robotics intelligence comes from all, serves all. littleprince.axisrobotics.ai

𝐪𝐮𝐢𝐜𝐤 𝐮𝐩𝐝𝐚𝐭𝐞 𝐨𝐧 Axis AI 𝐚𝐱𝐢𝐬 𝐫𝐨𝐛𝐨𝐭𝐢𝐜𝐬 𝐣𝐮𝐬𝐭 𝐫𝐞𝐥𝐞𝐚𝐬𝐞𝐝 𝐭𝐡𝐞𝐢𝐫 𝐟𝐢𝐫𝐬𝐭 𝐦𝐢𝐬𝐬𝐢𝐨𝐧 𝐜𝐚𝐥𝐥𝐞𝐝 𝐥𝐢𝐭𝐭𝐥𝐞 𝐩𝐫𝐢𝐧𝐜𝐞 𝐫𝐨𝐬𝐞. your task is to teach a robot arm how to find a kettle in the noise, grasp it, water a pot, and

Little Prince’s Rose — 24‑Hour Log In the first 24 hours after launch, we saw 10,000+ unique, real addresses participate, and collected ~3,500 data trajectories that are ready for model training. The community has been creating their own versions of the rose — and the world

LLM-empowered task generation: Verified We are using GenAI to turn our simulation asset library into a massive engine for high-quality training tasks. More diverse tasks → More diverse data → A model that actually generalizes. The pilot works perfectly. Next step: Massive

Axis Delta 002 Raw User Input vs. Optimized Training Signal Real-world user demonstrations can be messy (e.g., discontinuities, hand shake, inconsistent motion). But as demonstrated in our results below, our data optimizer significantly cleans and stabilizes this raw input. By

Two days since the launch of our "Little Prince's Rose" experiment, over 5,000 Little Princes have already watered their roses in the simulation. Thanks to the community's active participation and support, we’ve reserved some little surprises for our devoted caretakers.🌹

最近深入研究了 Axis Robotics 的一些资料 今天根据理解用心做了视频 并且打算聊聊我所认知的Axis： 为什么机器人需要在仿真中学习，而不是在现实中试错？ axis最让我兴奋的地方是什么？ 小王子的玫瑰实验到底验证了什么？ ⬇️ #axisai ChrisF(✱,✱) 郡主Christine (✱,✱) zagen扎根(✱,✱) PL 漂亮 (✱,✱)

What we’re especially excited about is that all of this data was contributed by retail users, entirely through our browser‑based data collection platform. We’ve already started training policies with this dataset and are moving toward real‑world deployment. We expect to have a

Thoughts on Value Capture: The "Sim-Real" Recipe We strongly believe the industry is converging on a Hybrid Training future. However, unlike the "Video + Real" approach, we are betting on a "Sim + Real" trajectory. For us, the ultimate moat isn't just the data itself, but the

Spot on from Marc Andreessen 🇺🇸: "The atomic unit of work is the task." A job is just a bundle of tasks. At Axis AI , we are taking this logic one layer deeper into the physical world: A Job = A bundle of Tasks. An Axis Task = A bundle of Trajectories. For us, the Trajectory is the

Introducing our open-sourced simulation task generation pipeline. Repo: github.com/AxisAIOrg/Axis… The trajectory is the true atomic unit of intelligence—a digitized instance of human wisdom captured through physical execution. Our task generation pipeline serves as the engine

The "Impossible Trinity" of Data Collection: Low Cost, Massive Scale, High Quality. We just cracked it. We analyzed thousands of retail user trajectories from the "Little Prince's Rose" task to answer one question: What makes data valuable? By rigorously optimizing dimensions

Axis Delta 003 Infinite Scenarios via Task Generation Engine Following our Asset Generation (Delta 001), we are now solving for Environment Diversity. We have completed our automated Task Generation Engine. The Pipeline: Input: Raw OBJ/URDF assets. Output: Physically valid,

Axis Delta 004 Data Scoring is LIVE We ported our evaluation logic to JS for instant feedback. Users now get an instant Quality Score immediately after task completion. The logic computes 4 dimensions: 1️⃣ Length: Trajectory efficiency. 2️⃣ Completion: Task success verification.

Our whole team is excited about using our platform to help pre-train general-purpose robotics foundation model by gathering diverse data through community efforts

Announcing DreamDojo: our open-source, interactive world model that takes robot motor controls and generates the future in pixels. No engine, no meshes, no hand-authored dynamics. It's Simulation 2.0. Time for robotics to take the bitter lesson pill. Real-world robot learning is

AXIS AI Beta Test Phase 1上线第四天 今天给大家做了一个视频教程， 简单讲一下整个Test是怎么参与、怎么操作的。 如果你还没体验过，可以先看一遍，上手会更丝滑。 第一次做视频，许多地方做得不好，请大家多多包涵。 欢迎大家一起加入 Axis AI 共同建设。 郡主Christine (✱,✱) ChrisF(✱,✱)

Welcome to the new world

Our beta test pre-launched for community members with 27 tasks during the Spring Festival. Within just 5 days, it collected over 100,000 valid data trajectories and engaged 18,000 participants.