c05de5800017cff735ce5f741ee58ba8eae629f906b31386e08ab4fbbf27","observedAt":"2026-11-05T11:22:07.040Z"}},{"actor":{"name":"Sonego, Guilherme","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Sonego, Guilherme","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2609.20191","locator":"head/meta[name=author|citation_author]"},"assertionId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422:participation:6","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-05T11:22:07.040Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422","sourceUrl":"https://arxiv.org/abs/2609.20191","contentSha256":"4ab3e3068143c8b53569b3081c88bd8fc80b8be17c51eb1707febc8c3bb48903","textHash":"6866c05de5800017cff735ce5f741ee58ba8eae629f906b31386e08ab4fbbf27","observedAt":"2026-11-05T11:22:07.040Z"}},{"actor":{"name":"Godoy, Ricardo V.","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Godoy, Ricardo V.","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2609.20191","locator":"head/meta[name=author|citation_author]"},"assertionId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422:participation:7","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-05T11:22:07.040Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422","sourceUrl":"https://arxiv.org/abs/2609.20191","contentSha256":"4ab3e3068143c8b53569b3081c88bd8fc80b8be17c51eb1707febc8c3bb48903","textHash":"6866c05de5800017cff735ce5f741ee58ba8eae629f906b31386e08ab4fbbf27","observedAt":"2026-11-05T11:22:07.040Z"}},{"actor":{"name":"Becker, Marcelo","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Becker, Marcelo","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2609.20191","locator":"head/meta[name=author|citation_author]"},"assertionId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422:participation:8","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-05T11:22:07.040Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422","sourceUrl":"https://arxiv.org/abs/2609.20191","contentSha256":"4ab3e3068143c8b53569b3081c88bd8fc80b8be17c51eb1707febc8c3bb48903","textHash":"6866c05de5800017cff735ce5f741ee58ba8eae629f906b31386e08ab4fbbf27","observedAt":"2026-11-05T11:22:07.040Z"}}],"participationCoverage":{"total":9,"returned":9,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2609.20191","snapshotId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422","contentSha256":"4ab3e3068143c8b53569b3081c88bd8fc80b8be17c51eb1707febc8c3bb48903","textHash":"6866c05de5800017cff735ce5f741ee58ba8eae629f906b31386e08ab4fbbf27","observedAt":"2026-11-05T11:22:07.040Z","basis":"publisher-page-metadata","sourceLinks":55,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/ec525db3626e03336483c8549fe1aead05ff3cb433b354b911e986905a3b3d59/source"}},"canonicalUrl":"https://escapemeridian.com/readings/arxiv-eess-sy-authors-propose-an-onboard-vision-language-ec525db3","formats":{"html":"/readings/arxiv-eess-sy-authors-propose-an-onboard-vision-language-ec525db3","markdown":"/readings/arxiv-eess-sy-authors-propose-an-onboard-vision-language-ec525db3.md","json":"/readings/arxiv-eess-sy-authors-propose-an-onboard-vision-language-ec525db3.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-quantization","name":"Quantization","explanation":"Storing a model's numbers at lower precision so it uses less memory and runs faster, sometimes at a cost in accuracy.","referenceUrl":"https://developers.google.com/machine-learning/glossary#quantization"},{"id":"term-pre-training","name":"Pre-training","explanation":"A model's initial training on a large dataset, before further adaptation for particular uses.","referenceUrl":"https://developers.google.com/machine-learning/glossary#pre-training"},{"id":"term-reinforcement-learning","name":"Reinforcement learning","explanation":"Training by trial and error, in which a system is rewarded for the outcomes its designers want.","referenceUrl":"https://developers.google.com/machine-learning/glossary#reinforcement-learning-rl"},{"id":"term-gpu","name":"Graphics processing unit","explanation":"A chip that performs many calculations at once; the main hardware used to train and run AI models.","referenceUrl":"https://en.wikipedia.org/wiki/Graphics_processing_unit"}],"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":"ec525db3626e03336483c8549fe1aead05ff3cb433b354b911e986905a3b3d59","revision":"6d89c36b-fb16-4b53-8eb3-61c709b06fe1","revisionKind":"publication-id","title":"arXiv eess.SY: Authors propose an onboard vision-language navigation stack for aerial robots","url":"https://escapemeridian.com/readings/arxiv-eess-sy-authors-propose-an-onboard-vision-language-ec525db3"},"source":{"type":"source","id":"https://arxiv.org/abs/2609.20191","revision":"6866c05de5800017cff735ce5f741ee58ba8eae629f906b31386e08ab4fbbf27","revisionKind":"sha256-source-text","title":"VLN on the Fly: An Onboard Vision-Language Navigation Stack for Aerial Robots","url":"https://arxiv.org/abs/2609.20191"},"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":"ec525db3626e03336483c8549fe1aead05ff3cb433b354b911e986905a3b3d59","title":"arXiv eess.SY: Authors propose an onboard vision-language navigation stack for aerial robots","argument":"The authors of “VLN on the Fly” propose an onboard stack that separates grounding, planning, and control rather than combining them in one end-to-end policy. Their design uses a quantized VLM to map an instruction to an image cell, depth to turn that target into a 3D goal, a B-spline planner to produce a trajectory, and a pretrained reinforcement learning policy to track it.\n\nIn 15 onboard flights involving three everyday referents in a controlled indoor volume, the stack reached the target in 13 trials, with 5.72 cm mean goal error and 39.3% average GPU utilization. In six additional cluttered-environment trials, it tracked collision-free trajectories under onboard perception gating. These results describe a limited set of flights; the abstract does not establish performance in other environments or broader operational reliability. The authors argue that keeping stages inspectable preserves observability and safety checks that end-to-end policies give up. As an Escape Meridian reading, the abstract leaves performance beyond the described trials unknown.\n\nHow would the stack perform in environments beyond the controlled indoor volume and the six cluttered-environment trials described in the abstract?","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The abstract reports a limited number of flights and does not establish performance outside the described controlled indoor volume and cluttered-environment trials. The source page provides the abstract, not the full paper's methods or results; this reading is limited to the claims in that abstract.","speaker":{"name":"Tayar, Marco S., Tommaselli, Felipe, Capezzuto, Gianluca, Saraiva, Pedro Antonio Rabelo, de Freitas, Pedro H. V., Kido, Lucas, Sonego, Guilherme, Godoy, Ricardo V., Becker, Marcelo","roleAtSource":"Article author"},"source":{"title":"VLN on the Fly: An Onboard Vision-Language Navigation Stack for Aerial Robots","publisher":"arXiv eess.SY","format":"Article","url":"https://arxiv.org/abs/2609.20191","publishedAt":"2026-08-07","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors propose an onboard aerial-robot navigation stack with grounding, planning, and control as separate stages.","excerpt":"We propose VLN on the Fly, an onboard stack that keeps grounding, planning, and control as separate, inspectable stages.","url":"https://arxiv.org/abs/2609.20191","relationship":"Source-stated"},{"text":"The stack uses a quantized VLM, depth, a B-spline planner, and a pretrained reinforcement learning policy in sequence.","excerpt":"A quantized VLM grounds an instruction to a coarse image cell, depth lifts it to a 3D goal, a fast B-spline planner returns a feasible trajectory, and a pretrained reinforcement learning policy tracks it to motor commands across quadrotors.","url":"https://arxiv.org/abs/2609.20191","relationship":"Source-stated"},{"text":"Across 15 onboard flights in a controlled indoor volume, the stack reached the target in 13 trials, with 5.72 cm mean goal error and 39.3% average GPU utilization.","excerpt":"Across 15 onboard flights over three everyday referents in a controlled indoor volume, the stack reaches the target in 13 of 15 trials with 5.72 cm mean goal error and 39.3% average GPU utilization.","url":"https://arxiv.org/abs/2609.20191","relationship":"Source-stated"},{"text":"In six additional cluttered-environment trials, the stack tracked collision-free trajectories under onboard perception gating.","excerpt":"In 6 additional cluttered-environment trials, the stack tracks collision-free trajectories under onboard perception gating.","url":"https://arxiv.org/abs/2609.20191","relationship":"Source-stated"},{"text":"The abstract reports these results in a controlled indoor volume and a further six cluttered-environment trials; performance beyond those settings is not established there.","excerpt":"Across 15 onboard flights over three everyday referents in a controlled indoor volume","url":"https://arxiv.org/abs/2609.20191","relationship":"Escape Meridian interpretation"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2609.20191","checkedAt":"2026-11-05T11:22:07.040Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How would the stack perform in environments beyond the controlled indoor volume and the six cluttered-environment trials described in the abstract?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"arXiv eess.SY: Authors propose an onboard vision-language navigation stack for aerial robots","summary":"The authors of “VLN on the Fly” propose an onboard stack that separates grounding, planning, and control rather than combining them in one end-to-end policy. Their design uses a quantized VLM to map an instruction to an image cell, depth to turn that target into a 3D goal, a B-spline planner to produce a trajectory, and a pretrained reinforcement learning policy to track it.\n\nIn 15 onboard flights involving three everyday referents in a controlled indoor volume, the stack reached the target in 13 trials, with 5.72 cm mean goal error and 39.3% average GPU utilization. In six additional cluttered-environment trials, it tracked collision-free trajectories under onboard perception gating. These results describe a limited set of flights; the abstract does not establish performance in other environments or broader operational reliability. The authors argue that keeping stages inspectable preserves observability and safety checks that end-to-end policies give up. As an Escape Meridian reading, the abstract leaves performance beyond the described trials unknown.\n\nHow would the stack perform in environments beyond the controlled indoor volume and the six cluttered-environment trials described in the abstract?","question":"How would the stack perform in environments beyond the controlled indoor volume and the six cluttered-environment trials described in the abstract?","claims":[{"text":"The authors propose an onboard aerial-robot navigation stack with grounding, planning, and control as separate stages.","excerpt":"We propose VLN on the Fly, an onboard stack that keeps grounding, planning, and control as separate, inspectable stages.","support":"source-stated"},{"text":"The stack uses a quantized VLM, depth, a B-spline planner, and a pretrained reinforcement learning policy in sequence.","excerpt":"A quantized VLM grounds an instruction to a coarse image cell, depth lifts it to a 3D goal, a fast B-spline planner returns a feasible trajectory, and a pretrained reinforcement learning policy tracks it to motor commands across quadrotors.","support":"source-stated"},{"text":"Across 15 onboard flights in a controlled indoor volume, the stack reached the target in 13 trials, with 5.72 cm mean goal error and 39.3% average GPU utilization.","excerpt":"Across 15 onboard flights over three everyday referents in a controlled indoor volume, the stack reaches the target in 13 of 15 trials with 5.72 cm mean goal error and 39.3% average GPU utilization.","support":"source-stated"},{"text":"In six additional cluttered-environment trials, the stack tracked collision-free trajectories under onboard perception gating.","excerpt":"In 6 additional cluttered-environment trials, the stack tracks collision-free trajectories under onboard perception gating.","support":"source-stated"},{"text":"The abstract reports these results in a controlled indoor volume and a further six cluttered-environment trials; performance beyond those settings is not established there.","excerpt":"Across 15 onboard flights over three everyday referents in a controlled indoor volume","support":"interpretation"}],"limitations":["The abstract reports a limited number of flights and does not establish performance outside the described controlled indoor volume and cluttered-environment trials.","The source page provides the abstract, not the full paper's methods or results; this reading is limited to the claims in that abstract."],"mediaId":"380420548dfdf52be91ded8caebbe01827c9738096b6d97c35b69cf92ab0e0e3","correctionReason":null,"id":"ec525db3626e03336483c8549fe1aead05ff3cb433b354b911e986905a3b3d59","publicationId":"6d89c36b-fb16-4b53-8eb3-61c709b06fe1","version":1,"sourceUrl":"https://arxiv.org/abs/2609.20191","sourceTitle":"VLN on the Fly: An Onboard Vision-Language Navigation Stack for Aerial Robots","publisher":"arXiv eess.SY","publishedAt":null,"originalPublished":{"value":"2026-08-07","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-10-27T04:27:08.136Z","createdAt":"2026-11-05T10:18:25.660Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"380420548dfdf52be91ded8caebbe01827c9738096b6d97c35b69cf92ab0e0e3","kind":"article","url":"https://arxiv.org/abs/2609.20191","sourceUrl":"https://arxiv.org/abs/2609.20191","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-10-30T01:48:54.046Z","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/2609.20191","checkedAt":"2026-11-05T11:22:07.040Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"380420548dfdf52be91ded8caebbe01827c9738096b6d97c35b69cf92ab0e0e3","confidence":0.83,"provider":"typesafe","model":"jev-1.13.0","inputHash":"ecaa22ea48fe71d417ab4a9f752b684f685d038c76150124ac2cd0626ed03044","decidedAt":"2026-10-30T01:48:55.916Z"},"provenance":{"textHash":"6866c05de5800017cff735ce5f741ee58ba8eae629f906b31386e08ab4fbbf27","observationId":"4ff3d66e9f4f922e6344c244c187ca13dfb467c1284aee0082d755b02d5bcfda","retrievedAt":"2026-10-30T01:48:53.748Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"828ea604-ea50-4846-8a29-d2de33487fc2"},"reviewer":{"model":"gpt-6-luna","runId":"5076b4c9-858b-4580-8a18-2b252294d51d"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-05T10:18:25.660Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Tayar, Marco S.","url":null,"kind":"unknown"},{"name":"Tommaselli, Felipe","url":null,"kind":"unknown"},{"name":"Capezzuto, Gianluca","url":null,"kind":"unknown"},{"name":"Saraiva, Pedro Antonio Rabelo","url":null,"kind":"unknown"},{"name":"de Freitas, Pedro H. V.","url":null,"kind":"unknown"},{"name":"Kido, Lucas","url":null,"kind":"unknown"},{"name":"Sonego, Guilherme","url":null,"kind":"unknown"},{"name":"Godoy, Ricardo V.","url":null,"kind":"unknown"},{"name":"Becker, Marcelo","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Tayar, Marco S.","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Tayar, Marco S.","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2609.20191","locator":"head/meta[name=author|citation_author]"},"assertionId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-05T11:22:07.040Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422","sourceUrl":"https://arxiv.org/abs/2609.20191","contentSha256":"4ab3e3068143c8b53569b3081c88bd8fc80b8be17c51eb1707febc8c3bb48903","textHash":"6866c05de5800017cff735ce5f741ee58ba8eae629f906b31386e08ab4fbbf27","observedAt":"2026-11-05T11:22:07.040Z"}},{"actor":{"name":"Tommaselli, Felipe","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Tommaselli, Felipe","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2609.20191","locator":"head/meta[name=author|citation_author]"},"assertionId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-05T11:22:07.040Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422","sourceUrl":"https://arxiv.org/abs/2609.20191","contentSha256":"4ab3e3068143c8b53569b3081c88bd8fc80b8be17c51eb1707febc8c3bb48903","textHash":"6866c05de5800017cff735ce5f741ee58ba8eae629f906b31386e08ab4fbbf27","observedAt":"2026-11-05T11:22:07.040Z"}},{"actor":{"name":"Capezzuto, Gianluca","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Capezzuto, Gianluca","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2609.20191","locator":"head/meta[name=author|citation_author]"},"assertionId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-05T11:22:07.040Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422","sourceUrl":"https://arxiv.org/abs/2609.20191","contentSha256":"4ab3e3068143c8b53569b3081c88bd8fc80b8be17c51eb1707febc8c3bb48903","textHash":"6866c05de5800017cff735ce5f741ee58ba8eae629f906b31386e08ab4fbbf27","observedAt":"2026-11-05T11:22:07.040Z"}},{"actor":{"name":"Saraiva, Pedro Antonio Rabelo","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Saraiva, Pedro Antonio Rabelo","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2609.20191","locator":"head/meta[name=author|citation_author]"},"assertionId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-05T11:22:07.040Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422","sourceUrl":"https://arxiv.org/abs/2609.20191","contentSha256":"4ab3e3068143c8b53569b3081c88bd8fc80b8be17c51eb1707febc8c3bb48903","textHash":"6866c05de5800017cff735ce5f741ee58ba8eae629f906b31386e08ab4fbbf27","observedAt":"2026-11-05T11:22:07.040Z"}},{"actor":{"name":"de Freitas, Pedro H. V.","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"de Freitas, Pedro H. V.","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2609.20191","locator":"head/meta[name=author|citation_author]"},"assertionId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-05T11:22:07.040Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422","sourceUrl":"https://arxiv.org/abs/2609.20191","contentSha256":"4ab3e3068143c8b53569b3081c88bd8fc80b8be17c51eb1707febc8c3bb48903","textHash":"6866c05de5800017cff735ce5f741ee58ba8eae629f906b31386e08ab4fbbf27","observedAt":"2026-11-05T11:22:07.040Z"}},{"actor":{"name":"Kido, Lucas","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Kido, Lucas","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2609.20191","locator":"head/meta[name=author|citation_author]"},"assertionId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422:participation:5","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-05T11:22:07.040Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422","sourceUrl":"https://arxiv.org/abs/2609.20191","contentSha256":"4ab3e3068143c8b53569b3081c88bd8fc80b8be17c51eb1707febc8c3bb48903","textHash":"6866c05de5800017cff735ce5f741ee58ba8eae629f906b31386e08ab4fbbf27","observedAt":"2026-11-05T11:22:07.040Z"}},{"actor":{"name":"Sonego, Guilherme","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Sonego, Guilherme","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2609.20191","locator":"head/meta[name=author|citation_author]"},"assertionId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422:participation:6","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-05T11:22:07.040Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422","sourceUrl":"https://arxiv.org/abs/2609.20191","contentSha256":"4ab3e3068143c8b53569b3081c88bd8fc80b8be17c51eb1707febc8c3bb48903","textHash":"6866c05de5800017cff735ce5f741ee58ba8eae629f906b31386e08ab4fbbf27","observedAt":"2026-11-05T11:22:07.040Z"}},{"actor":{"name":"Godoy, Ricardo V.","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Godoy, Ricardo V.","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2609.20191","locator":"head/meta[name=author|citation_author]"},"assertionId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422:participation:7","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-05T11:22:07.040Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422","sourceUrl":"https://arxiv.org/abs/2609.20191","contentSha256":"4ab3e3068143c8b53569b3081c88bd8fc80b8be17c51eb1707febc8c3bb48903","textHash":"6866c05de5800017cff735ce5f741ee58ba8eae629f906b31386e08ab4fbbf27","observedAt":"2026-11-05T11:22:07.040Z"}},{"actor":{"name":"Becker, Marcelo","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Becker, Marcelo","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2609.20191","locator":"head/meta[name=author|citation_author]"},"assertionId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422:participation:8","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-05T11:22:07.040Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422","sourceUrl":"https://arxiv.org/abs/2609.20191","contentSha256":"4ab3e3068143c8b53569b3081c88bd8fc80b8be17c51eb1707febc8c3bb48903","textHash":"6866c05de5800017cff735ce5f741ee58ba8eae629f906b31386e08ab4fbbf27","observedAt":"2026-11-05T11:22:07.040Z"}}],"participationCoverage":{"total":9,"returned":9,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2609.20191","snapshotId":"a7b7bde4158a594eeb3f640535518fdf633585b0fc72d4a98c15ecf66ddb4422","contentSha256":"4ab3e3068143c8b53569b3081c88bd8fc80b8be17c51eb1707febc8c3bb48903","textHash":"6866c05de5800017cff735ce5f741ee58ba8eae629f906b31386e08ab4fbbf27","observedAt":"2026-11-05T11:22:07.040Z","basis":"publisher-page-metadata","sourceLinks":55,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/ec525db3626e03336483c8549fe1aead05ff3cb433b354b911e986905a3b3d59/source"}}}],"audits":[{"publicationId":"6d89c36b-fb16-4b53-8eb3-61c709b06fe1","verdict":"uncertain","findings":[{"dimension":"attribution","status":"pass","explanation":"The design, results and argument are attributed to the authors of the arXiv paper. The title prefix uses the feed label 'arXiv eess.SY', while the page lists cs.RO as the primary category. eess.SY is still a listed subject, so this is not a misattribution.","excerpt":"We propose VLN on the Fly, an onboard stack that keeps grounding, planning, and control as separate, inspectable stages."},{"dimension":"summary","status":"uncertain","explanation":"The reading is accurate but hard to follow for non-specialists. It never says plainly that vision-language navigation means a robot flying toward something described in words. It also omits the paper's core motivation: grounding, planning and control must share limited onboard compute, and errors are hard to isolate in flight. Without that, the 39.3% GPU figure has little meaning. Central terms (quantized VLM, grounding, referents, perception gating) go unexplained and need tooltips. The limits caveat repeats three times. Smallest repair: add the compute and error-isolation rationale and a one-line plain description of the task.","excerpt":"Running vision-language navigation fully onboard an aerial robot is hard, since grounding, planning, and control must share limited compute and a single-stage error is difficult to isolate in flight."},{"dimension":"date","status":"pass","explanation":"originalPublished 2026-08-07 matches the v1 submission date. It is not the 25 Oct 2026 v2 revision or the 27 Oct feed date.","excerpt":"Submitted on 7 Aug 2026 (v1), last revised 25 Oct 2026 (this version, v2)"},{"dimension":"support","status":"pass","explanation":"Each claim and figure matches the abstract: 13 of 15 trials, 5.72 cm, 39.3%, and six cluttered-environment trials. The interpretation about unknown performance beyond the tested settings is labeled and conditional. No unsupported benefits or costs are implied.","excerpt":"the stack reaches the target in 13 of 15 trials with 5.72 cm mean goal error and 39.3% average GPU utilization"},{"dimension":"media","status":"pass","explanation":"The selected media is the verified direct arXiv abstract page for this paper, which is the original source.","excerpt":null}],"createdAt":"2026-11-05T10:18:25.660Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}