maple.backend.policy.openvla.OpenVLAPolicy
- class maple.backend.policy.openvla.OpenVLAPolicy
Backend for OpenVLA vision-language-action models.
OpenVLA is a generalist robot policy that conditions on visual observations and natural language instructions to predict robot actions. The model requires dataset-specific statistics for action unnormalization to convert normalized model outputs into executable robot commands.
The backend manages Docker containers running the OpenVLA inference server, which loads the model from HuggingFace and serves predictions via HTTP API.
Methods
act(handle, payload, instruction[, model_kwargs])Get action prediction for a single observation.
info()Get policy backend information and capabilities.
Attributes