Topic
Robotics and embodied intelligence, reported from Shenzhen
Robotics is where a model stops being a text prediction and starts moving something with mass. Control, sensing, locomotion and the reality gap between a simulator and a floor. Shenzhen is the right place to watch this, because the hardware iterates here first.
26 pieces
Teardowns
2- Sample Efficiency in Reinforcement Learning: A Foal Needs 62 Attempts, a Robot Needs Ten Billion Steps Reinforcement learning brute-forces locomotion in simulation. A foal walks after about 62 attempts. The gap is not compute, it is the structure a random policy does not arrive with.
- Self-Balancing Robot Simulation: What Transfers and What Does Not The inverted pendulum is the standard teaching model because it fails informatively. Simulating one is easy. Transferring the controller is not, and the reasons are always the same handful.
Essays
6- China's University Major Cuts Are AI Policy, and Nigeria Should Read the Fine Print China's education ministry revoked or suspended 12,200 undergraduate programme slots in five years and added 10,200. The number everyone repeated is real. What it means has been reported wrong.
- Made in China Quality: What the Label Actually Predicts in 2026 The label used to mean cheap and derivative. In batteries, solar, drones, robots and open-weight models it now marks the frontier. The perception lag is worth money to anyone testing on evidence.
- China Humanoid Robots Shipped 19,100 Units in Six Months. The Reducers Matter More. An ageing workforce and an explicit industrial policy point the same direction. Beneath the humanoid demos sits a component supply chain that quietly changed hands: reducers, sensors, actuators.
- Robot Safety Standards Are Written Before Any Law Arrives Robot governance is settled in ISO committees, insurer-funded biomechanics labs and tender documents, years before any parliament votes. In July 2026 the EU moved its own deadline to wait for the stan
- Wheel-Legged Robots: Buying Two Locomotion Modes, Paying for Three Wheels are cheap per metre. Legs handle discontinuity. The hybrid tries to buy both and pays in actuators, distal mass and a switching problem. An engineer's account of when that trade is worth it.
- Sensor Fusion in Sports Tracking and What 500Hz Is Actually For A 500Hz sensor in a football is a timestamping instrument, not a measuring one. What fusion actually buys you, and why clock alignment across devices breaks more systems than sampling rate ever does.
Tools
12 of 18- MuJoCo review Google DeepMind's rigid-body physics engine for robotics. Contact-rich dynamics, MJCF models, plus MJX and MuJoCo Warp for batched GPU rollouts during policy training.
- ONNX Runtime review Microsoft's cross-platform inference engine for ONNX models, with execution providers for CUDA, TensorRT, OpenVINO, CoreML, NNAPI, QNN, DirectML and XNNPACK.
- RoboBrain 2.5 review BAAI's embodied brain model (悟界·RoboBrain), Apache 2.0 at 4B and 8B, shipped in parallel NVIDIA and Moore Threads builds. It plans and points in 3D for a controller underneath.
- Roboflow review End-to-end computer vision platform for labelling, training and deploying detection models to cloud or edge hardware like Jetson, with a visual builder for multi-step vision pipelines.
- Runway review Video generation and in-context editing platform built on Gen-4.5 and Aleph 2.0, with a developer API fronting third-party models and a robotics simulation toolkit on top of GWM-1.
- AgiBot World review Open manipulation data and policy weights from AgiBot (智元机器人): a million-plus real robot trajectories, the GO-1 VLA and the Genie Envisioner world model, all non-commercial.
- CVAT review MIT-licensed annotation tool for images, video and 3D point clouds. Self-host via Docker, plug SAM or YOLO into Nuclio for pre-labelling, export to COCO, YOLO, KITTI and 20 other formats.
- Genesis World review Open-source multi-physics simulator for robot learning. Rigid, FEM, MPM, SPH and PBD solvers share one Python scene, with a photoreal renderer and heavy GPU batching.
- NVIDIA Jetson and TensorRT review NVIDIA's edge inference stack: Jetson modules from Orin Nano to Thor T5000, with TensorRT compiling models into hardware-specific FP4, FP8, INT8 and INT4 engines.
- SpatialLM review Manycore Tech (群核科技) of Hangzhou, behind Kujiale and Coohom, open-sourced its indoor spatial stack. SpatialLM turns point clouds into structured layouts, SpatialGen generates scenes.
- Ultralytics YOLO26 review Current YOLO model family from Ultralytics: five sizes, NMS-free end-to-end detection plus segmentation, pose and oriented boxes, exportable to TensorRT, ONNX, OpenVINO and TFLite.
- Unitree Robotics Developer Stack review The open repos behind Unitree's robots (宇树科技): BSD-3 C++ and Python SDKs, MuJoCo and IsaacLab RL training, ROS 2 packages and XR teleoperation for data capture.