simulo.ObservationComponent

One term of a task’s observation vector.

A task composes its observation space from components — each contributes dim(robot) values per environment, computed fresh every step. Structural mirror of simulo.core.observations.ObservationComponent.

initialize()

ObservationComponent.initialize(robot: RobotProtocol, device: str, num_envs: int) -> None

One-time setup: cache the robot handle and size buffers for num_envs on device.

dim()

ObservationComponent.dim(robot: RobotProtocol) -> int

Number of values this component contributes per environment.

compute()

ObservationComponent.compute(prev_actions: TensorLike) -> TensorLike

Return this component’s observation slice, shape (num_envs, dim).

prev_actions is the last applied action batch, for components that observe the policy’s own output.

---

Note

simulo.ObservationComponent is a mode-aware name: at submit time it resolves to a lean, torch-free stand-in so your app imports instantly; on the GPU worker it resolves to the real runtime class. The contract below is the definitive list of members every simulo.ObservationComponent provides, and the contract your own subclass fulfils.

Declared as:

@runtime_checkable
class ObservationComponentProtocol(Protocol)