FoveaVLA-Single

FoveaVLA-Single is the single-camera variant of FoveaVLA: Foveation Actions Replace the Wrist Camera in VLAs. Built on π₀.₅, it uses one external RGB-D camera with full views, foveal crops, and aligned depth. No wrist camera is required.

• Inputs: one external ZED stereo camera; robot state; language instruction.
• Outputs: 16 joint-position actions with gripper commands and one interaction box for t+8. Execute 8 actions before the next query.
• Training data: DROID demonstrations with RoboInter interaction annotations.
• Config: foveavla_single.

Use

After setting up the FoveaVLA server environment, run:

uv run --no-sync scripts/download_checkpoint.py --model=foveavla_single
XLA_PYTHON_CLIENT_MEM_FRACTION=0.70 uv run --no-sync scripts/serve.py \
  --model=foveavla_single --checkpoint=checkpoints/foveavla_single

See 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.