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 44 of 59

Mar 30, 2026Selective Memory as Geometry: Coreset Methods for Continual Medical Reinforcement Learning on Edge Devices Introduction The deployment of artificial intelligence in clinical radiology faces a fundamental tension. While deep learning models are typically trained on centralized, high-performance computing clusters with abundan Mar 30, 2026 · TREvent Kameralar ve FAITH: Otonom Navigasyonda Hesaplama Verimliliğinin Yeni ParadigmasıOtonom robot navigasyonunun temel bileşenlerinden biri, aracın hareket yönünün (course) gerçek zamanlı olarak tahmin edilmesidir. Özellikle mikro hava araçları (MAV) gibi kaynak kısıtlı platformlar söz konusu olduğunda, Mar 30, 2026The Focus of Expansion is No Longer a Blind Spot: Event-Based Binary Search for Visual NavigationAutonomous navigation for micro air vehicles (MAVs) demands robust course estimation capable of operating under severe computational constraints. The Focus of Expansion (FOE), the point in the visual field where optic fl Mar 30, 2026 · TRGörsel Dil Modelleri Gerçekten Vision Transformer'lara İhtiyaç Duyuyor mu? Durum Uzay Modellerinin GörselGörsel Dil Modelleri Gerçekten Vision Transformer'lara İhtiyaç Duyuyor mu? Durum Uzay Modellerinin Görsel Mar 30, 2026Do VLMs Need Vision Transformers? A Controlled Analysis of State Space Models as Visual Backbones The Tyranny of the Transformer in Multimodal Systems Vision-Language Models have settled into a comfortable architectural monoculture. The standard recipe is deceptively simple: take a Vision Transformer pretrained on I Mar 30, 2026 · TR Mar 30, 2026Function-Preserving Expansion and the Unsolved Composition Problem in Continual LearningCatastrophic forgetting remains a fundamental obstacle in the deployment of large language models. When practitioners adapt a pre-trained model to specialized domains, whether medical diagnosis or scientific reasoning, s Mar 29, 2026 · TRCOLD-Steer: Büyük Dil Modellerini Bağlam İçi Tek Adımlık Öğrenme Dinamikleriyle Yönlendirme Giriş: Davranışsal Kontrolün Verimlilik Çıkmazı Büyük dil modellerinin (LLM) davranışlarını değiştirmek için geliştirilen mevcut yöntemler, temel bir ikilemle karşı karşıya. Bir yanda, sınırlı sayıda örnek (sample) ile Mar 29, 2026Simulating Fine-Tuning at Inference Time: A Deep Dive into COLD-Steer Introduction The field of language model alignment faces a persistent tension between data efficiency and behavioral precision. When we wish to steer a large language model away from generating harmful stereotypes, for Mar 29, 2026 · TRDüşük Yoğunluklu Bölgelerde Difüzyon Modellerini Stabilize Etmek: AAPB ve Çoklu Anchor Perspektifi Giriş Metinden görüntüye üretim modellerindeki son gelişmeler, difüzyon tabanlı mimarilerin fotoğrafrealistik ve anlamsal olarak zengin görseller oluşturma kapasitesini önemli ölçüde artırdı. Ancak bu modellerin temelin Mar 29, 2026Navigating the Long Tail: Adaptive Prompt Blending for Low-Density Diffusion Sampling Introduction Text-to-image diffusion models have achieved remarkable fidelity in generating photorealistic imagery across diverse visual domains. Yet beneath this veneer of capability lies a persistent structural weakne Mar 29, 2026 · TRSadece Son Metin Kaldiginda: Cok-Ajanli Sistemlerde Gizli Yurutme Izleme ve IET Giris Yapay zeka ajanlarinin kurumsal altyapilara entegrasyonu hizlaniyor. Gartner'in 2025 raporlarina gore, 2026 yilina kadar kurumsal uygulamalarin %40'i gorev ozellikli AI ajanlarini barindiracak. Ancak bu buyumenin Mar 29, 2026Intrinsic Attribution: How Steganographic Signatures Are Reshaping Multi-Agent AccountabilityWhen enterprise AI systems fail, accountability often evaporates into a maze of delegation chains and lost metadata. Industry projections suggest that 40% of enterprise applications will incorporate task-specific AI agen Mar 29, 2026 · TRLLM Tabanlı Kod Üretimi Değerlendirmesinde Standartlaşma Çabaları: Gerekli Bir Çerçeve mi, Erken Bir Kısıtlama mı? Giriş Yazılım mühendisliği alanında büyük dil modellerinin (LLM) kod üretimi kapasitesi, son yılların en tartışmalı ve hızlı evrim geçiren konularından biri haline geldi. GitHub Copilot, CodeT5, CodeLlama ve GPT-4 gibi Mar 29, 2026The Impossibility of Stable Metrics in a Moving Target: Rethinking LLM Code Generation Evaluation Introduction The software engineering research community currently faces a methodological dilemma that strikes at the heart of empirical validity. On one hand, the proliferation of large language models for code generat Mar 29, 2026 · TRGörsel Geri Bildirimle Açık Sözlüklü 3D Sahne Üretimi: SceneAssistant ve Ajan Tabanlı Yaklaşımın Yükselişi Giriş 3D sahne üretimi, film yapımından oyun geliştirmeye, robotik simülasyonlardan somutlaştırılmış yapay zeka (embodied AI) araştırmalarına kadar geniş bir yelpazede kritik öneme sahip. Ancak metin tabanlı 3D sahne ol Mar 29, 2026From Monolithic Generation to Agentic Composition: The Case for Visual Feedback in 3D Scene SynthesisThe generation of complex 3D environments from natural language descriptions has long represented a frontier problem at the intersection of computer vision, graphics, and artificial intelligence. While recent years have Mar 29, 2026 · TRR-C2: Çok Modlu Akıl Yürütmede Döngüsel Tutarlılık ile Takviye Öğrenmesi Giriş Çok modlu büyük dil modelleri (Multimodal Large Language Models, MLLM'ler), metin ve görsel verileri birlikte işleyerek yapay zekada yeni ufuklar açtı. Web sayfalarını anlama, grafikleri yorumlama veya karmaşık gö