Stars
This is the official implementation of the paper "ConRFT: A Reinforced Fine-tuning Method for VLA Models via Consistency Policy".
EVOLVE-VLA: Test-Time Training from Environment Feedback for Vision-Language-Action Models
RLinf is a flexible and scalable open-source infrastructure designed for post-training foundation models (LLMs, VLMs, VLAs) via reinforcement learning.
A unified agent system for general-purpose robots.
Official implementation for Compliant Residual DAgger
Official Release of "Mixture of Horizons in Action Chunking"
A set of examples based on verl for end-to-end RL training recipes.
Code for [AAAI 2026] AffordDex: Towards Affordance-Aware Robotic Dexterous Grasping with Human-like Priors
Various retargeting optimizers to translate human hand motion to robot hand motion.
Real-Time VLAs via Future-state-aware Asynchronous Inference.
siiRL: Shanghai Innovation Institute RL Framework for Advanced LLMs and Multi-Agent Systems
[AAAI 26 Oral] SpatialActor: Exploring Disentangled Spatial Representations for Robust Robotic Manipulation
InternVLA-M1: A Spatially Guided Vision-Language-Action Framework for Generalist Robot Policy
Blender addon for Depth-Anything-3 3D reconstruction
Video-CoM: Interactive Video Reasoning via Chain of Manipulations
Dexbotic: Open-Source Vision-Language-Action Toolbox
NORA-1.5: A Vision-Language-Action Model Trained using World Model- and Action-based Preference Rewards
The repository provides code for running inference and finetuning with the Meta Segment Anything Model 3 (SAM 3), links for downloading the trained model checkpoints, and example notebooks that sho…
Evo-1: Lightweight Vision-Language-Action Model with Preserved Semantic Alignment
Running VLA at 30Hz frame rate and 480Hz trajectory frequency
Official Reporsitory of "RoboEngine: Plug-and-Play Robot Data Augmentation with Semantic Robot Segmentation and Background Generation"
A Comprehensive Survey on World Models for Embodied AI
Code for "PHUMA: Physically-Grounded Humanoid Locomotion Dataset"