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 36 of 59

Apr 10, 2026Emotional Framing in LLMs: Weak Signals and Context-Dependent Modulation Introduction: The Affective Turn in Human-LLM Interaction Human communication is inseparable from affect. When we ask for help, negotiate deadlines, or seek medical advice, our emotional state leaks into phrasing, tone, Apr 10, 2026CortexCompile and the Neuroscience of Code Generation: Separating Metaphor from MechanismThe pursuit of more capable automated programming systems has increasingly turned toward biological inspiration, specifically the modular architecture of the human brain. The recent paper, CortexCompile: Harnessing Corti Apr 10, 2026 Apr 10, 2026Adaptive Curriculum Design Through Contextual Bandits: A Technical Analysis of Personalized Learning SystemsThe scalability of personalized education remains one of the most persistent challenges in learning technology. While one-to-one human tutoring demonstrates consistent efficacy, the resource constraints of traditional cl Apr 10, 2026Navigating the Skies with Intelligence: A Critical Look at Reinforcement Learning for Autonomous UAVsThe proliferation of unmanned aerial vehicles across logistics, surveillance, and emergency response has precipitated a critical need for navigation systems that transcend pre-programmed waypoint following. While GPS gui Apr 10, 2026S2D2 and the Architecture of Self-Verifying Diffusion ModelsThe persistent tension between autoregressive and diffusion based language modeling has catalyzed significant architectural innovation over the past two years. While autoregressive models dominate reasoning intensive ben Apr 10, 2026 Apr 10, 2026Separable Neural Architectures: Unifying Prediction and Generation Through Factorised RepresentationsThe dominance of monolithic neural architectures, from convolutional networks to Transformers, has defined the trajectory of modern artificial intelligence. These models treat the systems they approximate as undifferenti Apr 10, 2026Emergent Stability: How Minimal Agents Learn Without Forgetting Introduction The stability plasticity dilemma has long haunted the field of continual learning. As artificial agents are deployed in environments where data streams continuously and retraining from scratch is impossible Apr 10, 2026Ruka-v2 and the Economics of Dexterity: Why Wrists Matter More Than We ThoughtThe robotics community has made remarkable strides in learning algorithms, yet hardware remains the stubborn bottleneck preventing widespread replication of human-level manipulation. While a graduate student can download Apr 10, 2026The Long Horizon Problem: Evaluating AI Assistants Through Cognitive SimulationThe vision of a universal digital assistant, capable of anticipating needs across months or years of interaction, remains the defining aspiration of artificial intelligence research. Yet as large language models (LLMs) d Apr 9, 2026Beyond Vision-Language Alignment: Why Pure Visual Self-Supervision Wins in Medical Imaging Foundation ModelsThe race to develop foundation models for radiology has largely followed the trajectory of natural language processing and general computer vision. The prevailing assumption has been that the path to clinically capable a Apr 9, 2026Rethinking Reinforcement Learning Environments: From Petri Dishes to Living EcosystemsThe field of reinforcement learning has made remarkable progress in recent years, with agents mastering complex games, controlling robotic systems, and even discovering new strategies in scientific domains. Yet despite t Apr 9, 2026Out of Sight, Out of Mind: Why Video World Models Struggle with Unobserved State EvolutionThe fundamental promise of video world models is compelling: generate coherent, evolving visual worlds that behave according to physical laws. But a critical question remains largely unexplored: do these models truly und Apr 9, 2026Generative AI and the Irreducible Human Core of Data ScienceThe rise of generative artificial intelligence has sparked intense debate about the future of data science education and practice. While many worry that AI will automate away the need for human data scientists, Nathan Ta Apr 9, 2026Learning to Present: How Inverse Specification Rewards Transform Automated Slide GenerationThe challenge of automated presentation generation sits at a fascinating intersection of artificial intelligence capabilities. While we've made remarkable progress in language generation and visual design, creating coher Apr 9, 2026Exploiting the Blind Spots of LLM Safety: How Encoding Attacks Bypass System Instruction ProtectionLarge Language Models have become the backbone of modern AI agents, from customer service chatbots to sophisticated enterprise automation systems. These agents rely on system instructions to define their behavior, safety Apr 9, 2026Non-stationary Reinforcement Learning with Function Approximation: A Step Forward with Practical LimitationsThe intersection of non-stationary environments and function approximation represents one of the most challenging frontiers in reinforcement learning theory. While most theoretical work assumes static environments where