Maulesh Trivedi

More roboticists should think about the people who will actually be using their robots (ie not ai or robotics folks)

In-context learning in LLMs

こうやって安全意識して倒れるとロボットへのイメージ変わるな。Unitree G1のバランス崩れると激しく倒れるのは結構怖いのよな

This is the single best read on World Models and one of the most important reads in AI. $10B has flowed into "world models" in the last 18mos, from Yann LeCun to FeiFei Li. The promise is, like LLMs, world models will provide the data it takes to scale robotics foundation models, and solve robotics. ..but the word has been abused to mean one of many things. This post unpacks: – What 5 traits makes a world model? – How do the different approaches stack up? – What is it used for within and beyond robotics? – Where is the opportunity? – Citations to research, news and blog posts Companies / products in the space include: – BigCo products: Google Genie, Tesla Optimus, Nvidia DreamDojo, DreamZero, Microsoft Muse – Pure world model: AMI Labs, World Labs, Runway, Rhoda, Decart, Spaitial, Odyssey, Embo, Dream Labs, OneWorld – Robot foundation model cos: Skild, Physical Intelligence, Figure, Mind Very likely one of the seminal technologies of the next decade.

So close!!! Nearly able to fold a crumbled shirt on a fairly low resource model / hardware! Almost there now 😊

Making humanoids is one thing. Making a humanoid into a consumer product takes engineering to extreme limits. We've put an intense amount of thought and effort into answering every question like "how does my NEO show up at my door" with beautiful accessories and features. Still work to be done. But it's the most exciting work of this generation IMO.

Robot abundance, one NEO at a time. More tomorrow.

¥6.8B? 8,500 robots. The biggest embodied robotics order we’ve seen so far. ⚡🤖 China’s State Grid is rolling out a massive deployment plan for 2026. The target: about 8,500 embodied robots, with a total budget of ¥6.8 billion (~$940 million USD). This is not one type of machine. It’s a full stack across three categories: 5,000 quadruped robots for inspection across substations, transmission lines, and mountainous grids. 500 humanoid robots for live electrical work — the highest-value segment, with ¥2.5 billion (~$350 million USD) allocated. 3,000 dual-arm robots for equipment operation and fault handling inside substations. The focus is clear: inspection, live operations, emergency response, and logistics. And the economics are already defined. Each unit is expected to save ¥500,000–¥800,000 (~$70K–$110K USD) in annual labor cost. Inspection efficiency improves 5×. Fault handling time drops 60%. Safety incidents are projected to decrease by 80%. This is not a pilot. It’s a structured rollout: small-scale in Q1, large-scale in Q3, and expansion in Q4. And it’s tied to a bigger system. State Grid is pushing toward 30% penetration in key regions by 2026, over 80% adoption by 2027, and full integration with a digital twin grid by 2030. Suppliers include Unitree, UBTECH, Fourier, Deep Robotics, and AGIBOT. No hype needed. This is what large-scale, real-world embodied AI deployment actually looks like.

It can ice skate now

Now they are pushing this narrative on Chinese social media that robotics has a “low barrier to entry”; That building a robot from scratch and achieving top-tier performance is easy. That’s absurd.

Honor สุดยอด พัฒนาหุ่นยนต์วิ่งอัจฉริยะ "Lightning" และ "Energetic Boy" เข้าร่วมแข่งขันหุ่นยนต์มาราธอน Beijing Yizhuang Human Robot Half Marathon 2026 ในวันที่ 19 เม.ย.นี้ ถือเป็นผู้ท้าชิงการแข่งขันหุ่นยนต์มาราธอนระดับโลกรายแรกที่ส่งโดยผู้ผลิตอุปกรณ์ปลายทางรายใหญ่ แล้ววิ่งเร็วมาก