openvla 开源项目推荐（43 个）

datawhalechina /every-embodied

仅需Python基础，从0构建自己的具身智能机器人；从0逐步构建VLA/OpenVLA/SmolVLA/Pi0，深入理解具身智能

arpitg1304 /forge

机器人数据工具包 |在机器人数据集格式（RLDS、LeRobot v2/v3、Zarr、HDF5、Rosbag）之间转换。检查、可视化和分析数据集。可配合HuggingFace Hub使用。专为OpenVLA、Octo、LeRobot和扩散策略工作流打造。

IP127000 /openVLA-Qwen2-0.5B

OpenVLA Lightweight Version(0.5B). It uses qwen2-0.5B and fine-tunes using mllm format, without occupying LLM's inherent tokens. It represents actions with an additional 271 tokens.

rsasaki0109 /vla_ros2

ROS2-native runtime, benchmark, and adapter hub for Vision-Language-Action models.

miladfa7 /PickAgent

PickAgent: OpenVLA-powered Pick and Place Agent | Gradio&Simulation | Vision Language Action Model

ArchishmanSengupta /mlx-vla

Train and deploy Vision-Language-Action models natively on Apple Silicon.

lz-googlefycy /vla-lab

Independent VLA research notes: OpenVLA / π0 / Spirit paper reviews, LIBERO reproduction, XLeRobot integration. Transitioning from AD motion planning to embodied AI.

Masoudjafaripour /VLM-VLA-Embodied-Agents

Using VLM-based visual question answering to perceive scenes and control robots in MuJoCo simulation.

GimpelZhang /genesis-world-tour

A full-stack Embodied AI simulation suite powered by Genesis World: featuring OpenVLA closed-loop evaluation, procedural scene generation, massively parallel GPU RL (Unitree Go2), and multi-physics coupling (PBD/SPH).

kalaiselvan-t /openvla-fr3

FR3 robot in Gazebo integrated with 4-bit quantized OpenVLA and MoveIt

ravijo /vla-grasp

OpenVLA-based grasping in a custom PyBullet environment

lz-googlefycy /openvla-libero

Reproducing OpenVLA-7B on LIBERO 4-suite on a single commodity GPU. 400 rollouts, 4 demo videos, 6 debugging gotchas, all open.

lusunn111 /RoboNix-Retrieval-Augmented-Toolkit

A RoboNix Skill for experience-memory retrieval and verified action reuse across OpenVLA and π0.

tc-huang /lerobot-policy-openvla-oft

OpenVLA-OFT ported to LeRobot as an out-of-tree policy plugin, with LIBERO checkpoints.

VihanAggarwal /shift-safe-monitors

Safety monitors for learned robot policies under distribution shift. A benchmark for which monitor still works once the deployment distribution moves.

Rick0525 /vla-rl-post-training

Reproducing 8-GPU VLA RL post-training (SimpleVLA-RL GRPO recipe on RLinf) with 2xA800 + LoRA: 93.8 on LIBERO-Spatial at 23% of the official data budget

chaoshengsc /libero-camera-ablation

Visual-dependence study on LIBERO-Spatial: four-policy blackout exploration, three-policy perturbations, focused pi0/pi0.5 wrist controls, and an upstream camera-slot bug-fix PR.

StaryMoon /OpenVLA-Unofficial

Unofficial PyTorch reproduction for OpenVLA: An Open-Source Vision-Language-Action Model.

farouk15160 /vla-arm-sim

Simulated UR5e + MoveIt 2 cell for testing vision-language-action policies in ROS 2 Jazzy. Swap SmolVLA, OpenVLA-7B (4-bit) or GR00T N1.7 behind one ZeroMQ protocol — all runnable on a 6 GB GPU.

FERBIN12 /vla-manipulation-ablations

VLA manipulation ablations, held-out generalization, and an OpenVLA-7B feasibility study, all measured on a 6GB laptop GPU