From simulation to physical walking
Follow the model, policy and hardware checks behind sim-to-real locomotion, with a measured OP3 walking study and explicit limits.
Choose a physical mechanism, a hardware question, a research method or an event edition.
Build a model, inspect its physics, train a controller and measure the gap to a physical robot.
Follow the model, policy and hardware checks behind sim-to-real locomotion, with a measured OP3 walking study and explicit limits.
Learn what robot logs can identify, how to compare trajectories and how PACE and ASAP use physical data in two different correction methods.
Load a pinned Unitree G1 model, inspect 29 actuators and record a joint response with a Python example executed in MuJoCo 3.15.0.
Understand Isaac Sim 6.1, Isaac Lab 3.0 Early Access, current GPU requirements and the source configuration behind a G1 locomotion task.
Compare eight simulation tools by version, operating system, hardware, model format and licence, then place learning and ROS tools in the right layer.
Use a pinned MuJoCo Playground G1 task to define observations, actions, rewards and held-out trials, with empty evaluation templates.
Choose learning resources and understand the scope of product conformity.
Compare real robotics learning resources, their costs and credential conditions, then build six projects with observable engineering outputs.
Understand the scope of five ISO documents, CE marking, the EU machinery transition and the evidence needed to check a robot certification claim.
Follow the measurements and commands behind walking, grasping and teleoperation.
Follow one humanoid step from IMU and joint measurements to foot placement, motor commands and recovery, with equations and measured G1 trials.
How humanoid fighting combines joint motion, balance and contact sensing, with official arena, scoring and safety rules and limits on autonomy claims.
Follow object detection, camera calibration, grasp planning, finger actuation and tactile feedback, with measured results and failure cases.
How VR tracking, gloves, exoskeletons and motion capture drive robot joints, with command rates, network delays, safety limits and learning data.
How cameras, depth sensing, SLAM and object tracking give humanoid controllers usable geometry, with a documented Atlas example and failure cases.
Follow a bottle-to-table task through perception, planning, learned policies, joint control, result checks and stops, using documented research architectures.
Identify a model revision, read its component list and check what the purchase includes.
Listed Unitree R1 and G1 prices, development access and the extra costs to include in a hardware quote.
What 1X NEO order terms, basic autonomy, Expert Mode and privacy documentation establish for a prospective home buyer.
Separate balance, navigation, manipulation, task planning and recovery using published Atlas and NEO control examples.
Compare G1, Atlas and Digit 5 battery claims. Learn how heat, charging and task failures affect working time, with a reproducible eight-hour shift calculation.
How GelSight and DIGIT use optical tactile sensing, with examples of hidden contacts, small-tool handling and sensor wear.
Compare T800 variants, joint control, sensors and published specifications, including conflicts between EngineAI's product pages and developer manuals.
Understand electric Atlas hardware, hands, perception, battery swapping and factory training, with historical robots and product specifications kept separate.
Run a first simulation, distinguish learning methods and interpret a paper or recording.
A practical worksheet for identifying the task, control mode, edits and missing attempts in a robot video.
A compact reading method for robotics papers that keeps hardware, test conditions, success counts and missing evidence together.
Start with a small Python and MuJoCo control exercise, then learn where ROS 2 messages fit into a robot software system.
What the two terms mean, where they overlap, and the evidence to ask for behind a robotics claim.
Imitation learning, reinforcement learning and simulation explained through ALOHA and a robotic hand experiment.
How RT-2 and OpenVLA turn images and instructions into robot actions, and what their results do not establish.
Find the edition, rules and published results behind an event name.
Selected running, sorting and handling results from the first Beijing Games, with the reported control mode and limits of the closing record.
The 2026 Games combined track events with manipulation tasks. Read the event counts, a reported race result and the limits of each.
The 2025 Beijing conference programme, named humanoid exhibits and its distinction from the separate August Games.
What the 2026 Beijing conference launched, who organized the application programme and what its annual target does and does not establish.
Tiangong Ultra won the 2025 race in 2 hours 40 minutes 42 seconds. Here is what the time tells us and which conditions remain unknown.
Lightning won the 2026 race in 50 minutes 26 seconds. The scoring rules explain how autonomy, remote control and support affect comparisons.
Locate two IROS 2025 humanoid papers, identify their hardware and separate a biped walking experiment from cart-mounted manipulation.
Find the official IROS 2026 dates, programme access and journal-presentation policy, with the distinction between a paper date and a conference session.