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 21 of 59

May 3, 2026The Protocol Problem: Why MLLM Classification Metrics Are Fundamentally UnstableRecent literature presents contradictory portraits of Multimodal Large Language Model (MLLM) capabilities. Some studies report that models like GPT-4V or LLaVA lag far behind supervised vision models on standard classifi May 3, 2026The Self-Driving Portfolio: Agentic Architecture and the Elevation of Human Judgment in Asset Management Introduction The most persistent myth in quantitative finance holds that investment performance suffers primarily from insufficient data or inadequate model sophistication. In their working paper, "The Self Driving Port May 3, 2026Learning to Shape Rewards by Inferring Environment Dynamics Introduction Sparse reward signals remain one of the most persistent obstacles in reinforcement learning. When an agent receives feedback only upon completing complex tasks, exploration becomes prohibitively difficult, May 3, 2026APOLLO and the Rise of Virtual Patients: When Representation Quality Surpasses Task-Specific Optimization Introduction Modern healthcare generates data at a staggering pace, yet the sector remains trapped in a paradox of abundance and underutilization. While electronic health records capture nearly 30% of global data volume May 3, 2026 May 3, 2026When Should You Trust an LLM Judge? Per-Instance Reliability Through the Lens of Conformal Prediction and TransitivityLarge language models deployed as automated judges have become the invisible infrastructure of modern NLP evaluation. From ranking summarization systems to scoring chatbot responses, LLM-as-judge frameworks promise scala May 3, 2026The Sparse Core: How Neural Network Pruning Reveals the Concentration of Interpretability Introduction Modern convolutional neural networks contain millions of parameters, yet empirical evidence suggests that the vast majority of these weights are superfluous. Techniques for sparse pruning can remove 90% of May 3, 2026 May 3, 2026Learning to Learn Online: Regret Bounds as Meta-ObjectivesOnline learning algorithms power much of modern machine learning infrastructure, from recommendation systems to financial forecasting. The Online Gradient Algorithm (OGA) and Exponentially Weighted Aggregation (EWA) repr May 2, 2026The Calibration Problem: Why AI Literacy Alone Cannot Fix Overreliance in Educational SettingsThe integration of generative AI into higher education has moved beyond experimentation into routine practice. Students now regularly encounter large language models as tutors, debugging assistants, and writing collabora May 2, 2026VideoSeek and the Logic of Sparse Attention in Long-Horizon Video UnderstandingThe computational cost of video understanding has long been the elephant in the room for multimodal AI. As large multimodal models (LMMs) process increasingly lengthy video content, the default strategy has been to sampl May 2, 2026Spatial Features Without Action Grounding: An Analysis of R3D and 3D Policy LearningThe transition from 2D to 3D representations in robotic imitation learning has long been viewed as a necessary evolution. While Vision-Language-Action (VLA) models and diffusion policies built upon 2D backbones have achi May 2, 2026The Format Trap: How Instruction Tuning Creates Hidden Fragility in Large Language ModelsInstruction tuning has become the standard method for transforming base language models into helpful assistants, yet the robustness of this helpfulness remains poorly understood. The paper "One Token Away from Collapse: May 2, 2026Scale Space Diffusion: Rethinking Resolution in Generative Models Introduction Diffusion models have established themselves as the dominant paradigm in generative modeling, yet they carry a fundamental inefficiency that has persisted since their inception. When we add noise to an imag May 2, 2026Beyond Chain-of-Thought: Retrieving Reasoning to Reduce Token CostsThe rise of reasoning centric large language models, exemplified by OpenAI's o1 series and DeepSeek-R1, has fundamentally altered how we approach complex problem solving in artificial intelligence. These models explicitl May 2, 2026DROID-W: Geometric Contradiction as a Universal Prior for Dynamic SLAM Introduction Simultaneous Localization and Mapping (SLAM) in unconstrained environments remains one of computer vision's persistent challenges. Dynamic objects, from pedestrians and vehicles to swaying vegetation, viola May 2, 2026Scale-Adaptive Diffusion Models for Climate Downscaling: Architecture Reuse Across Spatiotemporal Scales The Scale Fragmentation Problem in Environmental Modeling Climate science operates across a fragmented landscape of spatial and temporal resolutions. Satellite observations, radar mosaics, gauge networks, and reanalysis May 2, 2026Repurposing Language Models for Vision-Free Spatial Intelligence: A Technical Analysis of IMU-to-4DThe proliferation of AI assistants capable of understanding human activity has historically tethered us to visual perception. Cameras, while informationally rich, introduce persistent constraints regarding privacy, energ