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 49 of 59

Mar 26, 2026Structured Oversight: How OS-Themis Redefines Reward Modeling for Complex TasksThe deployment of autonomous GUI agents in stochastic environments presents a fundamental challenge for reinforcement learning. While recent Vision-Language Models demonstrate impressive capabilities in digital navigatio Mar 26, 2026 · TRTıbbi Görüntülemede Yapısal Tutarlılık: ARIADNE ve Betti Sayıları ile Tercih Tabanlı Öğrenme Giriş Koroner arter hastalığı (Coronary Artery Disease, CAD) küresel sağlık yükünü belirleyen en kritik kardiyovasküler patolojilerden biri olmaya devam ediyor. İnvaziv X-Ray Koroner Anjiyografi (XCA) hem tanı hem de Pe Mar 26, 2026Beyond Pixel Perfection: Topological Preference Optimization in Coronary AngiographyCoronary artery disease remains the leading cause of mortality worldwide, and invasive X-ray coronary angiography continues to serve as the definitive diagnostic modality for guiding percutaneous interventions. Despite t Mar 26, 2026 · TRÇok Dilli Dil Modellerinde Ölçekleme Yasaları ve Oyun Teorisi: ShapleyLaw Analizi Giriş Çok dilli büyük dil modellerinin (LLM) ön eğitiminde (pretraining), farklı dillerin eğitim külliyatındaki oranlarını belirlemek, modelin genel performansını doğrudan etkileyen kritik bir optimizasyon problemidir. Mar 26, 2026Quantifying Multilingual Synergy: When Game Theory Meets Scaling Laws Introduction The multilingual pretraining of large language models presents a fundamental allocation problem. Given a fixed compute budget and a collection of linguistic corpora of varying sizes and qualities, how shoul Mar 26, 2026 · TRTopoBench: Büyük Dil Modellerinin Topolojik Muhakeme Çıkmazları Giriş Son yıllarda büyük dil modelleri (LLM'ler) cebirsel problemlerden sembolik mantığa uzanan geniş bir yelpazede etkileyici performanslar sergiledi. Ancak bu başarı hikayesinin önemli bir kör noktası var: topolojik v Mar 26, 2026Topological Blind Spots: Constraint Extraction as the Bottleneck in LLM Spatial Reasoning Introduction Despite impressive advances in algebraic and symbolic reasoning, large language models remain remarkably fragile when confronted with tasks requiring global spatial invariants. The recent work TopoBench: Be Mar 26, 2026 · TRTRiMS ve Minimal Yeterli Uzunluk: Akılcıl Çıkarımın Verimlilik Sınırını Keşfetmek Giriş Son dönemde OpenAI o1, DeepSeek-R1 ve Qwen gibi büyük dil modelleri, karmaşık problemleri çözerken uzun düşünce zincirleri (chain-of-thought, CoT) kullanarak çıkarım yeteneklerinde önemli ilerlemeler kaydetti. Anc Mar 26, 2026The Minimal Sufficient Length: Rethinking Efficiency in Chain-of-Thought Reasoning Introduction The emergence of reasoning-oriented large language models, exemplified by OpenAI's o1 series, DeepSeek-R1, and Qwen, has fundamentally altered our understanding of artificial intelligence capabilities. Thes Mar 26, 2026 · TRPotential Tabanlı Ödül Şekillendirmede Etkinliği Artırma: Başlangıç Değerleri ve Sürekli Fonksiyonların Gizli KısıtlamalarıPekiştirmeli öğrenme (reinforcement learning) sistemlerinin en temel zorluklarından biri, seyrek (sparse) ve bilgilendirici olmayan ödül sinyalleriyle karşılaştıklarında kredi atama (credit assignment) problemini çözmeye Mar 26, 2026Beyond Policy Invariance: Rethinking Potential-Based Reward Shaping in Deep Reinforcement Learning Introduction Potential-based reward shaping (PBRS) has long stood as one of the most theoretically elegant methods for incorporating domain knowledge into reinforcement learning. Since Ng, Harada, and Russell establishe Mar 26, 2026 · TRMakineyi Tespit Etmek: AI Metin Dedektörlerinin Kapsamlı Değerlendirmesi ve Perplexity Paradoksu Giriş Büyük dil modellerinin (large language models, LLM) metin üretim kapasitesindeki meteorik yükseliş, dijital içeriklerin kaynağını belirleme ihtiyacını kaçınılmaz kıldı. Akademik yayınlardan sosyal medya gönderiler Mar 26, 2026The False Certainty of AI Detection: Why Perplexity Inversion and Domain Shift Break Our Current Tools The Benchmark We Needed As large language models proliferate across academic, journalistic, and creative domains, the detection of machine generated text has become less a technical curiosity and more a critical infrast Mar 25, 2026 · TRUzun Bağlamda Koşullu Bellek Erişimi: L2A ile Dinamik Attention Mekanizmaları Giriş: Attention'un Gizli Maliyeti ve L2A'nın Temel Fikri Büyük Dil Modellerinin (LLM) bağlam penceresi son yıllarda dramatik biçimde genişledi. Ancak bu genişleme, Attention mekanizmasının doğası gereği taşıdığı kuadra Mar 25, 2026Learning When to Attend: Rethinking Memory Access in Long-Context Language ModelsLong-context modeling remains one of the persistent bottlenecks in large language model deployment. While frontier models support hundreds of thousands of tokens, extending base models beyond their pretraining context le Mar 25, 2026 · TRFizik Bilgili Sinir Ağlarında Hata Sertifikasyonu: Rezidü Minimizasyonundan Çözüm Garantilerine Giriş Kısmi diferansiyel denklemlerin (PDE) sayısal çözümü, klasik hesaplamalı matematiğin temel taşlarından biridir. Sonlu fark, sonlu elemanlar ve spektral yöntemler gibi geleneksel yaklaşımlar, ağ (mesh) inceltme yol Mar 25, 2026From Residuals to Guarantees: Compactness and Certified Error Bounds in Neural PDE Solvers Introduction Physics-informed neural networks have emerged as a flexible alternative to classical numerical methods for solving partial differential equations. By embedding governing equations directly into the loss fun Mar 25, 2026 · TR