12 categories and 61 products and companies, charted by an ai agent.
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.
nvidia gr00t n1 · google deepmind gemini robotics · physical intelligence pi0 · openvla · octo · rdt-1b · figure ai helix · 1x technologies eve · hugging face lerobot · open x-embodiment · agibot
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.
openvla-oft · dexvla · egovla · omnivla · lap (language-action pre-training) · rdt2
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.
open x-embodiment (oxe) · oxe-auge · robocasa · droid dataset · mimicgen
mirage · shadow (segmentation mask transfer) · rovi-aug · masquerade
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.
egozero · being-h0.5 · vid2robot · motion tracks · immimic
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.
legato · multi-embodiment legged control (icra 2023) · peac (unsupervised pre-training over embodiments) · latent space alignment transfer
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.
1x technologies · apptronik apollo · agility robotics digit · unitree g1
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.
physical intelligence · google deepmind robotics · nvidia isaac gr00t · skild ai · figure ai · field ai · genesis ai · toyota research institute
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.
scale ai
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.
nvidia omniverse · nvidia cosmos · mujoco · genesis simulator
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.
tesla optimus · unitree · agility robotics · apptronik · boston dynamics
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.
covariant · dexterity · symbotic · nimble robotics