What is a world model in robotics?
A world model is an AI model that learns how the physical world behaves, so it can predict what happens next and let a robot plan or train in imagination.
Updated:
Definition
A world model learns the dynamics of the environment — how objects move, fall, deform and respond to a robot's actions — usually from large amounts of video and robot data.
How it works
Given the current observation and a candidate action, the model predicts the next frames or states. Robots can use it to plan, and developers can use it as a learned simulator to generate data and evaluate policies, complementing sim-to-real.
Why it matters
Physical data is scarce. Companies scaling human video — like Figure's Index dataset behind Helix 2.5 (story) — are betting that learning how the world works is the path to general embodied AI.
Frequently asked questions
How is a world model different from a VLA model?
A VLA model maps what the robot sees and is told directly to actions; a world model predicts future states of the world, which can then be used to choose or train actions.
What are world models used for?
Generating synthetic training video, testing policies in a learned simulator, and planning by imagining outcomes before acting.
Are world models the same as simulators?
Related but different: a physics simulator is hand-built from equations; a world model is learned from data such as video.
Coverage
Hesai's Kosmo turns real rooms into robot training worlds in minutes
The lidar maker launched Kosmo One (8K spatial camera) plus Kosmo World cloud, partnering with Galbot and Dexmal; a Dairy Queen scan cut modeling from ~300 hours to 15 minutes.
AWS opens a Physical AI Toolchain with NVIDIA stack hooks
On Oct 8, 2026 AWS launched an open-source Physical AI Toolchain that stitches synthetic data, SageMaker training, simulation and edge deploy — with NVIDIA Isaac and Cosmos in the mix — after RoboMaker's 2025 shutdown.
Skild's S1 learns soccer by playing itself for 140 simulated years
After launching S1 — a robotics foundation model that copies a single video demo without fine-tuning — Skild showed Physical Self-Play: the same base policy trained against itself in Isaac Sim, then transferred to a humanoid.
Figure's Helix 2.5 cleans 30 homes it has never seen
Figure says its humanoid tidied rooms, folded towels and made beds in 30 unseen Bay Area homes with zero local data, backed by a $3.5B compute deal with Nscale.
Newsletter
Robrief Weekly
One email a week with the robotics news that matters. Free, unsubscribe anytime.