| license: apache-2.0 | |
| language: | |
| - en | |
| tags: | |
| - robotics | |
| - vision-language-action | |
| - foveation | |
| - openpi | |
| - pi05 | |
| - droid | |
| - jax | |
| # FoveaVLA-Dual | |
| FoveaVLA-Dual is the dual-camera variant of **FoveaVLA: Foveation Actions Replace the Wrist Camera in VLAs**. Built on π₀.₅, it uses two external RGB-D cameras with full views, foveal crops, and aligned depth. **No wrist camera is required.** | |
| - **Inputs:** two external ZED stereo cameras in physical-left / physical-right order; robot state; language instruction. | |
| - **Outputs:** 16 joint-position actions with gripper commands and two ordered interaction boxes for t+8. Execute 8 actions before the next query. | |
| - **Training data:** DROID demonstrations with RoboInter interaction annotations. | |
| - **Config:** `foveavla_dual`. | |
| ## Use | |
| After setting up the [FoveaVLA server environment](https://github.com/Stanford-TML/FoveaVLA/blob/main/docs/deployment.md), run: | |
| ```bash | |
| uv run --no-sync scripts/download_checkpoint.py --model=foveavla_dual | |
| XLA_PYTHON_CLIENT_MEM_FRACTION=0.70 uv run --no-sync scripts/serve.py \ | |
| --model=foveavla_dual --checkpoint=checkpoints/foveavla_dual | |
| ``` | |
| See [Stanford-TML/FoveaVLA](https://github.com/Stanford-TML/FoveaVLA) for data preparation, training, and real-robot deployment. The deployment guide covers stereo-depth setup, GPU memory, camera calibration, and supported robot configurations. | |