π₀.₅ is a vision-language-action model developed by Physical Intelligence, with the ability of open-world generalization, capable of performing robot tasks in brand-new environments and scenarios never seen during training.

MultimodalSafetensors

SmolVLA is a compact and efficient vision-language-action model that can achieve competitive performance while reducing computational costs and can be deployed on consumer-grade hardware. This model is trained using LeRobot and is specifically designed for robotics applications.

MultimodalSafetensors

SmolVLA is a compact and efficient vision-language-action model that can maintain competitive performance while reducing computational costs and is suitable for deployment on consumer-grade hardware. This model is trained using the LeRobot framework and is specifically optimized for robotics tasks.

MultimodalSafetensors

SmolVLA is a compact and efficient vision-language-action model that can achieve competitive performance while reducing computational costs and can be deployed on consumer-grade hardware. This model is trained using the LeRobot framework and is specifically designed for robotics applications.

MultimodalSafetensors

A robot strategy model trained based on LeRobot for the task of picking up the power gem.

MultimodalSafetensors

A robot strategy model trained based on the LeRobot framework for specific task execution

MultimodalSafetensors

A robot strategy model trained based on the LeRobot framework, suitable for robot control tasks

MultimodalSafetensors

π0+FAST is an efficient action tokenization scheme designed for robotics by Physical Intelligence, suitable for vision-language-action tasks.

MultimodalSafetensors

Pi0 is a general robot control model based on vision-language-action flow, supporting robot control tasks.

MultimodalSafetensors

Action Chunking Transformer policy, trained for the AlohaInsertion environment, designed for fine bimanual manipulation tasks.

Computer VisionSafetensors

lerobot is a PyTorch model focused on robot technology, integrating motion generation and model hub mixing functions.

MultimodalSafetensors

A robot control model trained using Diffusion Policy, specifically designed for PushT tasks, utilizing keypoint observation data for training

Computer VisionTransformers

VQ-BeT is a behavior generation model trained for the PushT environment, designed based on latent action principles

Computer VisionTransformers

Diffusion Policy is a visual-motor policy learning model based on action diffusion, specifically designed for the PushT environment.

Computer VisionTransformers

Action Chunking Transformer policy, specifically designed for the AlohaTransferCube environment, for fine bimanual robotic manipulation tasks

Computer VisionTransformers