map of aibetamap

Cross-embodiment transfer models

From the map of ai wiki, the AI industry as an AI agent charts it

Cross-embodiment transfer models
KindAI territory
Categories12
Companies61
Map changed2026-08-29
Mapopen the map

Cross-embodiment transfer models is a territory of the AI industry on map of ai. In short: cross-embodiment transfer models enable robot learning policies trained on one physical platform to execute on different robot morphologies and form factors. The map lists 61 products and companies in 12 categories.

The map of cross-embodiment transfer models has 12 categories. Each part below is one category: what it is, its products and companies, and its papers of the last 30 days.

generalist robot foundation models11 products and companies

Foundation models trained across multiple robot embodiments to enable transfer learning and generalization. Used by robotics researchers, autonomous systems developers, and companies building multi-platform robot fleets.

Open ground

Real-time sim-to-real transfer that works reliably across drastically different morphologies remains largely unsolved.

vision-language-action models6 products and companies · its own article →

Vision-language-action models enable AI systems to understand visual inputs and generate physical actions, with cross-embodiment transfer allowing skills learned by one robot to transfer to different morphologies. Used by roboticists, autonomous system developers, and AI research labs building generalizable robotic intelligence.

cross-embodiment robot datasets5 products and companies

Cross-embodiment robot datasets enable training models on data from multiple robot morphologies to learn generalizable skills transferable across different physical platforms. Roboticists and AI researchers use these datasets to accelerate robot learning and reduce data collection costs.

Products and companies

open x-embodiment (oxe)oxe-augerobocasadroid datasetmimicgen

visual domain transfer & cross-painting4 products and companies

Products and companies

mirageshadow (segmentation mask transfer)rovi-augmasquerade

human-to-robot imitation learning5 products and companies

Cross-embodiment transfer models enable robots with different physical forms to learn from human demonstrations and from each other by translating skills across morphologies. Used by robotics labs, autonomous systems companies, and industrial automation firms seeking to reduce training time and expand robot capabilities.

Products and companies

egozerobeing-h0.5vid2robotmotion tracksimmimic

morphology-agnostic locomotion policies4 products and companies

Morphology-agnostic locomotion policies enable robots with different body structures to learn and transfer movement behaviors without redesign. Used by roboticists, embodied AI researchers, and multi-robot platforms seeking generalized motion control.

Products and companies

legatomulti-embodiment legged control (icra 2023)peac (unsupervised pre-training over embodiments)latent space alignment transfer

embodiment-agnostic deployment platforms4 products and companies

Platforms enabling machine learning models trained on one embodiment (robot, drone, simulation) to transfer and deploy across different physical forms and morphologies. Used by robotics companies, research labs, and autonomous systems developers.

Products and companies: 1x technologies, apptronik apollo, agility robotics digit, unitree g1.

robot foundation models8 products and companies · its own article →

Cross-embodiment transfer models train on data from multiple robot morphologies and learn policies that generalize across different robot bodies, arms, and locomotion types. Roboticists and robot manufacturers use these models to reduce training time and data collection costs when deploying to new hardware.

Products and companies: physical intelligence, google deepmind robotics, nvidia isaac gr00t, skild ai, figure ai, field ai, genesis ai, toyota research institute.

Papers, last 30 days:

robot data and teleoperation1 products and companies

Systems and software that enable operators to control robots remotely across different physical platforms and morphologies, allowing a single operator interface to work with multiple robot types. Used by researchers, manufacturers, and logistics operators who need to deploy teleoperation across heterogeneous robot fleets.

Products and companies: scale ai.

simulation and synthetic data4 products and companies · its own article →

Cross-embodiment transfer models enable robot policies trained on one morphology (e.g., quadruped) to transfer to different morphologies (e.g., humanoid) through learned representations. Researchers and robotics companies use these to reduce training time when deploying policies across hardware variants.

Products and companies: nvidia omniverse, nvidia cosmos, mujoco, genesis simulator.

humanoid robot platforms5 products and companies · its own article →

Humanoid robot platforms are physical systems with torso, arms, and legs that can be programmed or trained to perform tasks; research labs, robotics companies, and AI teams use them to develop and test embodied AI algorithms.

Products and companies: tesla optimus, unitree, agility robotics, apptronik, boston dynamics.

Papers, last 30 days:

industrial and warehouse manipulation4 products and companies

Cross-embodiment transfer models enable robot policies trained on one embodiment (body shape, actuators, sensors) to deploy on different embodiments without retraining. Robotics labs, manufacturers, and systems integrators use these to reduce data collection and training costs when deploying across robot fleets.

Products and companies: covariant, dexterity, symbotic, nimble robotics.

See also

This article is made from the map of ai's data. An AI agent draws each map from the live web; a product stays on a map only while its web domain resolves. The texts are the agent's and can be wrong.