Blog
Notes on teaching robots new tricks
How the studio works, why tiny robots come first, what the checks catch, and how a policy gets from simulation to a servo.
workflowopen-duck-miniSep 29, 2026 · 6 min readFrom a sentence to a walking duck: how kualia turns a plain-language brief into a training taskA walk through the whole kualia workflow, from a one-sentence brief to an ONNX policy, using the Open Duck Mini and a small ramp.Read the post 
Sep 27, 2026 · 5 minWhy we start with tiny robotsDesk-sized robots train on CPU in minutes to hours, cost little to own and cross the sim-to-real gap without heroics. That is why kualia's shelf starts with them.
Sep 25, 2026 · 6 min13 checks before a single training stepalia writes the training task, but it does not get to grade its own work. Every task passes 13 sandbox checks before training can start. Here is what each one catches.
Sep 23, 2026 · 5 minFrom simulation to servo: running your ONNX policy on a real SO-101What is inside a kualia ONNX export, how to read its metadata, a minimal onnxruntime loop for the SO-101, and the sim-to-real checklist before you power the servos.
Sep 21, 2026 · 5 minLet your coding agent train the robotkualia is a remote MCP server. Point Claude Code, Cursor or Codex at it and your agent can pick a robot, build the task, train it and fetch the ONNX policy from your editor.Research digest
Short reads on new robotics and AI papers, written by our research assistant and reviewed before publishing. 1,060 so far.
Page 45 of 59
Mar 29, 2026Cycle-Consistent Reinforcement Learning: Treating Multimodal Inconsistency as a Supervisory Signal Introduction Consider a multimodal large language model (MLLM) presented with a webpage. When shown a screenshot of the page, it identifies one product as the correct answer to a query. When presented with the raw HTML Mar 29, 2026 · TROtonom İnsansız Hava Araçlarında Şifreli Pekiştirmeli Öğrenme: Gerçek Zamanlı Navigasyonun Güvenlik Mimarisi Giriş Otonom İnsansız Hava Araçları (UAV'ler), sınır güvenliğinden afet müdahalesine, lojistik dağıtımından kritik altyapı denetimine kadar geniş bir yelpazede operasyonel vazgeçilmezlik haline geldi. Bu araçların navig Mar 29, 2026Encrypted Eyes in the Sky: Fully Homomorphic Encryption for Autonomous UAV NavigationAutonomous Unmanned Aerial Vehicles have evolved from experimental platforms into critical infrastructure components deployed across defense, disaster response, and commercial logistics. These systems increasingly rely o Mar 29, 2026 · TRFiziksel Olabilirlik ve Metin Uyumu Arasında Denge: PhysMoDPO ile Humanoid Hareket Optimizasyonu Giriş Diffusion modellerinin text-conditioned motion generation alanındaki ilerlemesi, büyük ölçekli insan hareketi verileri üzerinde eğitilen ve kinematik uzayda çalışan sistemlerle şekilleniyor. Bu modeller, metin açı Mar 29, 2026 Mar 29, 2026 · TRV0.5: Seyrek RL Rollout'lar için Genelleştirilmiş Değer Modeli ve Uyarlanabilir Bütçe Yönetimi Giriş Büyük Dil Modellerinin (LLM) post-training aşamasında karmaşık akıl yürütme yeteneklerini geliştirmek için kullanılan Doğrulanabilir Ödüllerle Pekiştirmeli Öğrenme (RLVR), son dönemde standart bir paradigma haline Mar 29, 2026 Mar 29, 2026 · TR Mar 29, 2026Traces Propagation: Rethinking Spatial Credit Assignment in Spiking Neural Networks Through Local Contrastive LearningSpiking Neural Networks represent a fundamental departure from conventional artificial neural networks. By employing discrete spike events rather than continuous activations, SNNs offer a computational framework that mir Mar 29, 2026 · TR Mar 29, 2026HiAR: Rethinking Temporal Continuity Through Hierarchical Denoising in Autoregressive Video Generation Introduction The quest for open-ended video generation represents one of the final frontiers in generative modeling. While current diffusion transformers can produce visually stunning clips of a few seconds, extending t Mar 29, 2026 · TRDinamik Videolarda 3D Geometrik Tutarlılığı Ölçmek: SGC Metriği ve Piksel Merkezli Değerlendirme Krizinin Aşılması Giriş Son dönemde üretken video modelleri, metin tabanlı açıklamalardan gerçekçi görüntü dizileri oluşturma konusunda kayda değer ilerlemeler kaydetti. CogVideo, VideoPoet ve benzeri sistemler, benchmark süreçlerinde yü Mar 29, 2026Beyond Pixels: Evaluating 3D Spatial Geometric Consistency in Generated VideosThe current generation of video synthesis models produces imagery of striking visual fidelity. Text to video systems from industry leaders Mar 28, 2026 · TR Mar 28, 2026The Certification Gap in Adaptive Safety Systems: When Online Learning Meets Safety Assurance Introduction Real time safety assessment of dynamic systems operating in non stationary environments represents one of the most pressing challenges in modern control theory and machine learning. As industrial automation Mar 28, 2026 · TR Mar 28, 2026Colon-Bench and the Illusion of Solved Scarcity: Rethinking Annotation Pipelines and Clinical Reasoning in Medical AI Introduction Colon cancer remains the second leading cause of cancer mortality globally, a statistic made more tragic by the preventability of most cases through early screening and lesion removal. Yet the colonoscopy p Mar 28, 2026 · TRBatch Normalization'un Gizli Geometrisi: Veri Uyumlu Bölümlemeler ve Akıllı Başlatma Giriş Derin öğrenme literatüründe neredeyse herhangi bir state-of-the-art modelin mimarisine baktığımızda, batch normalization (BN) katmanlarıyla karşılaşırız. Ioffe ve Szegedy'nin 2015 yılında ortaya attığı bu teknik,