{"article":{"annotation":{"article_id":"ntt-robot-physical-ai-shelf-loading-test-2026-11-11","available_at":"2026-11-12T03:09:10.045741Z","geography_revision":"ef6ee0010dda5fe7b240fd514749b43e214c57da73222d2e40ec83b276dd0977","id":"8bc3ed3d9bf8b87830f66d0dce96e374aa4cdc257c45085d5a46b5d7796108f4","input_sha256":"24fe7ac16cab4084cc6dce84fd8052f616a88334fe0be9c8848b05910dbcd197","limitations":["Rule-derived subjects are candidates, not verified editorial labels or calibrated model probabilities.","Coverage is distinct from event location. Unknown country/city/coordinates remain absent; no automatic geocoding.","Article geometries keep their stated precision and methodology; reproducing them is not independent coordinate verification."],"method":"controlled-rules-v1","places":[],"rules_sha256":"fb5ad3c27c91cb840f7a0a05f7f64a5cb26ad0374409ff9457f277c7092e22b4","schema_version":1,"subjects":[{"concept_id":"medtop:13000000","confidence":null,"confidence_method":null,"evidence":[{"field":"category+topics","language":"en","note":"Preserved editorial category/topic; broad candidate only","source_urls":["https://group.ntt/en/newsrelease/2026/11/11/261111b.html","https://openaccess.thecvf.com/content/ICCV2025/papers/Song_OCK_Unsupervised_Dynamic_Video_Prediction_with_Object-Centric_Kinematics_ICCV_2025_paper.pdf","https://openaccess.thecvf.com/content/ICCV2025/supplemental/Song_OCK_Unsupervised_Dynamic_ICCV_2025_supplemental.pdf"]}],"method":"controlled-rules-v1","rule_id":"technology-broad","status":"candidate"}],"taxonomy_revision":"1f2db31d811fca07e99b16d402c10e2cc2965c9df1c7b41d9be348b40a24e7fb","taxonomy_version":"IPTC Media Topics 2026-Q2"},"api_url":"https://newsverum.com/api/v1/articles/ntt-robot-physical-ai-shelf-loading-test-2026-11-11?lang=en","category":"technology","content":{"key_points":[],"paragraphs":["NTT said on 11 November that a physics-informed world model improved the success rate of a shelf-loading task performed by a small robot from 73.1 percent to 92.3 percent. The task simulated stacking items, and NTT described the result as a comparison with and without its technology. The company also reported a lower LPIPS score than two reference systems on the OBJ3D video benchmark. Its first planned field use is airport baggage handling, but field trials with robotics company Mujin are still ahead. The release does not describe a commercial deployment.","The system is designed to review candidate robot actions before movement. NTT says its model estimates object position, orientation, velocity and mass from video, then uses equations of motion to forecast how objects may move or interact. A proprietary physics control adapter connects those calculations to a video generation model. For each action proposed by existing control software, the system generates a possible future scene and scores outcomes such as a collision, a dropped object or a collapsing stack. It supplements an existing controller rather than establishing that robots can safely operate without supervision.","The shelf-loading result is notable, but the absolute gain is more informative than a broad claim that the robot became better at work. The reported success rate rose by 19.2 percentage points, calculated as 92.3 minus 73.1. Relative to the starting rate, that is about 26.3 percent, calculated as 19.2 divided by 73.1. Both calculations use rounded values in NTT's release. NTT does not state how many trials were run, how success was defined, whether the items or starting conditions varied, or how much results varied across runs. The release also provides no uncertainty interval.","NTT's Table 1 gives more detail on a separate prediction benchmark. It lists an LPIPS score multiplied by 100 of 7.81 for the NTT model, 7.92 for OCK, an academic model, and 21.90 for NVIDIA Cosmos Predict 2.5, a commercially available video generation model. Lower LPIPS is better. The NTT value is 0.11 lower than the OCK figure, or about 1.4 percent relative to 7.92. That arithmetic describes the displayed metric only. The release supplies no statistical interval or protocol comparison that would establish whether the small gap is robust across settings.","The benchmark itself needs context. A 2025 ICCV paper by Yeon-Ji Song and co-authors describes OBJ3D as a synthetic dataset with rigid objects, neutral grey scenes, 2,920 training sequences and 200 test sequences. It was designed to examine object geometry and interactions such as collisions. NTT says it evaluated its model on the public OBJ3D benchmark, but does not say whether it used the same data split, test protocol or training setup as the OCK paper. This result therefore measures video prediction in a controlled benchmark, not baggage handling in a real airport.","The technical distinction NTT emphasizes is explicit physical state. A video model can produce visually plausible frames while missing the forces or motion that determine whether a load will tip. NTT says its adapter represents mass and motion and applies equations of motion to generate candidate futures. That approach could help a robot reject a risky plan before acting, especially when each box or suitcase differs. The press release does not provide a latency figure, hardware configuration, memory demand, or safety test, so it cannot establish whether these calculations are fast and reliable enough for an operating warehouse.","NTT calls luggage and parcel handling an intended application and says it plans field trials with Mujin. It also says the technology is scheduled for display at NTT R&D Forum 2026. These are development plans, not evidence that airport operators have selected the system, installed it or measured labour savings. The release gives no customer, project budget, installation date, throughput target, error rate in live work, or price. The step from a simulated stacking task to mixed baggage at a belt or cargo area would require tests under changing lighting, occlusion, conveyor movement and human presence.","The announcement is best read as a research milestone with a concrete but narrow test result. The robot task suggests a possible operational use, while OBJ3D offers a separate check of predicted video quality. Neither result alone establishes general reliability across logistics, manufacturing or retail, the sectors NTT lists for possible expansion. Before deployment, decision-makers would need repeated trials, transparent definitions of task success, safety controls, performance across unseen objects and a comparison of the full system with current alternatives. NTT has described a route toward those trials, not their outcome."],"sections":[]},"corrections":[],"country":null,"data":{"blocks":[{"description":{"ar":"المعدلات التي أبلغت عنها NTT لمهمة تكديس محاكاة بروبوت صغير.","de":"Von NTT gemeldete Quoten für den simulierten Stapeltest mit einem kleinen Roboter.","en":"Rates NTT reports for its simulated stacking task with a small robot.","es":"Tasas informadas por NTT para su tarea simulada de apilamiento con un robot pequeño.","fr":"Taux rapportés par NTT pour sa tâche simulée d’empilement avec un petit robot.","pt":"Taxas informadas pela NTT para sua tarefa simulada de empilhamento com um robô pequeno.","tr":"NTT’nin küçük robotla yürüttüğü simüle istif görevinde bildirdiği oranlar."},"id":"shelf-loading-task-success-rate","items":[{"basis":"observed","label":{"ar":"معدل نجاح تحميل الرف بالروبوت الصغير دون تقنية NTT","de":"Erfolgsquote beim Regalstapeln ohne NTT-Technologie","en":"Small-robot shelf-loading success rate without NTT technology","es":"Tasa de éxito de carga en estantería sin tecnología NTT","fr":"Taux de réussite du chargement d’étagère sans la technologie NTT","pt":"Taxa de sucesso no carregamento da prateleira sem a tecnologia da NTT","tr":"NTT teknolojisi olmadan küçük robotlu raf yükleme başarı oranı"},"measure":{"aggregation":"rate","frequency":"irregular","geography":"world","id":"ntt-shelf-loading-success-rate-without","periodRole":"announcement-date","unitCode":"percent"},"period":"2026-11-11","unit":"percent","value":73.1,"valueQualifier":"exact"},{"basis":"observed","label":{"ar":"معدل نجاح تحميل الرف بالروبوت الصغير باستخدام تقنية NTT","de":"Erfolgsquote beim Regalstapeln mit NTT-Technologie","en":"Small-robot shelf-loading success rate with NTT technology","es":"Tasa de éxito de carga en estantería con tecnología NTT","fr":"Taux de réussite du chargement d’étagère avec la technologie NTT","pt":"Taxa de sucesso no carregamento da prateleira com a tecnologia da NTT","tr":"NTT teknolojisiyle küçük robotlu raf yükleme başarı oranı"},"measure":{"aggregation":"rate","frequency":"irregular","geography":"world","id":"ntt-shelf-loading-success-rate-with","periodRole":"announcement-date","unitCode":"percent"},"period":"2026-11-11","unit":"percent","value":92.3,"valueQualifier":"exact"}],"methodology":{"ar":"تورد NTT معدلي نجاح المهمة دون التقنية وباستخدامها، ولا تكشف عدد التجارب أو معيار النجاح أو التباين. 92.3 ناقص 73.1 يساوي 19.2 نقطة مئوية. وقسمة 19.2 على 73.1 تعطي زيادة نسبية تقارب 26.3 في المئة. تعتمد الحسابات على قيم الشركة المقربة وليست قياسات منفصلة.","de":"NTT meldet Erfolgsquoten mit und ohne Technologie, nennt aber weder Versuchszahl noch Erfolgskriterium oder Streuung. 92,3 minus 73,1 ergibt 19,2 Prozentpunkte. 19,2 geteilt durch 73,1 entspricht einem relativen Anstieg von rund 26,3 Prozent. Die Berechnungen beruhen auf gerundeten Unternehmensangaben und sind keine zusätzlichen Messungen.","en":"NTT reports task success rates without and with its technology; it does not disclose the number of trials, success criterion or variability. 92.3 minus 73.1 equals 19.2 percentage points. 19.2 divided by 73.1 is about a 26.3 percent relative increase. These calculations use rounded company values and are not separate measurements.","es":"NTT informa las tasas de éxito con y sin su tecnología, pero no publica el número de pruebas, el criterio de éxito ni la variabilidad. 92,3 menos 73,1 son 19,2 puntos porcentuales. 19,2 dividido por 73,1 equivale a un aumento relativo aproximado del 26,3 %. Los cálculos usan valores redondeados de la empresa y no son mediciones independientes.","fr":"NTT rapporte les taux de réussite avec et sans sa technologie, sans publier le nombre d’essais, le critère de succès ni la variabilité. 92,3 moins 73,1 donne 19,2 points de pourcentage. 19,2 divisé par 73,1 correspond à une hausse relative d’environ 26,3 %. Ces calculs utilisent des valeurs arrondies de l’entreprise et ne sont pas des mesures distinctes.","pt":"A NTT informa as taxas de sucesso com e sem sua tecnologia, mas não divulga o número de testes, o critério de sucesso nem a variabilidade. 92,3 menos 73,1 são 19,2 pontos percentuais. 19,2 dividido por 73,1 equivale a um aumento relativo aproximado de 26,3%. Os cálculos usam valores arredondados da empresa e não são medições separadas.","tr":"NTT, görev başarısı oranını teknoloji olmadan ve teknolojiyle bildirdi; deneme sayısı, başarı ölçütü ve değişkenlik açıklanmadı. 92,3 eksi 73,1 eşittir 19,2 yüzde puan. 19,2 bölü 73,1 yaklaşık yüzde 26,3 göreli artış verir. Bu hesaplar yuvarlanmış şirket değerlerine dayanır ve ayrı ölçüm değildir."},"sourceUrls":["https://group.ntt/en/newsrelease/2026/11/11/261111b.html"],"title":{"ar":"معدلات نجاح محاكاة تحميل الرف","de":"Erfolgsquoten der Regalladesimulation","en":"Shelf-loading simulation success rates","es":"Tasas de éxito de la simulación de carga en estantería","fr":"Taux de réussite de la simulation de chargement d’étagère","pt":"Taxas de sucesso da simulação de carregamento em prateleira","tr":"Raf yükleme simülasyonu başarı oranları"},"type":"metrics","updatedAt":"2026-11-12T06:05:54+03:00"},{"columns":[{"key":"model_name","label":{"ar":"النموذج","de":"Modell","en":"Model","es":"Modelo","fr":"Modèle","pt":"Modelo","tr":"Model adı"},"language":"en","type":"text"},{"key":"model_class","label":{"ar":"نوع النموذج","de":"Modelltyp","en":"Model type","es":"Tipo de modelo","fr":"Type de modèle","pt":"Tipo de modelo","tr":"Model türü"},"language":"en","type":"text"},{"key":"reported_score","label":{"ar":"درجة LPIPS المعلنة مضروبة في 100","de":"Gemeldeter LPIPS-x100-Wert","en":"Reported LPIPS x100 score","es":"Puntuación LPIPS x100 publicada","fr":"Score LPIPS x100 publié","pt":"Pontuação LPIPS x100 informada","tr":"Bildirilen LPIPS x100 skoru"},"language":"en","type":"text"}],"description":{"ar":"درجات النماذج الواردة في جدول NTT. والقيمة الأقل في LPIPS أفضل.","de":"Modellwerte aus der von NTT veröffentlichten Tabelle. Ein niedrigerer LPIPS-Wert ist besser.","en":"Model scores shown in NTT’s published table. Lower LPIPS is better.","es":"Puntuaciones de modelos incluidas en la tabla de NTT. Un LPIPS menor es mejor.","fr":"Scores des modèles présentés dans le tableau publié par NTT. Un LPIPS plus faible est préférable.","pt":"Pontuações dos modelos na tabela publicada pela NTT. Valores menores de LPIPS são melhores.","tr":"NTT’nin yayımladığı tabloda gösterilen model skorları. LPIPS’te düşük değer daha iyidir."},"id":"obj3d-video-benchmark-lpips","methodology":{"ar":"يعرض جدول NTT قيم LPIPS مضروبة في 100، والدرجة الأقل أفضل. وتحفظ خلايا الدرجات القيم المنشورة 7.81 و7.92 و21.90 مع بيان المقياس. ولا يقدم البيان فترات عدم يقين أو مقارنة مباشرة بين بروتوكولي NTT وOCK.","de":"NTTs Tabelle 1 multipliziert LPIPS mit 100; niedrigere Werte sind besser. Die Zellen bewahren die veröffentlichten Werte 7,81, 7,92 und 21,90 samt Skalenangabe. Die Mitteilung enthält weder Unsicherheitsintervalle noch einen direkten Protokollvergleich zwischen NTT und OCK.","en":"NTT’s Table 1 reports LPIPS multiplied by 100, with lower scores better. Score cells preserve the published values 7.81, 7.92 and 21.90 with the scale label. The release gives no uncertainty intervals or direct protocol comparison between NTT and OCK.","es":"La Tabla 1 de NTT multiplica LPIPS por 100 y un valor menor es mejor. Las celdas conservan las cifras publicadas 7,81, 7,92 y 21,90 con la escala indicada. El comunicado no ofrece intervalos de incertidumbre ni una comparación directa de los protocolos de NTT y OCK.","fr":"Le tableau 1 de NTT multiplie le LPIPS par 100, un score plus faible étant préférable. Les cellules conservent les valeurs publiées 7,81, 7,92 et 21,90 avec l’échelle indiquée. Le communiqué ne donne ni intervalle d’incertitude ni comparaison directe des protocoles NTT et OCK.","pt":"A Tabela 1 da NTT multiplica o LPIPS por 100, e valores menores são melhores. As células preservam os valores publicados 7,81, 7,92 e 21,90 com a escala indicada. O comunicado não fornece intervalos de incerteza nem comparação direta entre os protocolos da NTT e do OCK.","tr":"NTT’nin Tablo 1’i LPIPS değerlerini 100 ile çarpılmış biçimde verir ve düşük skor daha iyidir. Skor hücreleri yayımlanan 7,81, 7,92 ve 21,90 değerlerini birim açıklamasıyla aynen korur. Duyuru belirsizlik aralığı veya NTT ile OCK protokollerinin doğrudan karşılaştırmasını vermiyor."},"rowTranslations":[{"model_class":{"ar":"نموذج الشركة","de":"Unternehmensmodell","en":"Company model","es":"Modelo de la empresa","fr":"Modèle de l’entreprise","pt":"Modelo da empresa","tr":"Şirket modeli"},"model_name":{"ar":"نموذج NTT للعالم","de":"NTT-Weltmodell","en":"NTT world model","es":"modelo del mundo de NTT","fr":"modèle du monde de NTT","pt":"modelo de mundo da NTT","tr":"NTT dünya modeli"}},{"model_class":{"ar":"نموذج أكاديمي استشهدت به NTT","de":"Von NTT zitierter akademischer Ansatz","en":"Academic model cited by NTT","es":"Modelo académico citado por NTT","fr":"Modèle académique cité par NTT","pt":"Modelo acadêmico citado pela NTT","tr":"NTT’nin atıf yaptığı akademik model"},"model_name":{"ar":"OCK","de":"OCK","en":"OCK","es":"OCK","fr":"OCK","pt":"OCK","tr":"OCK"}},{"model_class":{"ar":"نموذج تجاري لتوليد الفيديو","de":"Kommerzielles Videogenerierungsmodell","en":"Commercial video generation model","es":"Modelo comercial de generación de vídeo","fr":"Modèle commercial de génération vidéo","pt":"Modelo comercial de geração de vídeo","tr":"Ticari video üretim modeli"},"model_name":{"ar":"NVIDIA Cosmos Predict 2.5","de":"NVIDIA Cosmos Predict 2.5","en":"NVIDIA Cosmos Predict 2.5","es":"NVIDIA Cosmos Predict 2.5","fr":"NVIDIA Cosmos Predict 2.5","pt":"NVIDIA Cosmos Predict 2.5","tr":"NVIDIA Cosmos Predict 2.5"}}],"rows":[{"model_class":"Company model","model_name":"NTT world model","reported_score":"7.81 LPIPS x100"},{"model_class":"Academic model cited by NTT","model_name":"OCK","reported_score":"7.92 LPIPS x100"},{"model_class":"Commercial video generation model","model_name":"NVIDIA Cosmos Predict 2.5","reported_score":"21.90 LPIPS x100"}],"sourceUrls":["https://group.ntt/en/newsrelease/2026/11/11/261111b.html","https://openaccess.thecvf.com/content/ICCV2025/papers/Song_OCK_Unsupervised_Dynamic_Video_Prediction_with_Object-Centric_Kinematics_ICCV_2025_paper.pdf"],"title":{"ar":"مقارنة LPIPS على اختبار الفيديو OBJ3D","de":"LPIPS-Vergleich im OBJ3D-Videobenchmark","en":"OBJ3D video benchmark LPIPS comparison","es":"Comparación LPIPS en el benchmark de vídeo OBJ3D","fr":"Comparaison LPIPS sur le benchmark vidéo OBJ3D","pt":"Comparação LPIPS no benchmark de vídeo OBJ3D","tr":"OBJ3D video benchmark LPIPS karşılaştırması"},"type":"table","updatedAt":"2026-11-12T06:05:54+03:00"}],"version":1},"data_downloads":[{"formats":{"csv":"https://newsverum.com/api/v1/articles/ntt-robot-physical-ai-shelf-loading-test-2026-11-11/data/shelf-loading-task-success-rate?lang=en&format=csv","json":"https://newsverum.com/api/v1/articles/ntt-robot-physical-ai-shelf-loading-test-2026-11-11/data/shelf-loading-task-success-rate?lang=en&format=json"},"id":"shelf-loading-task-success-rate"},{"formats":{"csv":"https://newsverum.com/api/v1/articles/ntt-robot-physical-ai-shelf-loading-test-2026-11-11/data/obj3d-video-benchmark-lpips?lang=en&format=csv","json":"https://newsverum.com/api/v1/articles/ntt-robot-physical-ai-shelf-loading-test-2026-11-11/data/obj3d-video-benchmark-lpips?lang=en&format=json"},"id":"obj3d-video-benchmark-lpips"}],"data_types":["metrics","table"],"editorial_illustration":{"alt":"Conceptual illustration of technology, processors and connected information.","asset_id":"newsverum-v1-technology","avif_variants":[{"height":320,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-480.avif","width":480},{"height":480,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-720.avif","width":720},{"height":640,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-960.avif","width":960},{"height":800,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-1200.avif","width":1200},{"height":1024,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-1536.avif","width":1536}],"caption":null,"credit":"Newsverum","documentary":false,"height":1024,"kind":"symbolic-editorial-illustration","language":"en","scope":"topic","url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-1536.webp","variants":[{"height":320,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-480.webp","width":480},{"height":640,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-960.webp","width":960},{"height":1024,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-1536.webp","width":1536}],"width":1536},"event_id":"event-422c5316-ef7a-5c6d-af93-a203aa3e8195","event_period":"2026-11-11","event_period_basis":"event-or-reference-period","id":"ntt-robot-physical-ai-shelf-loading-test-2026-11-11","image":null,"importance":0.6,"language":"en","modified_at":"2026-11-12T03:09:09+00:00","publication_basis":"publication-record","published_at":"2026-11-12T03:09:09+00:00","reading_minutes":4,"region":"world","relations":[],"section_illustration":{"alt":"Conceptual illustration of technology, processors and connected information.","asset_id":"newsverum-v1-technology","avif_variants":[{"height":320,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-480.avif","width":480},{"height":480,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-720.avif","width":720},{"height":640,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-960.avif","width":960},{"height":800,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-1200.avif","width":1200},{"height":1024,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-1536.avif","width":1536}],"caption":null,"credit":"Newsverum","documentary":false,"height":1024,"kind":"symbolic-editorial-illustration","language":"en","scope":"section","url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-1536.webp","variants":[{"height":320,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-480.webp","width":480},{"height":640,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-960.webp","width":960},{"height":1024,"url":"https://newsverum.com/assets/newsverum/illustrations/v1/technology-1536.webp","width":1536}],"width":1536},"slug":"ntt-robot-physical-ai-shelf-loading-test-2026-11-11","sources":[{"checkedAt":"2026-11-12","name":"NTT, NTT Develops Physical AI Technology That Predicts the Future Before Robots Move to Prevent Failures","publishedAt":"2026-11-11","url":"https://group.ntt/en/newsrelease/2026/11/11/261111b.html"},{"checkedAt":"2026-11-12","name":"Computer Vision Foundation, OCK: Unsupervised Dynamic Video Prediction with Object-Centric Kinematics, ICCV 2025 paper","url":"https://openaccess.thecvf.com/content/ICCV2025/papers/Song_OCK_Unsupervised_Dynamic_Video_Prediction_with_Object-Centric_Kinematics_ICCV_2025_paper.pdf"},{"checkedAt":"2026-11-12","name":"Computer Vision Foundation, OCK supplementary material","url":"https://openaccess.thecvf.com/content/ICCV2025/supplemental/Song_OCK_Unsupervised_Dynamic_ICCV_2025_supplemental.pdf"}],"summary":"NTT says a physics-informed model raised success in a small-robot shelf-loading simulation from 73.1% to 92.3%. Field trials with Mujin remain planned, and the release does not report a commercial deployment.","text":"NTT says a physics-informed model raised success in a small-robot shelf-loading simulation from 73.1% to 92.3%. Field trials with Mujin remain planned, and the release does not report a commercial deployment.\n\nNTT said on 11 November that a physics-informed world model improved the success rate of a shelf-loading task performed by a small robot from 73.1 percent to 92.3 percent. The task simulated stacking items, and NTT described the result as a comparison with and without its technology. The company also reported a lower LPIPS score than two reference systems on the OBJ3D video benchmark. Its first planned field use is airport baggage handling, but field trials with robotics company Mujin are still ahead. The release does not describe a commercial deployment.\n\nThe system is designed to review candidate robot actions before movement. NTT says its model estimates object position, orientation, velocity and mass from video, then uses equations of motion to forecast how objects may move or interact. A proprietary physics control adapter connects those calculations to a video generation model. For each action proposed by existing control software, the system generates a possible future scene and scores outcomes such as a collision, a dropped object or a collapsing stack. It supplements an existing controller rather than establishing that robots can safely operate without supervision.\n\nThe shelf-loading result is notable, but the absolute gain is more informative than a broad claim that the robot became better at work. The reported success rate rose by 19.2 percentage points, calculated as 92.3 minus 73.1. Relative to the starting rate, that is about 26.3 percent, calculated as 19.2 divided by 73.1. Both calculations use rounded values in NTT's release. NTT does not state how many trials were run, how success was defined, whether the items or starting conditions varied, or how much results varied across runs. The release also provides no uncertainty interval.\n\nNTT's Table 1 gives more detail on a separate prediction benchmark. It lists an LPIPS score multiplied by 100 of 7.81 for the NTT model, 7.92 for OCK, an academic model, and 21.90 for NVIDIA Cosmos Predict 2.5, a commercially available video generation model. Lower LPIPS is better. The NTT value is 0.11 lower than the OCK figure, or about 1.4 percent relative to 7.92. That arithmetic describes the displayed metric only. The release supplies no statistical interval or protocol comparison that would establish whether the small gap is robust across settings.\n\nThe benchmark itself needs context. A 2025 ICCV paper by Yeon-Ji Song and co-authors describes OBJ3D as a synthetic dataset with rigid objects, neutral grey scenes, 2,920 training sequences and 200 test sequences. It was designed to examine object geometry and interactions such as collisions. NTT says it evaluated its model on the public OBJ3D benchmark, but does not say whether it used the same data split, test protocol or training setup as the OCK paper. This result therefore measures video prediction in a controlled benchmark, not baggage handling in a real airport.\n\nThe technical distinction NTT emphasizes is explicit physical state. A video model can produce visually plausible frames while missing the forces or motion that determine whether a load will tip. NTT says its adapter represents mass and motion and applies equations of motion to generate candidate futures. That approach could help a robot reject a risky plan before acting, especially when each box or suitcase differs. The press release does not provide a latency figure, hardware configuration, memory demand, or safety test, so it cannot establish whether these calculations are fast and reliable enough for an operating warehouse.\n\nNTT calls luggage and parcel handling an intended application and says it plans field trials with Mujin. It also says the technology is scheduled for display at NTT R&D Forum 2026. These are development plans, not evidence that airport operators have selected the system, installed it or measured labour savings. The release gives no customer, project budget, installation date, throughput target, error rate in live work, or price. The step from a simulated stacking task to mixed baggage at a belt or cargo area would require tests under changing lighting, occlusion, conveyor movement and human presence.\n\nThe announcement is best read as a research milestone with a concrete but narrow test result. The robot task suggests a possible operational use, while OBJ3D offers a separate check of predicted video quality. Neither result alone establishes general reliability across logistics, manufacturing or retail, the sectors NTT lists for possible expansion. Before deployment, decision-makers would need repeated trials, transparent definitions of task success, safety controls, performance across unseen objects and a comparison of the full system with current alternatives. NTT has described a route toward those trials, not their outcome.","title":"NTT reports 92.3% success in robot shelf-loading test; field trials remain ahead","topics":[],"translations":[{"api_url":"https://newsverum.com/api/v1/articles/ntt-robot-physical-ai-shelf-loading-test-2026-11-11?lang=tr","language":"tr","url":"https://newsverum.com/tr/news/ntt-robot-physical-ai-shelf-loading-test-2026-11-11"},{"api_url":"https://newsverum.com/api/v1/articles/ntt-robot-physical-ai-shelf-loading-test-2026-11-11?lang=en","language":"en","url":"https://newsverum.com/en/news/ntt-robot-physical-ai-shelf-loading-test-2026-11-11"},{"api_url":"https://newsverum.com/api/v1/articles/ntt-robot-physical-ai-shelf-loading-test-2026-11-11?lang=ar","language":"ar","url":"https://newsverum.com/ar/news/ntt-robot-physical-ai-shelf-loading-test-2026-11-11"},{"api_url":"https://newsverum.com/api/v1/articles/ntt-robot-physical-ai-shelf-loading-test-2026-11-11?lang=fr","language":"fr","url":"https://newsverum.com/fr/news/ntt-robot-physical-ai-shelf-loading-test-2026-11-11"},{"api_url":"https://newsverum.com/api/v1/articles/ntt-robot-physical-ai-shelf-loading-test-2026-11-11?lang=es","language":"es","url":"https://newsverum.com/es/news/ntt-robot-physical-ai-shelf-loading-test-2026-11-11"},{"api_url":"https://newsverum.com/api/v1/articles/ntt-robot-physical-ai-shelf-loading-test-2026-11-11?lang=de","language":"de","url":"https://newsverum.com/de/news/ntt-robot-physical-ai-shelf-loading-test-2026-11-11"},{"api_url":"https://newsverum.com/api/v1/articles/ntt-robot-physical-ai-shelf-loading-test-2026-11-11?lang=pt","language":"pt","url":"https://newsverum.com/pt/news/ntt-robot-physical-ai-shelf-loading-test-2026-11-11"}],"url":"https://newsverum.com/en/news/ntt-robot-physical-ai-shelf-loading-test-2026-11-11"},"version":"1"}
