What is embodied AI?
Embodied AI is artificial intelligence that has a body: it perceives the physical world through sensors and acts in it through a robot, learning from that interaction.
Updated:
Definition
Embodied AI refers to AI systems that live in a physical body, typically a robot, and learn to act through sensing and interaction rather than just text or images.
Key ingredients
- Perception: cameras, depth, touch, force, audio.
- Decision-making: often a robot foundation model such as a VLA model.
- Action: motor commands to arms, hands, legs or wheels.
- Data: real-world, simulated and human demonstration data.
Why it matters
Language models learned from the internet. Robots have no internet-scale dataset of physical actions, so how embodied AI collects data and represents actions is the central research question.
Frequently asked questions
How is embodied AI different from chatbots?
A chatbot only produces text or images. Embodied AI must turn perception into physical actions in real time, where mistakes have physical consequences.
Is embodied AI the same as physical AI?
They are closely related and often used interchangeably. 'Physical AI' is popular in industry; 'embodied AI' is more common in research.
What data does embodied AI learn from?
Real robot recordings, simulation, and human demonstrations captured with video, data gloves or sensor suits.
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.
NEURA's Neuron AI adds touch, sound and memory to the robot foundation model
NEURA Robotics launched Neuron AI, a 'perception-language-action' model that fuses 3D vision, touch, force, audio, robot state and language with memory, and will demo an 11-minute laundry task on its 4NE1 humanoid.
Astribot raises ~$149M as cable-driven humanoids hit a $2.1B valuation
Shenzhen's Astribot closed an RMB 1 billion round at about RMB 15 billion valuation — its fourth raise in six months — to scale cable-driven S1 and T1 robots.
Physical Intelligence opens its π0 robot models to everyone via an API
The Pi API beta lets robot builders upload data, fine-tune π0/π0.5 on Pi's infrastructure and run the result through hosted inference.
BYD's humanoid design patent surfaces while a formal debut waits
China's IP office published BYD's embodied-intelligent-robot design patent on Oct 9, 2026 — a bipedal look aimed at service, education and industry — but the company has still not formally unveiled the product.
Ultra raises $62M for warehouse robots-as-a-service — with Physical Intelligence as the brain
Brooklyn's Ultra closed a $50M Series A (Framework Ventures) plus earlier seed for $62M total, leasing OP1 packers while Physical Intelligence supplies the AI.
Mecka raises $60M Series B to digitize human motion for robot training
Sequoia led Mecka's $60M Series B with NVIDIA, Microsoft's M12, Qualcomm Ventures and Samsung; the startup captures human demonstration data and says it passed $100M run-rate revenue in June 2026.
Minerva leaves stealth with ~$10M to put specialists behind VR, not blast suits
General Catalyst led Minerva Humanoids' ~$10M pre-seed for Roger, a semi-autonomous humanoid aimed at oil & gas and EOD; founders include ex-Tesla Optimus and Boston Dynamics Stretch talent.
Vela: a VLA that predicts smooth trajectories instead of action chunks
A Fudan/CUHK team's Vela model outputs spline curves with an adaptive time horizon, beating π0.5 on LIBERO-X and EBench and on real egg-cake cooking and potato-shredding tasks.
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.
Apptronik's Robot Park: Apollo 2 fleets collecting data for Gemini Robotics
The Austin humanoid firm expanded its ~90,000 sq ft training hub and wheeled/bipedal Apollo 2 platforms with Google DeepMind, Mercedes-Benz and GXO sites feeding the commercial Apollo 3.
Newsletter
Robrief Weekly
One email a week with the robotics news that matters. Free, unsubscribe anytime.