{"id":57,"date":"2026-05-08T03:58:25","date_gmt":"2026-05-08T03:58:25","guid":{"rendered":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/?page_id=57"},"modified":"2026-05-08T23:50:24","modified_gmt":"2026-05-08T23:50:24","slug":"results","status":"publish","type":"page","link":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/results\/","title":{"rendered":"Results & Conclusions"},"content":{"rendered":"\n\n\n

Results<\/h2>\n\n

We build on 3DFA (Gkanatsios et al., 2025) as our base 3D policy and introduce \"\\delta, a learned orthogonal correction matrix that adapts the 3D RoPE positional encodings to account for unknown camera–robot extrinsics. Under fixed per-lab miscalibration — the realistic setting where each data source carries a consistent but unknown extrinsic error — our \"\\delta formulation recovers to near the performance of a model trained with ground-truth calibration, substantially outperforming a miscalibrated 3DFA without correction and a 2D baseline with no 3D coordinates. This confirms that the model can implicitly identify and compensate for a per-source calibration offset from the data alone. However, when miscalibration is randomized per episode, \"\\delta provides no benefit, suggesting the correction relies on consistent error structure across a source rather than frame-level adaptation. We further extend this approach to GR00T N1.5 by injecting 3D RoPE into its cross-attention layers and finetuning on a 1.4K-rollout subset of DROID. In this setting, 2D, 3D, and 3D+\"\\delta variants converge to similar training loss, pointing to the need for larger-scale or more calibration-diverse data for the 3D signal to provide a clear advantage over the pretrained 2D backbone.<\/p>\n\n\n\n\n

<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n
\"\"<\/figure>\n\n\n\n\n

Conclusion<\/h2>\n\n

We presented a method for training 3D Vision-Language-Action models on large-scale robot datasets with unknown or noisy camera-robot extrinsics. Our key contribution is \"\\delta, a learned orthogonal correction to 3D RoPE positional encodings that is predicted from image content alone and enforced to lie in \"SO(d)\" by construction via the matrix exponential. Under fixed per-source miscalibration, \"\\delta recovers most of the performance gap relative to a model with ground-truth calibration, without any explicit supervision on the extrinsics. This opens a practical path toward training large 3D-VLAs on datasets like DROID and OpenX, where calibration data is sparse, inconsistent, or entirely absent. A key limitation is that \"\\delta relies on consistent error structure within a source — randomly varying miscalibration remains unsolved, and likely requires a fundamentally different formulation, such as per-frame calibration estimation from multi-view geometric cues. We leave this as an important direction for future work.<\/p>\n\n\n\n\n

<\/p>\n","protected":false},"excerpt":{"rendered":"

Results We build on 3DFA (Gkanatsios et al., 2025) as our base 3D policy and introduce , a learned orthogonal correction matrix that adapts the 3D RoPE positional encodings to account for unknown camera–robot extrinsics. Under fixed per-lab miscalibration — the realistic setting where each data source carries a consistent but unknown extrinsic error — […]<\/p>\n","protected":false},"author":281,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"footnotes":""},"class_list":["post-57","page","type-page","status-publish","hentry"],"yoast_head":"\nResults & Conclusions - Learning 3D-VLAs with Noisy Miscalibrated Data<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/results\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Results & Conclusions - Learning 3D-VLAs with Noisy Miscalibrated Data\" \/>\n<meta property=\"og:description\" content=\"Results We build on 3DFA (Gkanatsios et al., 2025) as our base 3D policy and introduce , a learned orthogonal correction matrix that adapts the 3D RoPE positional encodings to account for unknown camera–robot extrinsics. Under fixed per-lab miscalibration — the realistic setting where each data source carries a consistent but unknown extrinsic error — […]\" \/>\n<meta property=\"og:url\" content=\"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/results\/\" \/>\n<meta property=\"og:site_name\" content=\"Learning 3D-VLAs with Noisy Miscalibrated Data\" \/>\n<meta property=\"article:modified_time\" content=\"2026-05-08T23:50:24+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/latex.codecogs.com\/svg.image?%5Cdelta%20M\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data1\" content=\"4 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf8\\\/results\\\/\",\"url\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf8\\\/results\\\/\",\"name\":\"Results & Conclusions - Learning 3D-VLAs with Noisy Miscalibrated Data\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf8\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf8\\\/results\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf8\\\/results\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/latex.codecogs.com\\\/svg.image?%5Cdelta%20M\",\"datePublished\":\"2026-05-08T03:58:25+00:00\",\"dateModified\":\"2026-05-08T23:50:24+00:00\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf8\\\/results\\\/#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf8\\\/results\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf8\\\/results\\\/#primaryimage\",\"url\":\"https:\\\/\\\/latex.codecogs.com\\\/svg.image?%5Cdelta%20M\",\"contentUrl\":\"https:\\\/\\\/latex.codecogs.com\\\/svg.image?%5Cdelta%20M\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf8\\\/results\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf8\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Results & Conclusions\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf8\\\/#website\",\"url\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf8\\\/\",\"name\":\"Learning 3D-VLAs with Noisy Miscalibrated Data\",\"description\":\"\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/mscvprojects.ri.cmu.edu\\\/2026teamf8\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"Results & Conclusions - Learning 3D-VLAs with Noisy Miscalibrated Data","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/results\/","og_locale":"en_US","og_type":"article","og_title":"Results & Conclusions - Learning 3D-VLAs with Noisy Miscalibrated Data","og_description":"Results We build on 3DFA (Gkanatsios et al., 2025) as our base 3D policy and introduce , a learned orthogonal correction matrix that adapts the 3D RoPE positional encodings to account for unknown camera–robot extrinsics. Under fixed per-lab miscalibration — the realistic setting where each data source carries a consistent but unknown extrinsic error — […]","og_url":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/results\/","og_site_name":"Learning 3D-VLAs with Noisy Miscalibrated Data","article_modified_time":"2026-05-08T23:50:24+00:00","og_image":[{"url":"https:\/\/latex.codecogs.com\/svg.image?%5Cdelta%20M","type":"","width":"","height":""}],"twitter_card":"summary_large_image","twitter_misc":{"Est. reading time":"4 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/results\/","url":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/results\/","name":"Results & Conclusions - Learning 3D-VLAs with Noisy Miscalibrated Data","isPartOf":{"@id":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/#website"},"primaryImageOfPage":{"@id":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/results\/#primaryimage"},"image":{"@id":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/results\/#primaryimage"},"thumbnailUrl":"https:\/\/latex.codecogs.com\/svg.image?%5Cdelta%20M","datePublished":"2026-05-08T03:58:25+00:00","dateModified":"2026-05-08T23:50:24+00:00","breadcrumb":{"@id":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/results\/#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/results\/"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/results\/#primaryimage","url":"https:\/\/latex.codecogs.com\/svg.image?%5Cdelta%20M","contentUrl":"https:\/\/latex.codecogs.com\/svg.image?%5Cdelta%20M"},{"@type":"BreadcrumbList","@id":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/results\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/"},{"@type":"ListItem","position":2,"name":"Results & Conclusions"}]},{"@type":"WebSite","@id":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/#website","url":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/","name":"Learning 3D-VLAs with Noisy Miscalibrated Data","description":"","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"}]}},"_links":{"self":[{"href":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/wp-json\/wp\/v2\/pages\/57","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/wp-json\/wp\/v2\/users\/281"}],"replies":[{"embeddable":true,"href":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/wp-json\/wp\/v2\/comments?post=57"}],"version-history":[{"count":6,"href":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/wp-json\/wp\/v2\/pages\/57\/revisions"}],"predecessor-version":[{"id":80,"href":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/wp-json\/wp\/v2\/pages\/57\/revisions\/80"}],"wp:attachment":[{"href":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf8\/wp-json\/wp\/v2\/media?parent=57"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}